1 2017-12-22 Mike Stump <mikestump@comcast.net>
2 Eric Botcazou <ebotcazou@adacore.com>
4 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
6 2017-12-21 Jakub Jelinek <jakub@redhat.com>
9 * config/i386/i386.c (ix86_function_arg_boundary): Return
10 PARM_BOUNDARY for TYPE_EMPTY_P types.
13 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
14 if navail is >= dir.len.
16 2017-12-21 Steve Ellcey <sellcey@cavium.com>
18 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
21 2017-12-21 Jakub Jelinek <jakub@redhat.com>
23 PR rtl-optimization/80747
24 PR rtl-optimization/83512
25 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
26 succ edge from ENTRY, copy partition from e->dest to the newly
28 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
29 ENTRY, use BB_PARTITION of its successor block as current_partition.
30 Don't copy partition when splitting succ edge from ENTRY.
32 PR tree-optimization/83523
33 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
34 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
35 (convert_mult_to_fma): Likewise.
37 PR tree-optimization/83521
38 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
39 gimple_build_assign without code on result of
40 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
43 2017-12-21 Andrew Pinski <apinski@cavium.com>
44 Steve Ellcey <sellcey@cavium.com>
46 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
49 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
52 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
53 constraints to operand 2.
54 (*ashl<mode>3_mask_1): Ditto.
55 (*<shift_insn><mode>3_mask): Ditto.
56 (*<shift_insn><mode>3_mask_1): Ditto.
57 (*<rotate_insn><mode>3_mask): Ditto.
58 (*<rotate_insn><mode>3_mask_1): Ditto.
60 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
62 * reorg.c (make_return_insns): Reemit each insn with its own location.
64 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
66 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
67 generation for cases where splatting a value is not useful.
68 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
69 across a vec_duplicate and a paradoxical subreg forming a vector
72 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
74 * combine.c (simplify_set): Do not transform subregs to zero_extends
75 if the destination is not a scalar int mode.
77 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
80 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
81 shared zero if the input has overflowed.
83 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
85 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
86 support for Cortex-A55 and Cortex-A75.
88 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
90 * common/config/arm/arm-common.c (compare_opt_names): Add function
91 comment. Use strcmp instead of manual loop.
93 2017-12-21 Martin Liska <mliska@suse.cz>
96 * gcov-dump.c (dump_gcov_file): Do not read info about
97 support_unexecuted_blocks for gcda files.
99 2017-12-21 Jakub Jelinek <jakub@redhat.com>
101 PR rtl-optimization/82973
102 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
103 (valid_for_const_vector_p): ... this.
104 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
105 (valid_for_const_vector_p): ... this. Adjust function comment.
106 (gen_vec_duplicate): Adjust caller.
107 * optabs.c (expand_vector_broadcast): Likewise.
108 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
109 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
111 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
112 instead of GET_CODE == CONST_FIXED.
113 (simplify_subreg): Use CONST_FIXED_P macro instead of
114 GET_CODE == CONST_FIXED.
117 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
118 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
119 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
120 (ix86_option_override_internal): Adjust for
121 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
122 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
123 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
124 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
125 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
126 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
127 def_builtin2 for CET builtins.
128 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
129 up formatting in IX86_BUILTIN_RDPID code.
130 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
131 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
132 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
133 ix86_isa_flags2 to ix86_isa_flags.
134 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
136 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
137 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
138 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
140 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
141 Or in OPTION_MASK_ISA_AVX512F_SET.
142 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
143 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
144 (ix86_handle_option): Adjust for
145 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
146 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
148 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
149 Alan Hayward <alan.hayward@arm.com>
150 David Sherwood <david.sherwood@arm.com>
152 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
153 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
154 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
155 Track polynomial offsets.
157 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
158 Alan Hayward <alan.hayward@arm.com>
159 David Sherwood <david.sherwood@arm.com>
161 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
162 Treat drb->init as a poly_int. Fail if its misalignment wrt
163 vector_alignment isn't known.
165 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
166 Alan Hayward <alan.hayward@arm.com>
167 David Sherwood <david.sherwood@arm.com>
169 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
171 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
172 (vect_loop_versioning): Take the loop versioning threshold as a
174 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
175 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
176 versioning_threshold.
177 (vect_analyze_loop_2): Compute the loop versioning threshold
178 whenever loop versioning is needed, and store it in the new
179 field rather than combining it with the cost model threshold.
180 (vect_transform_loop): Update call to vect_loop_versioning.
181 Try to combine the loop versioning and cost thresholds here.
183 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
184 Alan Hayward <alan.hayward@arm.com>
185 David Sherwood <david.sherwood@arm.com>
187 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
188 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
189 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
190 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
191 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
192 (group_compare_offset): Update accordingly.
193 (split_small_address_groups_p): Likewise.
194 (record_use): Take addr_offset as a poly_uint64 rather than
195 an unsigned HOST_WIDE_INT.
196 (strip_offset): Return the offset as a poly_uint64 rather than
197 an unsigned HOST_WIDE_INT.
198 (record_group_use, split_address_groups): Track polynomial offsets.
199 (add_iv_candidate_for_use): Likewise.
200 (addr_offset_valid_p): Take the offset as a poly_int64 rather
201 than a HOST_WIDE_INT.
202 (strip_offset_1): Return the offset as a poly_int64 rather than
205 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
206 Alan Hayward <alan.hayward@arm.com>
207 David Sherwood <david.sherwood@arm.com>
209 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
210 rather than a HOST_WIDE_INT.
211 * tree.c (get_binfo_at_offset): Likewise.
213 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
214 Alan Hayward <alan.hayward@arm.com>
215 David Sherwood <david.sherwood@arm.com>
217 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
218 rather than a HOST_WIDE_INT.
219 * tree-sra.c (build_ref_for_offset): Likewise.
221 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
222 Alan Hayward <alan.hayward@arm.com>
223 David Sherwood <david.sherwood@arm.com>
225 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
227 * tree.c (mem_ref_offset): Likewise.
228 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
229 * builtins.c (get_object_alignment_2): Likewise.
230 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
231 * gimple-fold.c (get_base_constructor): Likewise.
232 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
233 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
235 * ipa-polymorphic-call.c
236 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
237 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
238 (get_ancestor_addr_info): Likewise.
239 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
240 * match.pd: Likewise.
241 * tree-data-ref.c (dr_analyze_innermost): Likewise.
242 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
243 * tree-eh.c (tree_could_trap_p): Likewise.
244 * tree-object-size.c (addr_object_size): Likewise.
245 * tree-ssa-address.c (copy_ref_info): Likewise.
246 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
247 (indirect_refs_may_alias_p): Likewise.
248 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
249 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
250 (non_rewritable_mem_ref_base): Likewise.
251 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
252 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
253 * varasm.c (decode_addr_const): Likewise.
255 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
256 Alan Hayward <alan.hayward@arm.com>
257 David Sherwood <david.sherwood@arm.com>
259 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
260 offsets for MEM_REFs.
262 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
263 Alan Hayward <alan.hayward@arm.com>
264 David Sherwood <david.sherwood@arm.com>
266 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
267 as a poly_uint64 rather than an unsigned int.
268 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
270 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
271 Alan Hayward <alan.hayward@arm.com>
272 David Sherwood <david.sherwood@arm.com>
274 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
276 (decode_addr_const): Update accordingly.
278 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
279 Alan Hayward <alan.hayward@arm.com>
280 David Sherwood <david.sherwood@arm.com>
282 * tree.h (bit_field_size, bit_field_offset): New functions.
283 * hsa-gen.c (gen_hsa_addr): Use them.
284 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
285 (simplify_vector_constructor): Likewise.
286 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
287 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
288 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
289 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
292 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
293 Alan Hayward <alan.hayward@arm.com>
294 David Sherwood <david.sherwood@arm.com>
296 * expr.h (emit_group_load, emit_group_load_into_temps)
297 (emit_group_store): Take the size as a poly_int64 rather than an int.
298 * expr.c (emit_group_load_1, emit_group_load): Likewise.
299 (emit_group_load_into_temp, emit_group_store): Likewise.
301 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
302 Alan Hayward <alan.hayward@arm.com>
303 David Sherwood <david.sherwood@arm.com>
305 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
306 an unsigned int to a poly_uint64.
307 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
308 sizes as poly_uint64s rather than unsigned ints.
309 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
312 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
313 Alan Hayward <alan.hayward@arm.com>
314 David Sherwood <david.sherwood@arm.com>
316 * lra-constraints.c (emit_inc): Change inc_amount from an int
319 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
320 Alan Hayward <alan.hayward@arm.com>
321 David Sherwood <david.sherwood@arm.com>
323 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
325 (add_stack_var, stack_var_cmp, partition_stack_vars)
326 (dump_stack_var_partition): Update accordingly.
327 (alloc_stack_frame_space): Take the size as a poly_int64 rather
328 than a HOST_WIDE_INT.
329 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
330 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
331 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
332 rather than a HOST_WIDE_INT.
334 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
335 Alan Hayward <alan.hayward@arm.com>
336 David Sherwood <david.sherwood@arm.com>
338 * target.def (return_pops_args): Treat both the input and output
339 sizes as poly_int64s rather than HOST_WIDE_INTS.
340 * targhooks.h (default_return_pops_args): Update accordingly.
341 * targhooks.c (default_return_pops_args): Likewise.
342 * doc/tm.texi: Regenerate.
343 * emit-rtl.h (incoming_args): Change pops_args, size and
344 outgoing_args_size from int to poly_int64_pod.
345 * function.h (expr_status): Change x_pending_stack_adjust and
346 x_stack_pointer_delta from int to poly_int64.
347 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
348 (ARGS_SIZE_RTX): Update accordingly.
349 * calls.c (highest_outgoing_arg_in_use): Change from int to
351 (stack_usage_watermark, stored_args_watermark): New variables.
352 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
353 (emit_call_1): Change the stack_size and rounded_stack_size
354 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
356 (save_fixed_argument_area): Check stack_usage_watermark.
357 (initialize_argument_information): Change old_pending_adj from
358 a HOST_WIDE_INT * to a poly_int64_pod *.
359 (compute_argument_block_size): Return the size as a poly_int64
361 (finalize_must_preallocate): Track polynomial argument sizes.
362 (compute_argument_addresses): Likewise.
363 (internal_arg_pointer_based_exp): Track polynomial offsets.
364 (mem_overlaps_already_clobbered_arg_p): Rename to...
365 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
366 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
367 Check stored_args_used_watermark.
368 (load_register_parameters): Update accordingly.
369 (check_sibcall_argument_overlap_1): Likewise.
370 (combine_pending_stack_adjustment_and_call): Take the unadjusted
371 args size as a poly_int64 rather than an int. Return a bool
372 indicating whether the optimization was possible and return
373 the new adjustment by reference.
374 (check_sibcall_argument_overlap): Track polynomail argument sizes.
375 Update stored_args_watermark.
376 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
377 (expand_call): Likewise. Maintain stack_usage_watermark and
378 stored_args_watermark. Update calls to
379 combine_pending_stack_adjustment_and_call.
380 (emit_library_call_value_1): Handle polynomial argument sizes.
381 Call stack_region_maybe_used_p and mark_stack_region_used.
382 Maintain stack_usage_watermark.
383 (store_one_arg): Likewise. Update call to
384 mem_overlaps_already_clobbered_arg_p.
385 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
387 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
388 * config/microblaze/microblaze.c (microblaze_function_prologue):
390 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
391 TARGET_RETURN_POPS_ARGS interface.
392 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
394 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
395 * config/i386/i386.c (ix86_return_pops_args): Update for new
396 TARGET_RETURN_POPS_ARGS interface.
397 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
398 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
399 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
400 TARGET_RETURN_POPS_ARGS interface.
401 * config/vax/vax.c (vax_return_pops_args): Likewise.
402 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
403 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
404 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
405 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
406 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
407 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
408 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
409 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
410 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
411 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
412 and x_stack_pointer_delta from int to poly_int64.
413 * dojump.c (do_pending_stack_adjust): Update accordingly.
414 * explow.c (allocate_dynamic_stack_space): Handle polynomial
415 stack_pointer_deltas.
416 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
417 (pad_to_arg_alignment): Track polynomial offsets.
418 (assign_parm_find_stack_rtl): Likewise.
419 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
420 * toplev.c (output_stack_usage): Update reference to
421 current_function_pushed_stack_size.
423 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
424 Alan Hayward <alan.hayward@arm.com>
425 David Sherwood <david.sherwood@arm.com>
427 * function.c (in_arg_offset, var_offset, dynamic_offset)
428 (out_arg_offset, cfa_offset): Change from int to poly_int64.
429 (instantiate_new_reg): Return the new offset as a poly_int64_pod
430 rather than a HOST_WIDE_INT.
431 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
432 (instantiate_virtual_regs_in_insn): Likewise.
434 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
435 Alan Hayward <alan.hayward@arm.com>
436 David Sherwood <david.sherwood@arm.com>
438 * rtl.h (get_args_size, add_args_size_note): New functions.
439 (find_args_size_adjust): Return a poly_int64 rather than a
441 (fixup_args_size_notes): Likewise. Make the same change to the
442 end_args_size parameter.
443 * rtlanal.c (get_args_size, add_args_size_note): New functions.
444 * builtins.c (expand_builtin_trap): Use add_args_size_note.
445 * calls.c (emit_call_1): Likewise.
446 * explow.c (adjust_stack_1): Likewise.
447 * cfgcleanup.c (old_insns_match_p): Update use of
448 find_args_size_adjust.
449 * combine.c (distribute_notes): Track polynomial arg sizes.
450 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
451 end_true_args_size, beg_delay_args_size and end_delay_args_size
452 from HOST_WIDE_INT to poly_int64.
453 (add_cfi_args_size): Take the args_size as a poly_int64 rather
454 than a HOST_WIDE_INT.
455 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
456 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
457 polynomial arg sizes.
458 * emit-rtl.c (try_split): Use get_args_size.
459 * recog.c (peep2_attempt): Likewise.
460 * reload1.c (reload_as_needed): Likewise.
461 * expr.c (find_args_size_adjust): Return the adjustment as a
462 poly_int64 rather than a HOST_WIDE_INT.
463 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
464 to a poly_int64 and change the return type in the same way.
465 (emit_single_push_insn): Track polynomial arg sizes.
467 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
468 Alan Hayward <alan.hayward@arm.com>
469 David Sherwood <david.sherwood@arm.com>
471 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
472 from HOST_WIDE_INT to poly_int64.
473 * expr.c (push_block, emit_push_insn): Likewise.
475 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
476 Alan Hayward <alan.hayward@arm.com>
477 David Sherwood <david.sherwood@arm.com>
479 * function.h (frame_space): Change start and length from HOST_WIDE_INT
481 (get_frame_size): Return the size as a poly_int64 rather than a
483 (frame_offset_overflow): Take the offset as a poly_int64 rather
484 than a HOST_WIDE_INT.
485 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
486 (assign_stack_temp): Likewise for the size.
487 * function.c (get_frame_size): Return a poly_int64 rather than
489 (frame_offset_overflow): Take the offset as a poly_int64 rather
490 than a HOST_WIDE_INT.
491 (try_fit_stack_local): Take the start, length and size as poly_int64s
492 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
493 rather than a HOST_WIDE_INT.
494 (add_frame_space): Take the start and end as poly_int64s rather than
496 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
497 (assign_stack_temp): Likewise for the size.
498 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
500 (find_temp_slot_from_address): Handle polynomial offsets.
501 (combine_temp_slots): Likewise.
502 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
504 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
505 poly_int64 rather than a HOST_WIDE_INT.
506 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
507 than a HOST_WIDE_INT.
508 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
509 polynomial frame offsets.
510 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
511 as a poly_int64 rather than an int.
512 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
513 * config/v850/v850-protos.h (compute_frame_size): Likewise.
514 * config/v850/v850.c (compute_frame_size): Likewise.
515 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
516 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
517 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
518 * config/pa/pa.c (pa_compute_frame_size): Likewise.
519 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
520 rather than a HOST_WIDE_INT.
521 * explow.c (get_dynamic_stack_base): Likewise.
522 * final.c (final_start_function): Use the constant lower bound
523 of the frame size for -Wframe-larger-than.
524 * ira.c (do_reload): Adjust for new get_frame_size return type.
525 * lra.c (lra): Likewise.
526 * reload1.c (reload): Likewise.
527 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
528 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
529 * rtlanal.c (get_initial_register_offset): Return the offset as
530 a poly_int64 rather than a HOST_WIDE_INT.
532 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
533 Alan Hayward <alan.hayward@arm.com>
534 David Sherwood <david.sherwood@arm.com>
536 * reload1.c (elim_table): Change initial_offset, offset and
537 previous_offset from HOST_WIDE_INT to poly_int64_pod.
538 (offsets_at): Change the target array's element type from
539 HOST_WIDE_INT to poly_int64_pod.
540 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
541 (elimination_costs_in_insn, update_eliminable_offsets)
542 (verify_initial_elim_offsets, set_offsets_for_label)
543 (init_eliminable_invariants): Update after above changes.
545 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
546 Alan Hayward <alan.hayward@arm.com>
547 David Sherwood <david.sherwood@arm.com>
549 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
550 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
551 (decomposition): Change start and end from HOST_WIDE_INT
553 (decompose, immune_p): Update accordingly.
554 (find_inc_amount): Return a poly_int64 rather than an int.
555 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
558 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
559 Alan Hayward <alan.hayward@arm.com>
560 David Sherwood <david.sherwood@arm.com>
562 * tree.h (get_inner_reference): Return the bitsize and bitpos
563 as poly_int64_pods rather than HOST_WIDE_INT.
564 * fold-const.h (ptr_difference_const): Return the pointer difference
565 as a poly_int64_pod rather than a HOST_WIDE_INT.
566 * expr.c (get_inner_reference): Return the bitsize and bitpos
567 as poly_int64_pods rather than HOST_WIDE_INT.
568 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
570 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
571 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
572 (optimize_bit_field_compare): Update call to get_inner_reference.
573 (decode_field_reference): Likewise.
574 (fold_unary_loc): Track polynomial offsets and sizes.
575 (split_address_to_core_and_offset): Return the bitpos as a
576 poly_int64_pod rather than a HOST_WIDE_INT.
577 (ptr_difference_const): Likewise for the pointer difference.
578 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
579 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
580 * dbxout.c (dbxout_expand_expr): Likewise.
581 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
582 (loc_list_from_tree_1, fortran_common): Likewise.
583 * gimple-laddress.c (pass_laddress::execute): Likewise.
584 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
585 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
586 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
587 * tree-affine.c (tree_to_aff_combination): Likewise.
588 (get_inner_reference_aff): Likewise.
589 * tree-data-ref.c (split_constant_offset_1): Likewise.
590 (dr_analyze_innermost): Likewise.
591 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
592 * tree-sra.c (ipa_sra_check_caller): Likewise.
593 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
594 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
595 (instrument_bool_enum_load, instrument_object_size): Likewise.
596 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
597 to get_inner_reference.
598 * hsa-gen.c (gen_hsa_addr): Likewise.
599 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
600 * tsan.c (instrument_expr): Likewise.
601 * match.pd: Update call to ptr_difference_const.
603 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
604 Alan Hayward <alan.hayward@arm.com>
605 David Sherwood <david.sherwood@arm.com>
607 * fold-const.c (fold_comparison): Track sizes and offsets as
608 poly_int64s rather than HOST_WIDE_INTs when folding address
611 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
612 Alan Hayward <alan.hayward@arm.com>
613 David Sherwood <david.sherwood@arm.com>
615 * expr.h (get_bit_range): Return the bitstart and bitend as
616 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
617 as a poly_int64 rather than a HOST_WIDE_INT.
618 * expr.c (get_bit_range): Likewise.
619 (expand_assignment): Update call accordingly.
620 * fold-const.c (optimize_bit_field_compare): Likewise.
622 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
623 Alan Hayward <alan.hayward@arm.com>
624 David Sherwood <david.sherwood@arm.com>
626 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
627 (aarch64_legitimate_address_p): Use it instead of an rtx code,
628 as an optional final parameter.
629 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
630 (aarch64_legitimate_address_p): Likewise.
631 (aarch64_print_address_internal): Take an aarch64_addr_query_type
632 instead of an rtx code.
633 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
634 (aarch64_legitimate_address_hook_p): Likewise.
635 (aarch64_print_ldpstp_address): Likewise.
636 (aarch64_print_operand_address): Likewise.
637 (aarch64_address_cost): Likewise.
638 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
639 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
640 (aarch64_mem_pair_lanes_operand): Likewise.
642 2017-12-20 Richard Biener <rguenther@suse.de>
644 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
645 update_stmt_if_modified.
647 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
649 PR tree-optimization/83491
650 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
651 before walking uses. Improve coding style and comments.
653 2017-12-20 Tom de Vries <tom@codesourcery.com>
655 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
657 2017-12-20 Jakub Jelinek <jakub@redhat.com>
660 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
662 (pass_ipa_free_fn_summary): Add small_p private data member,
663 initialize to false in the ctor.
664 (pass_ipa_free_fn_summary::clone,
665 pass_ipa_free_fn_summary::set_pass_param,
666 pass_ipa_free_fn_summary::gate): New methods.
667 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
668 | TODO_dump_symtab if small_p.
669 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
670 entry and add another instance of the pass with false parm after
672 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
673 ipa_free_fn_summary here.
675 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
677 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
679 2017-12-20 Martin Sebor <msebor@redhat.com>
682 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
683 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
684 (expand_builtin_strncmp): Same.
686 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
689 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
690 allowed debug stmts before labels.
691 (expand_gimple_basic_block): Likewise.
692 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
693 * gimple-iterator.h (gsi_after_labels): Likewise.
694 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
695 rename reused variable, and simplify using gsi_move_before.
696 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
697 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
698 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
699 (gimple_verify_flow_info, gimple_block_label): Likewise.
700 (make_blocks): Move debug markers after adjacent labels.
701 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
702 allowed debug insns outside blocks.
703 * df-scan.c (df_insn_delete): Likewise.
704 * lra-constraints.c (update_ebb_live_info): Likewise.
705 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
706 (vt_initialize, delete_vta_debug_insns): Likewise.
707 (reemit_marker_as_note): Drop BB parm. Adjust callers.
709 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
710 Alan Hayward <alan.hayward@arm.com>
711 David Sherwood <david.sherwood@arm.com>
713 * poly-int-types.h (round_down_to_byte_boundary): New macro.
714 (round_up_to_byte_boundary): Likewise.
715 * expr.h (get_bit_range): Add temporary shim.
716 * gimple-ssa-store-merging.c (store_operand_info): Change the
717 bitsize, bitpos, bitregion_start and bitregion_end fields from
718 unsigned HOST_WIDE_INT to poly_uint64.
719 (merged_store_group): Likewise load_align_base.
720 (compatible_load_p, compatible_load_p): Update accordingly.
721 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
722 (split_group, imm_store_chain_info::output_merged_store): Likewise.
723 (mem_valid_for_store_merging): Return the bitsize, bitpos,
724 bitregion_start and bitregion_end as poly_uint64s rather than
725 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
726 (handled_load): Take the bitsize, bitpos,
727 bitregion_start and bitregion_end as poly_uint64s rather than
728 unsigned HOST_WIDE_INTs.
729 (pass_store_merging::process_store): Update call to
730 mem_valid_for_store_merging.
732 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
733 Alan Hayward <alan.hayward@arm.com>
734 David Sherwood <david.sherwood@arm.com>
736 * builtins.c (get_object_alignment_2): Track polynomial offsets
737 and sizes. Update the alignment handling.
739 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
740 Alan Hayward <alan.hayward@arm.com>
741 David Sherwood <david.sherwood@arm.com>
743 * tree.h (get_inner_reference): Add a version that returns the
744 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
745 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
746 the case in which bitpos is not associated with the first byte.
748 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
749 Alan Hayward <alan.hayward@arm.com>
750 David Sherwood <david.sherwood@arm.com>
752 * tree-affine.h (get_inner_reference_aff): Return the size as a
754 * tree-affine.c (get_inner_reference_aff): Likewise.
755 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
756 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
758 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
759 Alan Hayward <alan.hayward@arm.com>
760 David Sherwood <david.sherwood@arm.com>
762 * fold-const.c (pointer_may_wrap_p): Take the offset as a
763 HOST_WIDE_INT rather than a poly_int64.
765 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
766 Alan Hayward <alan.hayward@arm.com>
767 David Sherwood <david.sherwood@arm.com>
769 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
770 HOST_WIDE_INT to poly_int64_pod.
771 (perform_symbolic_merge): Update accordingly.
772 (bswap_replace): Likewise.
774 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
775 Alan Hayward <alan.hayward@arm.com>
776 David Sherwood <david.sherwood@arm.com>
778 * tree-affine.h (aff_tree::offset): Change from widest_int
780 (wide_int_ext_for_comb): Delete.
781 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
782 constants as poly_widest_int rather than widest_int.
783 (aff_combination_constant_multiple_p): Return the multiplier
784 as a poly_widest_int.
785 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
787 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
788 version static and add an overload for poly_widest_int.
789 (aff_combination_const, aff_combination_add_cst)
790 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
791 the constants as poly_widest_int rather than widest_int.
792 (tree_to_aff_combination): Generalize INTEGER_CST case to
794 (aff_combination_to_tree): Track offsets as poly_widest_ints.
795 (aff_combination_add_product, aff_combination_mult): Handle
797 (aff_combination_constant_multiple_p): Return the multiplier
798 as a poly_widest_int.
799 * tree-predcom.c (determine_offset): Return the offset as a
801 (split_data_refs_to_components, suitable_component_p): Update
803 (valid_initializer_p): Update call to
804 aff_combination_constant_multiple_p.
805 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
806 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
807 as a poly_int64 rather than a HOST_WIDE_INT.
808 (get_address_cost): Handle polynomial offsets.
809 (iv_elimination_compare_lt): Likewise.
810 (rewrite_use_nonlinear_expr): Likewise.
812 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
813 Alan Hayward <alan.hayward@arm.com>
814 David Sherwood <david.sherwood@arm.com>
816 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
817 as a poly_int64_pod rather than a HOST_WIDE_INT.
818 (get_addr_base_and_unit_offset): Likewise.
819 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
820 (get_addr_base_and_unit_offset): Likewise.
821 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
822 to poly_int64 in example.
823 * fold-const.c (fold_binary_loc): Update call to
824 get_addr_base_and_unit_offset.
825 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
826 (maybe_canonicalize_mem_ref_addr): Likewise.
827 (gimple_fold_stmt_to_constant_1): Likewise.
828 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
830 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
831 * match.pd: Likewise.
832 * omp-low.c (lower_omp_target): Likewise.
833 * tree-sra.c (build_ref_for_offset): Likewise.
834 (build_debug_ref_for_model): Likewise.
835 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
836 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
837 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
838 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
839 (constant_pointer_difference): Likewise.
840 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
841 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
842 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
843 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
844 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
845 (set_ssa_val_to): Likewise.
846 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
847 (maybe_diag_stxncpy_trunc): Likewise.
848 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
849 * tree.c (build_simple_mem_ref_loc): Likewise.
850 (array_at_struct_end_p): Likewise.
852 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
853 Alan Hayward <alan.hayward@arm.com>
854 David Sherwood <david.sherwood@arm.com>
856 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
857 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
858 (get_ref_base_and_extent_hwi): Declare.
859 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
860 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
861 (get_ref_base_and_extent_hwi): New function.
862 * cfgexpand.c (expand_debug_expr): Update call to
863 get_ref_base_and_extent.
864 * dwarf2out.c (add_var_loc_to_decl): Likewise.
865 * gimple-fold.c (get_base_constructor): Return the offset as a
866 poly_int64_pod rather than a HOST_WIDE_INT.
867 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
868 * ipa-polymorphic-call.c
869 (ipa_polymorphic_call_context::set_by_invariant)
870 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
871 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
872 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
873 rather than get_ref_base_and_extent.
874 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
875 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
876 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
878 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
879 call to get_ref_base_and_extent.
880 * tree-sra.c (create_access, get_access_for_expr): Likewise.
881 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
882 (stmt_kills_ref_p): Likewise.
883 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
884 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
886 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
887 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
888 when calling native_encode_expr.
889 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
890 call to get_ref_base_and_extent.
891 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
892 get_ref_base_and_extent.
893 * var-tracking.c (track_expr_p): Likewise.
895 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
896 Alan Hayward <alan.hayward@arm.com>
897 David Sherwood <david.sherwood@arm.com>
899 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
900 HOST_WIDE_INT to poly_int64_pod.
901 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
902 polynomail parameter offsets.
904 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
905 Alan Hayward <alan.hayward@arm.com>
906 David Sherwood <david.sherwood@arm.com>
908 * gengtype.c (main): Handle poly_int64_pod.
909 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
910 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
911 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
912 from HOST_WIDE_INT to poly_int64_pod.
913 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
914 (copy_cfa): New function.
915 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
916 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
917 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
918 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
919 (initial_return_save): Treat offsets as poly_ints.
920 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
921 offset is nonconstant.
922 (reg_save): Take the offset as a poly_int64. Fall back to
923 DW_CFA_expression for nonconstant offsets.
924 (queue_reg_save): Take the offset as a poly_int64.
925 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
927 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
928 Alan Hayward <alan.hayward@arm.com>
929 David Sherwood <david.sherwood@arm.com>
931 * rtl.h (operand_subword, operand_subword_force): Take the offset
932 as a poly_uint64 an unsigned int.
933 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
935 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
936 Alan Hayward <alan.hayward@arm.com>
937 David Sherwood <david.sherwood@arm.com>
939 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
940 'p' format code. Use INT_LIST rather than SUBREG as the example of
941 a code with an XINT and an XEXP. Remove the implication that
942 accessing an rtx field using XINT is expected to work.
943 * rtl.def (SUBREG): Change format from "ei" to "ep".
944 * rtl.h (rtunion::rt_subreg): New field.
945 (XCSUBREG): New macro.
946 (SUBREG_BYTE): Use it.
947 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
948 Update constructor accordingly.
949 (subreg_shape::operator ==): Update accordingly.
950 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
951 than an unsigned int.
952 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
953 a poly_uint64 rather than an unsigned int.
954 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
955 than an unsigned int.
956 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
957 (subreg_size_highpart_offset): Return a poly_uint64 rather than
958 an unsigned int. Take the sizes as poly_uint64s.
959 (subreg_offset_from_lsb): Return a poly_uint64 rather than
960 an unsigned int. Take the shift as a poly_uint64 rather than
962 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
963 as a poly_uint64 rather than an unsigned int.
964 (simplify_subreg_regno): Likewise.
965 (byte_lowpart_offset): Return the memory offset as a poly_int64
967 (subreg_memory_offset): Likewise. Take the subreg offset as a
968 poly_uint64 rather than an unsigned int.
969 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
970 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
971 poly_uint64 rather than an unsigned int.
972 * rtl.c (rtx_format): Describe 'p' in comment.
973 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
974 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
975 offset as a poly_uint64 rather than an unsigned int.
976 (byte_lowpart_offset): Return the memory offset as a poly_int64
978 (subreg_memory_offset): Likewise. Take the subreg offset as a
979 poly_uint64 rather than an unsigned int.
980 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
981 mode sizes as poly_uint64s rather than unsigned ints. Return a
982 poly_uint64 rather than an unsigned int.
983 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
984 (copy_insn_1): Handle 'p'.
985 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
986 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
987 an unsigned int. Return the shift in the same way.
988 (subreg_lsb): Return the shift as a poly_uint64 rather than an
990 (subreg_size_offset_from_lsb): Take the sizes and shift as
991 poly_uint64s rather than unsigned ints. Return the offset as
993 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
994 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
996 * rtlhash.c (add_rtx): Handle 'p'.
997 * genemit.c (gen_exp): Likewise.
998 * gengenrtl.c (type_from_format, gendef): Likewise.
999 * gensupport.c (subst_pattern_match, get_alternatives_number)
1000 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1001 (subst_dup): Likewise.
1002 * gengtype.c (adjust_field_rtx_def): Likewise.
1003 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1004 (match_pattern_2): Likewise.
1005 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1006 (rtx_test::subreg_field): New function.
1007 (operator ==, safe_to_hoist_p, transition_parameter_type)
1008 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1009 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1010 * genpeep.c (match_rtx): Likewise.
1011 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1012 (rtx_writer::print_rtx_operand): Handle 'p'.
1013 (print_value): Handle SUBREG.
1014 * read-rtl.c (apply_int_iterator): Likewise.
1015 (rtx_reader::read_rtx_operand): Handle 'p'.
1016 * alias.c (rtx_equal_for_memref_p): Likewise.
1017 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1018 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1020 * calls.c (expand_call): Likewise.
1021 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1022 (make_extraction, gen_lowpart_for_combine): Likewise.
1023 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1025 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1026 rather than an unsigned int. Treat subreg offsets as poly_ints.
1027 (exp_equiv_p): Handle 'p'.
1028 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1029 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1030 * dse.c (find_shift_sequence): Likewise.
1031 * dwarf2out.c (rtl_for_decl_location): Likewise.
1032 * expmed.c (extract_low_bits): Likewise.
1033 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1034 (expand_expr_real_2): Likewise.
1035 * final.c (alter_subreg): Likewise.
1036 (leaf_renumber_regs_insn): Handle 'p'.
1037 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1038 Treat subreg offsets as poly_ints.
1039 * fwprop.c (forward_propagate_and_simplify): Likewise.
1040 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1041 * ira.c (get_subreg_tracking_sizes): Likewise.
1042 * ira-conflicts.c (go_through_subreg): Likewise.
1043 * ira-lives.c (process_single_reg_class_operands): Likewise.
1044 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1045 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1046 as a poly_uint64 rather than an unsigned int.
1047 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1048 subreg offsets as poly_ints.
1049 * lra-constraints.c (operands_match_p): Handle 'p'.
1050 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1051 * lra-spills.c (assign_mem_slot): Likewise.
1052 * postreload.c (move2add_valid_value_p): Likewise.
1053 * recog.c (general_operand, indirect_operand): Likewise.
1054 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1055 (copyprop_hardreg_forward_1): Likewise.
1056 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1057 (record_subregs_of_mode): Likewise.
1058 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1059 * reload.c (operands_match_p): Handle 'p'.
1060 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1061 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1062 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1063 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1064 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1065 (simplify_const_poly_int_tests<N>::run): Likewise.
1066 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1067 a poly_uint64 rather than an unsigned int.
1068 * valtrack.c (debug_lowpart_subreg): Likewise.
1069 * var-tracking.c (var_lowpart): Likewise.
1070 (loc_cmp): Handle 'p'.
1072 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1073 Alan Hayward <alan.hayward@arm.com>
1074 David Sherwood <david.sherwood@arm.com>
1076 * ira.c (get_subreg_tracking_sizes): New function.
1077 (init_live_subregs): Take an integer size rather than a register.
1078 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1079 to init_live_subregs.
1081 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1082 Alan Hayward <alan.hayward@arm.com>
1083 David Sherwood <david.sherwood@arm.com>
1085 * expr.c (store_constructor_field): Change bitsize from a
1086 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1087 HOST_WIDE_INT to a poly_int64.
1088 (store_constructor): Change size from a HOST_WIDE_INT to
1090 (store_field): Likewise bitsize and bitpos.
1092 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1093 Alan Hayward <alan.hayward@arm.com>
1094 David Sherwood <david.sherwood@arm.com>
1096 * expmed.h (store_bit_field): Change bitregion_start and
1097 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1098 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1099 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1100 (store_fixed_bit_field, store_split_bit_field): Likewise.
1101 * expr.c (store_constructor_field, store_field): Likewise.
1102 (optimize_bitfield_assignment_op): Likewise. Make the same change
1103 to bitsize and bitpos.
1104 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1105 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1106 same change in the constructor arguments.
1107 (get_best_mode): Change bitregion_start and bitregion_end from
1108 unsigned HOST_WIDE_INT to poly_uint64.
1109 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1110 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1112 (bit_field_mode_iterator::next_mode): Update for new types
1113 of m_bitregion_start and m_bitregion_end.
1114 (get_best_mode): Change bitregion_start and bitregion_end from
1115 unsigned HOST_WIDE_INT to poly_uint64.
1117 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1118 Alan Hayward <alan.hayward@arm.com>
1119 David Sherwood <david.sherwood@arm.com>
1121 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1122 accepts poly_uint64 offsets.
1123 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1124 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1125 * expmed.c (lowpart_bit_field_p): Likewise.
1126 (extract_bit_field_as_subreg): New function, split out from...
1127 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1128 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1129 extractions, check that BITSIZE matches the size of the extracted
1130 value and that BITNUM is an exact multiple of that size.
1131 If all else fails, try forcing the value into memory if
1132 BITNUM is variable, and adjusting the address so that the
1133 offset is constant. Split the part that can only handle constant
1134 bitsize and bitnum out into...
1135 (extract_integral_bit_field): ...this new function.
1136 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1137 rather than unsigned HOST_WIDE_INTs.
1139 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1140 Alan Hayward <alan.hayward@arm.com>
1141 David Sherwood <david.sherwood@arm.com>
1143 * expmed.h (store_bit_field): Take bitsize and bitnum as
1144 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1145 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1146 that returns the byte size.
1147 (store_bit_field_1): Take bitsize and bitnum as
1148 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1149 to simple_mem_bitfield_p. Split the part that can only handle
1150 constant bitsize and bitnum out into...
1151 (store_integral_bit_field): ...this new function.
1152 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1153 than unsigned HOST_WIDE_INTs.
1154 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1156 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1157 Alan Hayward <alan.hayward@arm.com>
1158 David Sherwood <david.sherwood@arm.com>
1160 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1161 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1163 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1164 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1165 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1166 offset as a poly_int64 rather than an int.
1167 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1168 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1169 * lra-constraints.c (equiv_address_substitution): Track offsets
1171 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1172 (curr_insn_transform): Handle the new form of sp_offset.
1173 * lra-eliminations.c (lra_elim_table): Change previous_offset
1174 and offset from HOST_WIDE_INT to poly_int64.
1175 (print_elim_table, update_reg_eliminate): Update accordingly.
1176 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1177 (get_elimination): Update accordingly.
1178 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1179 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1180 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1181 poly_int64 offsets generally.
1182 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1183 (mark_not_eliminable, init_elimination): Update accordingly.
1184 (remove_reg_equal_offset_note): Return a bool and pass the new
1185 offset back by pointer as a poly_int64.
1186 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1187 rather than a HOST_WIDE_INT.
1188 (do_remat): Track offsets poly_int64s.
1189 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1191 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1192 Alan Hayward <alan.hayward@arm.com>
1193 David Sherwood <david.sherwood@arm.com>
1195 * rtl.h (mem_attrs): Add a default constructor. Change size and
1196 offset from HOST_WIDE_INT to poly_int64.
1197 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1198 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1199 (widen_memory_access): Take the sizes and offsets as poly_int64s
1200 rather than HOST_WIDE_INTs.
1201 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1202 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1204 (adjust_offset_for_component_ref): Change the offset from a
1205 HOST_WIDE_INT to a poly_int64.
1206 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1207 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1208 * dce.c (find_call_stack_args): Likewise.
1209 * dse.c (record_store): Likewise.
1210 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1211 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1212 * read-rtl-function.c (test_loading_mem): Likewise.
1213 * rtlanal.c (may_trap_p_1): Likewise.
1214 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1215 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1216 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1217 (mem_attrs::mem_attrs): New function.
1218 (set_mem_attributes_minus_bitpos): Change bitpos from a
1219 HOST_WIDE_INT to poly_int64.
1220 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1221 (clear_mem_offset, clear_mem_size, change_address)
1222 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1223 initialize mem_attrs.
1224 (set_mem_offset, set_mem_size, adjust_address_1)
1225 (adjust_automodify_address_1, offset_address, widen_memory_access):
1226 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1228 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1229 Alan Hayward <alan.hayward@arm.com>
1230 David Sherwood <david.sherwood@arm.com>
1232 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1233 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1234 rather than 0 to represent an unknown size. Assert that the size
1235 is known when the mode isn't BLKmode.
1236 (may_trap_p_1): Use -1 for unknown sizes.
1237 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1239 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1240 Alan Hayward <alan.hayward@arm.com>
1241 David Sherwood <david.sherwood@arm.com>
1243 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1244 to poly_int64. Update commentary for positions_needed.large.
1245 (read_info_type): Change offset and width from HOST_WIDE_INT
1247 (set_usage_bits): Likewise.
1248 (canon_address): Return the offset as a poly_int64 rather than
1249 a HOST_WIDE_INT. Use strip_offset_and_add.
1250 (set_all_positions_unneeded, any_positions_needed_p): Use
1251 positions_needed.large to track stores with non-constant widths.
1252 (all_positions_needed_p): Likewise. Take the offset and width
1253 as poly_int64s rather than ints. Assert that rhs is nonnull.
1254 (record_store): Cope with non-constant offsets and widths.
1255 Nullify the rhs of an earlier store if we can't tell which bytes
1257 (find_shift_sequence): Take the access_size and shift as poly_int64s
1259 (get_stored_val): Take the read_offset and read_width as poly_int64s
1260 rather than HOST_WIDE_INTs.
1261 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1262 non-constant offsets and widths.
1264 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1265 Alan Hayward <alan.hayward@arm.com>
1266 David Sherwood <david.sherwood@arm.com>
1268 * inchash.h (inchash::hash::add_poly_int): New function.
1269 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1270 Use poly_int64 rather than HOST_WIDE_INT.
1271 (ao_ref::max_size_known_p): New function.
1272 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1273 rather than HOST_WIDE_INT.
1274 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1275 to temporaries until its interface is adjusted to match.
1276 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1277 (aliasing_component_refs_p, decl_refs_may_alias_p)
1278 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1279 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1280 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1282 * alias.c (ao_ref_from_mem): Likewise.
1283 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1284 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1285 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1286 (maybe_trim_complex_store, maybe_trim_constructor_store)
1287 (live_bytes_read, dse_classify_store): Likewise.
1288 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1289 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1290 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1291 (vn_reference_lookup_3): Likewise.
1292 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1294 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1295 Alan Hayward <alan.hayward@arm.com>
1296 David Sherwood <david.sherwood@arm.com>
1298 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1299 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1300 instead of ranges_overlap_p.
1302 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1303 Alan Hayward <alan.hayward@arm.com>
1304 David Sherwood <david.sherwood@arm.com>
1306 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1307 sizes as poly_int64s rather than HOST_WIDE_INTs.
1309 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1310 Alan Hayward <alan.hayward@arm.com>
1311 David Sherwood <david.sherwood@arm.com>
1313 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1314 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1315 * gimple-fold.c (fold_ctor_reference): Likewise.
1317 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1318 Alan Hayward <alan.hayward@arm.com>
1319 David Sherwood <david.sherwood@arm.com>
1321 * target.def (dwarf_poly_indeterminate_value): New hook.
1322 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1323 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1324 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1325 * doc/tm.texi: Regenerate.
1326 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1327 offset as a poly_int64.
1328 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1329 offset as a poly_int64.
1330 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1331 Take the offset as a poly_int64.
1332 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1333 (frame_pointer_fb_offset): Change to a poly_int64.
1334 (int_loc_descriptor): Take the offset as a poly_int64. Use
1335 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1336 (based_loc_descr): Take the offset as a poly_int64.
1337 Use strip_offset_and_add to handle (plus X (const)).
1338 Use new_reg_loc_descr instead of an open-coded version of the
1339 previous implementation.
1340 (mem_loc_descriptor): Handle CONST_POLY_INT.
1341 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1342 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1344 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1345 Alan Hayward <alan.hayward@arm.com>
1346 David Sherwood <david.sherwood@arm.com>
1348 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1349 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1350 * inchash.h (inchash::hash::add_poly_hwi): New function.
1351 * gengtype.c (main): Register poly_int64.
1352 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1353 offset as a poly_int.
1354 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1355 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1356 offset as a poly_int64.
1357 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1358 * print-rtl.c (print_poly_int): New function.
1359 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1361 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1363 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1364 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1365 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1366 Enforce track_offset_p for parts of a PARALLEL.
1367 (vt_add_function_parameter): Use const_offset for the final
1368 offset to track. Use get_tracked_reg_offset for the parts
1371 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1372 Alan Hayward <alan.hayward@arm.com>
1373 David Sherwood <david.sherwood@arm.com>
1375 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1376 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1377 * doc/tm.texi: Regenerate.
1378 * hooks.h (hook_bool_uint_uint_true): Delete.
1379 (hook_bool_puint64_puint64_true): Declare.
1380 * hooks.c (hook_bool_uint_uint_true): Delete.
1381 (hook_bool_puint64_puint64_true): New function.
1382 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1383 instead of unsigned ints.
1384 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1385 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1387 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1388 Alan Hayward <alan.hayward@arm.com>
1389 David Sherwood <david.sherwood@arm.com>
1391 * optabs.h (expand_operand): Add an int_value field.
1392 (create_expand_operand): Add an int_value parameter and use it
1393 to initialize the new expand_operand field.
1394 (create_integer_operand): Replace with a declaration of a function
1395 that accepts poly_int64s. Move the implementation to...
1396 * optabs.c (create_integer_operand): ...here.
1397 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1398 the mode preserves the value of int_value, instead of calling
1399 const_int_operand on the rtx. Use gen_int_mode to generate
1402 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1403 Alan Hayward <alan.hayward@arm.com>
1404 David Sherwood <david.sherwood@arm.com>
1406 * dumpfile.h (dump_dec): Declare.
1407 * dumpfile.c (dump_dec): New function.
1408 * pretty-print.h (pp_wide_integer): Turn into a function and
1409 declare a poly_int version.
1410 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1412 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1413 Alan Hayward <alan.hayward@arm.com>
1414 David Sherwood <david.sherwood@arm.com>
1416 * doc/generic.texi (POLY_INT_CST): Document.
1417 * tree.def (POLY_INT_CST): New tree code.
1418 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1419 * tree-core.h (tree_poly_int_cst): New struct.
1420 (tree_node): Add a poly_int_cst field.
1421 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1422 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1423 instead of a wide_int_ref.
1424 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1426 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1427 instead of an unsigned HOST_WIDE_INT.
1428 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1429 (ptrdiff_tree_p): Declare.
1430 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1431 extern inline implementations if the target doesn't use POLY_INT_CST.
1432 (poly_int_tree_p): New function.
1433 (wi::unextended_tree): New class.
1434 (wi::int_traits <unextended_tree>): New override.
1435 (wi::extended_tree): Add a default constructor.
1436 (wi::extended_tree::get_tree): New function.
1437 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1438 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1439 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1440 (wi::tree_to_poly_wide_ref): New typedefs.
1441 (wi::ints_for): Provide overloads for extended_tree and
1443 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1444 (wi::to_wide): New functions.
1445 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1446 * tree.c (poly_int_cst_hasher): New struct.
1447 (poly_int_cst_hash_table): New variable.
1448 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1449 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1451 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1452 (init_ttree): Initialize poly_int_cst_hash_table.
1453 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1454 a poly_int64 instead of a HOST_WIDE_INT.
1455 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1456 instead of an unsigned HOST_WIDE_INT.
1457 (wide_int_to_tree): Rename to...
1458 (wide_int_to_tree_1): ...this.
1459 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1460 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1461 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1462 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1463 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1464 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1466 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1467 (streamer_read_tree_body): Likewise.
1468 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1469 (streamer_write_tree_body): Likewise.
1470 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1471 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1472 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1473 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1474 * gimple-expr.h (is_gimple_constant): Likewise.
1475 * gimplify.c (maybe_with_size_expr): Likewise.
1476 * print-tree.c (print_node): Likewise.
1477 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1478 * tree-pretty-print.c (dump_generic_node): Likewise.
1479 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1480 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1481 * tree-vrp.c (compare_values_warnv): Likewise.
1482 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1483 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1484 (force_expr_to_var_cost): Likewise.
1485 * tree-ssa-loop.c (for_each_index): Likewise.
1486 * fold-const.h (build_invariant_address, size_int_kind): Take a
1487 poly_int64 instead of a HOST_WIDE_INT.
1488 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1489 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1491 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1492 (int_const_binop_2): New function, split out from...
1493 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1494 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1495 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1496 * gimple-ssa-strength-reduction.c (slsr_process_add)
1497 (slsr_process_mul): Check for INTEGER_CSTs before using them
1499 * stor-layout.c (bits_from_bytes): New function.
1500 (bit_from_pos): Use it.
1501 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1502 by BITS_PER_UNIT to get the TYPE_SIZE.
1503 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1504 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1506 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1507 Alan Hayward <alan.hayward@arm.com>
1508 David Sherwood <david.sherwood@arm.com>
1510 * doc/rtl.texi (const_poly_int): Document. Also document the
1511 rtl sharing behavior.
1512 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1513 * rtl.h (const_poly_int_def): New struct.
1514 (rtx_def::u): Add a cpi field.
1515 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1516 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1517 (wi::rtx_to_poly_wide_ref): New typedef
1518 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1519 (poly_int_rtx_p): New functions.
1520 (trunc_int_for_mode): Declare a poly_int64 version.
1521 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1522 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1524 (strip_offset): Declare.
1525 (strip_offset_and_add): New function.
1526 * rtl.def (CONST_POLY_INT): New rtx code.
1527 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1528 (shared_const_p): Use poly_int_rtx_p.
1529 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1531 (gen_int_shift_amount): Likewise.
1532 * emit-rtl.c (const_poly_int_hasher): New class.
1533 (const_poly_int_htab): New variable.
1534 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1535 (const_poly_int_hasher::hash): New function.
1536 (const_poly_int_hasher::equal): Likewise.
1537 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1538 (immed_wide_int_const): Rename to...
1539 (immed_wide_int_const_1): ...this and make static.
1540 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1541 instead of a wide_int_ref.
1542 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1543 (gen_lowpart_common): Handle CONST_POLY_INT.
1544 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1545 * cselib.c (cselib_hash_rtx): Likewise.
1546 * dwarf2out.c (const_ok_for_output_1): Likewise.
1547 * expr.c (convert_modes): Likewise.
1548 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1549 * rtlhash.c (add_rtx): Likewise.
1550 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1551 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1552 Handle existing CONST_POLY_INT rtxes.
1553 * expmed.h (expand_shift): Take a poly_int64 instead of a
1555 * expmed.c (expand_shift): Likewise.
1556 * rtlanal.c (strip_offset): New function.
1557 (commutative_operand_precedence): Give CONST_POLY_INT the same
1558 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1560 * rtl-tests.c (const_poly_int_tests): New struct.
1561 (rtl_tests_c_tests): Use it.
1562 * simplify-rtx.c (simplify_const_unary_operation): Handle
1564 (simplify_const_binary_operation): Likewise.
1565 (simplify_binary_operation_1): Fold additions of symbolic constants
1566 and CONST_POLY_INTs.
1567 (simplify_subreg): Handle extensions and truncations of
1569 (simplify_const_poly_int_tests): New struct.
1570 (simplify_rtx_c_tests): Use it.
1571 * wide-int.h (storage_ref): Add default constructor.
1572 (wide_int_ref_storage): Likewise.
1573 (trailing_wide_ints): Use GTY((user)).
1574 (trailing_wide_ints::operator[]): Add a const version.
1575 (trailing_wide_ints::get_precision): New function.
1576 (trailing_wide_ints::extra_size): Likewise.
1578 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1579 Alan Hayward <alan.hayward@arm.com>
1580 David Sherwood <david.sherwood@arm.com>
1582 * emit-rtl.h (gen_int_shift_amount): Declare.
1583 * emit-rtl.c (gen_int_shift_amount): New function.
1584 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1586 * calls.c (shift_return_value): Likewise.
1587 * cse.c (fold_rtx): Likewise.
1588 * dse.c (find_shift_sequence): Likewise.
1589 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1590 (expand_shift, expand_smod_pow2): Likewise.
1591 * lower-subreg.c (shift_cost): Likewise.
1592 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1593 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1594 (expand_vec_perm_var): Likewise.
1595 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1596 (simplify_binary_operation_1): Likewise.
1597 * combine.c (try_combine, find_split_point, force_int_to_mode)
1598 (simplify_shift_const_1, simplify_shift_const): Likewise.
1599 (change_zero_ext): Likewise. Use simplify_gen_binary.
1601 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1603 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1605 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1607 * doc/invoke.texi (ARM Options): Document accepted extension options
1608 for -march=armv8.3-a.
1610 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1613 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1614 or --with-float=softfp, set the default CPU to arm10e.
1616 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1618 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1619 * config/visium/predicates.md (const_shift_operand): Likewise.
1620 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1621 (visium_legitimize_reload_address): Likewise.
1623 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1626 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1628 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1631 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1633 2017-12-20 Julia Koval <julia.koval@intel.com>
1635 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1636 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1637 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1638 * config.gcc: Include vpclmulqdqintrin.h.
1639 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1640 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1642 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1643 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1644 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1645 * config/i386/i386.c (isa2_opts): Add -mcx16.
1646 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1647 (ix86_option_override_internal): Move mcx16 to flags2.
1648 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1649 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1650 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1651 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1652 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1653 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1654 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1655 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1656 * doc/invoke.texi: Add -mvpclmulqdq.
1658 2017-12-20 Tom de Vries <tom@codesourcery.com>
1661 * config/i386/i386.c (ix86_static_chain): Move
1662 DECL_STATIC_CHAIN test ...
1663 * calls.c (rtx_for_static_chain): ... here. New function.
1664 * calls.h (rtx_for_static_chain): Declare.
1665 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1666 instead of targetm.calls.static_chain.
1667 * df-scan.c (df_get_entry_block_def_set): Same.
1669 2017-12-19 Tom de Vries <tom@codesourcery.com>
1671 PR tree-optimization/83493
1672 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1675 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1677 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1678 inform with hardcoded english plural handling.
1680 2017-12-18 Jeff Law <law@redhat.com>
1682 PR tree-optimization/83477
1683 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1684 a non-virtual PHI, always push a new range.
1686 2017-12-19 Martin Sebor <msebor@redhat.com>
1689 * builtins.c (compute_objsize): Handle non-constant offsets.
1691 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1693 PR tree-optimization/83444
1694 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1695 character load case, if get_stridx on MEM_REF's operand doesn't
1696 look usable, retry with get_addr_stridx.
1698 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1701 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1702 (variable_tracking_main_1): Keep markers even when VTA fails.
1705 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1706 even if there are markers before it.
1707 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1709 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1711 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1712 typical order conditions.
1713 * sel-sched.c (extract_new_fences_from): Likewise.
1714 * config/visium/constraints.md (J, K, L): Likewise.
1715 * config/visium/predicates.md (const_shift_operand): Likewise.
1716 * config/visium/visium.c (visium_legitimize_address,
1717 visium_legitimize_reload_address): Likewise.
1718 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1719 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1720 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1721 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1722 SET_DUMP_DETAIL): Likewise.
1723 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1724 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1725 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1726 avr_set_core_architecture, avr_set_current_function,
1727 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1728 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1729 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1730 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1731 avr_map_decompose, avr_fold_builtin): Likewise.
1732 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1733 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1734 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1735 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1736 * config/m32c/m32c.c (m32c_conditional_register_usage,
1737 m32c_address_cost): Likewise.
1738 * config/m32c/predicates.md (shiftcount_operand,
1739 longshiftcount_operand): Likewise.
1740 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1741 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1742 can_use_cdx_ldstw): Likewise.
1743 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1744 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1746 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1747 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1748 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1749 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1750 * config/vax/vax.c (adjacent_operands_p): Likewise.
1751 * config/ft32/constraints.md (L, b, KA): Likewise.
1752 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1754 * cfgexpand.c (expand_stack_alignment): Likewise.
1755 * gcse.c (insert_expr_in_table): Likewise.
1756 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1757 * cgraphunit.c (cgraph_node::expand): Likewise.
1758 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1759 * emit-rtl.c (add_insn): Likewise.
1760 * input.c (dump_location_info): Likewise.
1761 * passes.c (NEXT_PASS): Likewise.
1762 * read-rtl-function.c (parse_note_insn_name,
1763 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1765 * sched-rgn.c (sched_rgn_init): Likewise.
1766 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1767 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1768 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
1770 * explow.c (eliminate_constant_term): Likewise.
1771 * final.c (leaf_renumber_regs_insn): Likewise.
1772 * cfgrtl.c (print_rtl_with_bb): Likewise.
1773 * genhooks.c (emit_init_macros): Likewise.
1774 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1775 * tree-data-ref.c (conflict_fn): Likewise.
1776 * selftest.c (assert_streq): Likewise.
1777 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1778 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1779 fold_binary_loc, multiple_of_p): Likewise.
1780 * reload.c (push_reload, find_equiv_reg): Likewise.
1781 * et-forest.c (et_nca, et_below): Likewise.
1782 * dbxout.c (dbxout_symbol_location): Likewise.
1783 * reorg.c (relax_delay_slots): Likewise.
1784 * dojump.c (do_compare_rtx_and_jump): Likewise.
1785 * gengtype-parse.c (type): Likewise.
1786 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1787 simplify_const_relational_operation): Likewise.
1788 * reload1.c (do_output_reload): Likewise.
1789 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1790 * gengtype.c (type_for_name): Likewise.
1791 * gimple-ssa-sprintf.c (format_directive): Likewise.
1793 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1796 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1797 accessing it. Adjust comment.
1799 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1802 * predict.c (zero_one_minusone): New function.
1803 (apply_return_prediction): Avoid return prediction for functions
1804 returning only -1, 0 and 1 values, unless they only return -1 and 0
1807 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1809 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1810 fall-through warning.
1811 (arc_compute_frame_size): Remove unused variables.
1812 (arc_print_operand): Fix fprintif format.
1813 (arc_can_follow_jump): Clean fall-through warning.
1815 2017-12-19 Marek Polacek <polacek@redhat.com>
1818 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1821 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1823 * config/arc/arc.c (overriderregs): New variable.
1824 (arc_override_options): Track fixed/call saved/call options.
1825 (arc_conditional_register_usage): Check against overrideregs
1826 variable whenever we change register properties.
1828 2017-12-19 Nathan Sidwell <nathan@acm.org>
1830 * opts.c (finish_options): Don't prefix dump_base_name if it
1831 already contains directories.
1833 2017-12-19 Martin Liska <mliska@suse.cz>
1835 PR rtl-optimization/82675
1836 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1837 more element in sbitmap.
1839 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1841 * gimplify.c (gimplify_expr): Use error_operand_p.
1843 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1846 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1849 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1851 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1852 the jump is to a label.
1854 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1856 PR tree-optimization/83444
1857 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1862 * ipa-inline.c (flatten_remove_node_hook): New function.
1863 (ipa_inline): Keep only nodes with flatten attribute at the end of
1864 the array in the order from ipa_reverse_postorder, only walk that
1865 portion of array for flattening, if there is more than one such
1866 node, temporarily register a removal hook and ignore removed nodes.
1868 PR tree-optimization/80631
1869 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1870 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1871 IFN_REDUC_MAX or IFN_REDUC_MIN.
1873 2017-12-18 Jeff Law <law@redhat.com>
1875 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1876 record symbolic equivalences from backedges in the CFG.
1879 2017-11-19 Jeff Law <law@redhat.com>
1881 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1882 of degenerates resulting from ignoring an edge.
1884 2017-12-18 Martin Sebor <msebor@redhat.com>
1887 PR tree-optimization/78450
1888 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1889 (handle_builtin_strlen): Call it.
1891 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1893 PR rtl-optimization/83424
1894 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1896 2017-12-18 Marek Polacek <polacek@redhat.com>
1899 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1900 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1903 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1905 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1906 (umaddsidi4, umaddsidi_split): Likewise.
1908 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1910 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1913 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1916 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1918 2017-12-18 Richard Biener <rguenther@suse.de>
1920 PR tree-optimization/81877
1921 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1922 (outermost_indep_loop): Adjust.
1923 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1924 (can_sm_ref_p): Adjust.
1926 2017-12-18 Richard Biener <rguenther@suse.de>
1929 * tree.c (array_at_struct_end_p): Return true if the underlying
1930 object has space for at least one element in excess of what
1931 the array domain specifies.
1933 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1935 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1936 example to avoid overfull hbox.
1937 * doc/invoke.texi (Option Summary): Add missing @gol.
1938 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1941 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1942 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1944 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1945 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1946 (AArch64 Options): Document -mverbose-cost-dump.
1947 (ARM Options): Likewise, plus -mflip-thumb.
1949 2017-12-17 Martin Sebor <msebor@redhat.com>
1952 * gimple-ssa-warn-restrict.c
1953 (builtin_memref::offset_out_of_bounds): Correct the handling of
1956 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1958 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1960 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1962 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1964 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1966 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1969 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1971 * doc/invoke.texi: Fix some typos.
1973 2017-12-16 Martin Sebor <msebor@redhat.com>
1975 PR tree-optimization/78918
1976 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1977 * builtins.c (check_sizes): Rename...
1978 (check_access): ...to this. Rename function arguments for clarity.
1979 (check_memop_sizes): Adjust names.
1980 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1981 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1982 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1983 (check_strncat_sizes, expand_builtin_strncat): Same.
1984 (expand_builtin_strncpy, expand_builtin_memset): Same.
1985 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1986 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1987 (maybe_emit_sprintf_chk_warning): Same.
1988 (expand_builtin_strcpy): Adjust.
1989 (expand_builtin_stpcpy): Same.
1990 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1991 in pointer-checking forms of memcpy, memmove, and mempcpy.
1992 (gcall_to_tree_minimal, max_object_size): Define new functions.
1993 * builtins.h (max_object_size): Declare.
1994 * calls.c (alloc_max_size): Call max_object_size instead of
1995 hardcoding ssizetype limit.
1996 (get_size_range): Handle new argument.
1997 * calls.h (get_size_range): Add a new argument.
1998 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1999 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2000 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2002 (gimple_fold_builtin_memory_chk): Same.
2003 (gimple_fold_builtin_stxcpy_chk): New function.
2004 * gimple-ssa-warn-restrict.c: New source.
2005 * gimple-ssa-warn-restrict.h: New header.
2006 * gimple.c (gimple_build_call_from_tree): Propagate location.
2007 * passes.def (pass_warn_restrict): Add new pass.
2008 * tree-pass.h (make_pass_warn_restrict): Declare.
2009 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2011 (handle_builtin_strcat): Same.
2012 (strlen_optimize_stmt): Rename...
2013 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2014 stpncpy, strncpy, and their checking forms.
2016 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2018 PR rtl-optimization/82849
2019 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2020 and get_max_loop_iterations_int.
2022 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2023 Alan Hayward <alan.hayward@arm.com>
2024 David Sherwood <david.sherwood@arm.com>
2026 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2027 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2028 size as a poly_uint64.
2029 (mode_for_vector, mode_for_int_vector): Take the number of vector
2030 elements as a poly_uint64.
2031 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2032 size as a poly_uint64.
2033 (mode_for_vector, mode_for_int_vector): Take the number of vector
2034 elements as a poly_uint64.
2036 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2037 Alan Hayward <alan.hayward@arm.com>
2038 David Sherwood <david.sherwood@arm.com>
2040 * machmode.h (MACRO_MODE): New macro.
2041 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2042 * targhooks.c (default_libcall_value, default_secondary_reload)
2043 (default_memory_move_cost, default_register_move_cost)
2044 (default_class_max_nregs): Likewise.
2046 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2047 Alan Hayward <alan.hayward@arm.com>
2048 David Sherwood <david.sherwood@arm.com>
2050 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2051 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2053 * genautomata.c (main): Likewise.
2054 * genconditions.c (write_header): Likewise.
2055 * genemit.c (main): Likewise.
2056 * genextract.c (print_header): Likewise.
2057 * genopinit.c (main): Likewise.
2058 * genoutput.c (output_prologue): Likewise.
2059 * genpeep.c (main): Likewise.
2060 * genpreds.c (write_insn_preds_c): Likewise.
2061 * genrecog.c (writer_header): Likewise.
2062 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2063 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2064 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2065 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2066 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2067 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2068 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2069 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2070 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2071 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2072 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2073 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2074 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2075 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2076 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2077 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2078 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2079 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2080 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2081 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2082 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2083 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2084 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2085 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2086 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2087 * config/darwin.c (IN_TARGET_CODE): Likewise.
2088 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2089 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2090 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2091 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2092 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2093 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2094 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2095 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2096 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2097 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2098 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2099 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2100 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2101 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2102 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2103 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2104 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2105 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2106 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2107 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2108 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2109 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2110 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2111 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2112 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2113 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2114 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2115 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2116 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2117 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2118 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2119 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2120 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2121 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2122 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2123 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2124 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2125 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2126 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2127 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2128 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2129 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2130 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2131 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2132 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2133 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2134 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2135 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2136 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2137 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2138 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2139 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2140 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2141 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2142 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2143 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2144 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2145 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2146 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2147 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2148 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2149 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2150 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2151 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2152 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2153 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2154 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2155 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2156 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2157 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2158 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2159 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2160 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2161 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2162 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2163 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2164 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2165 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2166 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2167 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2168 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2169 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2170 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2171 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2172 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2173 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2174 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2175 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2176 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2177 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2178 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2179 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2180 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2181 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2182 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2183 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2184 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2185 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2186 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2187 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2188 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2189 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2190 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2191 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2192 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2193 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2195 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2196 Alan Hayward <alan.hayward@arm.com>
2197 David Sherwood <david.sherwood@arm.com>
2199 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2202 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2203 Alan Hayward <alan.hayward@arm.com>
2204 David Sherwood <david.sherwood@arm.com>
2206 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2207 * doc/md.texi (vec_series@var{m}): Document.
2208 * tree.def (VEC_SERIES_EXPR): New tree code.
2209 * tree.h (build_vec_series): Declare.
2210 * tree.c (build_vec_series): New function.
2211 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2212 * tree-pretty-print.c (dump_generic_node): Likewise.
2213 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2214 * tree-inline.c (estimate_operator_cost): Likewise.
2215 * expr.c (expand_expr_real_2): Likewise.
2216 * optabs-tree.c (optab_for_tree_code): Likewise.
2217 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2218 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2219 * expmed.c (make_tree): Handle VEC_SERIES.
2220 * optabs.def (vec_series_optab): New optab.
2221 * optabs.h (expand_vec_series_expr): Declare.
2222 * optabs.c (expand_vec_series_expr): New function.
2223 * tree-vect-generic.c (expand_vector_operations_1): Check that
2224 the operands also have vector type.
2226 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2227 Alan Hayward <alan.hawyard@arm.com>
2228 David Sherwood <david.sherwood@arm.com>
2230 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2231 (VEC_COND_EXPR): Add missing @tindex.
2232 * doc/md.texi (vec_duplicate@var{m}): Document.
2233 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2234 * tree.c (build_vector_from_val): Add stubbed-out handling of
2235 variable-length vectors, using VEC_DUPLICATE_EXPR.
2236 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2237 * cfgexpand.c (expand_debug_expr): Likewise.
2238 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2239 * tree-inline.c (estimate_operator_cost): Likewise.
2240 * tree-pretty-print.c (dump_generic_node): Likewise.
2241 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2242 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2243 (test_vec_duplicate_folding): New function.
2244 (fold_const_c_tests): Call it.
2245 * optabs.def (vec_duplicate_optab): New optab.
2246 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2247 * optabs.h (expand_vector_broadcast): Declare.
2248 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2249 vec_duplicate_optab.
2250 * expr.c (store_constructor): Try using vec_duplicate_optab for
2252 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2254 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2257 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2258 div/mod latencies a bit.
2260 2017-12-15 Jeff Law <law@redhat.com>
2262 PR tree-optimization/36550
2263 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2264 (mark_threaded_blocks): Rewrite code to avoid block copying when
2265 optimizing for size. Don't pessimize blocks which will be
2266 copied, but all the statements will be dead.
2268 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2270 PR tree-optimization/81165
2271 * tree-ssa-threadupdate.c (uses_in_bb): New.
2272 (estimate_threading_killed_stmts): New.
2273 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2274 * tree-ssa-threadedge.c
2275 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2278 2017-12-15 Jeff Law <law@redhat.com>
2280 PR tree-optimization/83410
2281 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2282 threads when parallelizing loops.
2284 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2286 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2288 * config/alpha/alpha.c (vms_attribute_table): Swap
2289 affects_type_identity and handler fields, adjust comments.
2290 * config/mips/mips.c (mips_attribute_table): Likewise.
2291 * config/visium/visium.c (visium_attribute_table): Likewise.
2292 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2293 * config/microblaze/microblaze.c (microblaze_attribute_table):
2295 * config/spu/spu.c (spu_attribute_table): Likewise.
2296 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2297 * config/arc/arc.c (arc_attribute_table): Likewise.
2298 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2299 * config/v850/v850.c (v850_handle_interrupt_attribute,
2300 v850_handle_data_area_attribute): Formatting fixes.
2301 (v850_attribute_table): Swap affects_type_identity and handler
2302 fields, adjust comments.
2303 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2304 * config/arm/arm.c (arm_attribute_table): Likewise.
2305 * config/avr/avr.c (avr_attribute_table): Likewise.
2306 * config/s390/s390.c (s390_attribute_table): Likewise.
2307 * config/sh/sh.c (sh_attribute_table): Likewise.
2308 * config/i386/i386.c (ix86_handle_cconv_attribute,
2309 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2310 (ix86_attribute_table): Swap affects_type_identity and handler
2311 fields, adjust comments.
2312 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2313 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2314 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2315 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2316 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2317 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2318 * config/rx/rx.c (rx_attribute_table): Likewise.
2319 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2320 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2321 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2322 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2323 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2324 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2325 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2326 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2327 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2328 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2329 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2331 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2333 (register_scoped_attributes, decl_attributes): Formatting fixes.
2335 PR tree-optimization/83269
2336 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2337 subtraction in arg0's type if type is signed and arg0 is unsigned.
2341 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2342 convert? on @0 instead of convert. Check type of @1, not @0.
2343 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2344 convert? on @0 instead of convert. Check type of @1, not @0.
2345 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2346 only optimize if either both @1 and @2 types are narrower
2347 precision, or both are wider or equal precision, and in the former
2348 case only if both have undefined overflow.
2350 2017-12-15 Richard Biener <rguenther@suse.de>
2353 * internal-fn.def (IFN_NOP): Add.
2354 * internal-fn.c (expand_NOP): Do nothing.
2355 * lto-streamer-in.c (input_function): Instead of removing
2356 sanitizer calls replace them with IFN_NOP calls.
2358 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2359 Alan Hayward <alan.hayward@arm.com>
2360 David Sherwood <david.sherwood@arm.com>
2362 * dse.c (store_info, read_info_type): Replace begin and end with
2364 (print_range): New function.
2365 (set_all_positions_unneeded, any_positions_needed_p)
2366 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2368 (record_store): Likewise. Optimize the case in which all positions
2370 (get_stored_val): Replace read_begin and read_end with read_offset
2372 (replace_read): Update call accordingly.
2374 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2376 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2377 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2378 (loop_cand::loop_cand): Initialize above members.
2379 (loop_cand::supported_operations): Delete.
2380 (loop_cand::can_interchange_p): Inline above function.
2381 (loop_cand::classify_simple_reduction): Record number of constant
2382 initialized simple reductions.
2383 (should_interchange_loops): New parameters. Check stmt cost of loops
2385 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2386 Update call to should_interchange_loops.
2387 (should_interchange_loop_nest): Update call to
2388 should_interchange_loops.
2390 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2393 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2394 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2395 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2397 2017-12-15 Julia Koval <julia.koval@intel.com>
2399 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2400 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2402 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2403 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2404 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2406 2017-12-15 Julia Koval <julia.koval@intel.com>
2408 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2409 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2410 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2411 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2412 _mm_aesenc_epi128): New intrinsics.
2414 2017-12-15 Julia Koval <julia.koval@intel.com>
2416 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2417 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2419 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2420 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2421 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2423 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2425 * gimple-ssa-strength-reduction.c (analyze_increments):
2426 Distinguish replacement costs for constant strides from those for
2429 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2431 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2433 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2435 * doc/invoke.texi: Document -Wcast-function-type.
2436 * recog.h (stored_funcptr): Change signature.
2437 * tree-dump.c (dump_node): Avoid warning.
2438 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2440 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2443 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2446 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2448 PR tree-optimization/83312
2449 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2450 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2451 "val" param, and to cope with arbitrary basic blocks.
2452 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2453 handle NULL_TREE for "val", dropping "bb" param.
2454 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2455 drop "bb" param. Handle NULL_TREE for "val".
2456 (find_case_label_for_value): Make "switch_stmt" param const.
2457 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2459 (vrp_prop::check_all_array_refs): Reimplement as...
2460 (check_array_bounds_dom_walker::before_dom_children): ...this new
2461 vfunc. Replace linear search through BB block list, excluding
2462 those with non-executable in-edges via dominator walk.
2464 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2466 * config/arm/arm.opt (mverbose-cost-dump): New option.
2467 * config/arm/arm.c (arm_rtx_costs): Use it.
2469 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2472 * reload1.c (emit_input_reload_insns): Skip debug markers.
2474 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2476 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2480 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2481 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2482 * config/sh/sh.c: Skip debug insns besides notes.
2483 * config/sh/sh.md: Likewise.
2484 * config/sh/sh_treg_combine.cc: Likewise.
2485 * config/sh/sync.md: Likewise.
2487 2017-12-14 Tom de Vries <tom@codesourcery.com>
2489 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2490 Add item for weak_undefined.
2492 2017-12-14 Richard Biener <rguenther@suse.de>
2494 PR tree-optimization/67842
2495 * sese.h (bb_in_region): Remove #if 0'ed code.
2497 2017-12-14 Richard Biener <rguenther@suse.de>
2499 PR tree-optimization/83326
2500 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2501 may_be_zero parameter and handle it by not marking the first
2502 peeled copy as not exiting the loop.
2503 (try_peel_loop): Likewise.
2504 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2505 to handle the case of constant or zero iterations and perform
2506 loop header copying on-the-fly.
2508 2017-12-14 Richard Biener <rguenther@suse.de>
2510 PR tree-optimization/83418
2511 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2512 Instead of asserting we don't get unfolded comparisons deal with them.
2514 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2517 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2518 basic blocks. Assert debug bind insns don't appear outside of bbs,
2519 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2522 PR tree-optimization/83198
2523 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2524 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2525 value if arg is a REAL_CST with incompatible type.
2527 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2528 Bin Cheng <bin.cheng@arm.com>
2531 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2533 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2536 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2538 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2539 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2541 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2542 Alan Hayward <alan.hayward@arm.com>
2543 David Sherwood <david.sherwood@arm.com>
2545 * poly-int.h: New file.
2546 * poly-int-types.h: Likewise.
2547 * coretypes.h: Include them.
2548 (POLY_INT_CONVERSION): Define.
2549 * target.def (estimated_poly_value): New hook.
2550 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2551 * doc/tm.texi: Regenerate.
2552 * doc/poly-int.texi: New file.
2553 * doc/gccint.texi: Include it.
2554 * doc/rtl.texi: Describe restrictions on subreg modes.
2555 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2556 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2557 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2558 * targhooks.h (default_estimated_poly_value): Declare.
2559 * targhooks.c (default_estimated_poly_value): New function.
2560 * target.h (estimated_poly_value): Likewise.
2561 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2562 (wi::unary_traits): Delete.
2563 (wi::binary_traits::signed_shift_result_type): Define for
2564 offset_int << HOST_WIDE_INT, etc.
2565 (generic_wide_int::operator <<=): Define for all types and use
2566 wi::lshift instead of <<.
2567 (wi::hwi_with_prec): Add a default constructor.
2568 (wi::ints_for): New class.
2569 (operator <<): Define for all wide-int types.
2570 (operator /): New function.
2571 (operator %): Likewise.
2572 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2573 (ASSERT_MAYBE_NE_AT): New macros.
2575 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2576 Dominik Vogt <vogt@linux.vnet.ibm.com>
2579 * emit-rtl.c (init_emit): Remove ??? comment.
2580 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2581 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2582 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2583 alignment of 3 virtual registers to BITS_PER_WORD.
2585 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2587 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2589 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2590 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2592 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2595 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2597 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2599 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2600 with the exception of debug begin stmt markers appear before
2604 * final.c (rest_of_handle_final): Call variable_tracking_main only
2605 if !flag_var_tracking.
2607 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2608 Jakub Jelinek <jakub@redhat.com>
2612 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2613 labels debug stmts that can only appear after labels.
2615 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2617 PR rtl-optimization/82398
2618 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2619 EXPR_USEFULNESS in priority comparison.
2621 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2623 PR rtl-optimization/83393
2624 * combine.c (move_deaths): If reg_stat points to a too new insn in
2625 last_death, do not use it: find the proper insn instead.
2627 2017-12-12 Jeff Law <law@redhat.com>
2629 PR tree-optimization/83298
2630 PR tree-optimization/83362
2631 PR tree-optimization/83383
2632 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2633 push_value_range a public interface. Add new argument to
2634 record_ranges_from_stmt.
2635 * gimple-ssa-evrp-analyze.c
2636 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2637 Update comments. Handle recording temporary equivalences.
2638 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2639 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2640 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2641 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2642 gimple-ssa-evrp-analyze.h.
2643 (record_temporary_equivalences_from_phis): Add new argument. When
2644 the PHI arg is an SSA_NAME, set the result's range to the range
2646 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2647 from statements too.
2648 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2649 Pass it down to children as needed.
2650 (thread_outgoing_edges): Likewise.
2651 (thread_across_edge): Likewise. Push/pop range state as needed.
2652 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2654 2017-12-12 Julia Koval <julia.koval@intel.com>
2656 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2657 (PTA_CANNONLAKE): Remove PTA_CLWB.
2659 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2662 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2663 (vcondv2div2df): Likewise.
2664 (vconduv2dfv2di): Likewise.
2666 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2668 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2669 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2670 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2671 ATTR_MATHFN_FPROUNDING_ERRNO.
2673 2017-12-12 Richard Biener <rguenther@suse.de>
2675 PR tree-optimization/83385
2676 * graphite-scop-detection.c (get_order, order): Remove.
2677 (bb_to_rpo): New global.
2679 (build_scops): Sort pbbs in RPO order.
2681 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2683 * combine.c (simplify_set): Do not transform subregs to zero_extends
2684 if the destination mode is a vector mode.
2686 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2688 PR tree-optimization/83359
2689 * tree-cfg.h (fold_loop_internal_call): Declare.
2690 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2691 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2692 (find_loop_dist_alias): New function.
2693 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2694 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2697 PR tree-optimization/80631
2698 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2699 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2700 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2701 hardcoding zero as the value if COND_EXPR is never true. For
2702 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2703 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2704 hardcoding MAX_EXPR as the reduction operation.
2705 (is_nonwrapping_integer_induction): Allow negative step.
2706 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2707 vect_create_epilog_for_reduction, if no value is suitable, don't
2708 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2710 2017-12-12 Richard Biener <rguenther@suse.de>
2712 PR tree-optimization/81889
2713 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2714 range info from the non-wrapping IV instead of just the range
2717 2017-12-12 Julia Koval <julia.koval@intel.com>
2719 * config.gcc: Add vaesintrin.h.
2720 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2722 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2723 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2725 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2726 * config/i386/immintrin.h: Include vaesintrin.h.
2727 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2728 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2729 _mm_aesdec_epi128): New intrinsics.
2731 2017-12-12 Julia Koval <julia.koval@intel.com>
2733 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2734 OPTION_MASK_ISA_VAES_UNSET): New.
2735 (ix86_handle_option): Handle -mvaes.
2736 * config/i386/cpuid.h: Define bit_VAES.
2737 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2738 * config/i386/i386-c.c (__VAES__): New.
2739 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2740 (ix86_valid_target_attribute_inner_p): Ditto.
2741 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2742 * config/i386/i386.opt: Add -mvaes.
2743 * doc/invoke.texi: Ditto.
2745 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2747 * debug.h (gcc_debug_hooks): Add inline_entry.
2748 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2749 * debug.c (do_nothing_debug_hooks): Likewise.
2750 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2751 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2752 (dwarf2_lineno_debug_hooks): Likewise.
2754 * common.opt (gstatement-frontiers): New, setting
2755 debug_nonbind_markers_p.
2756 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2757 * toplev.c (process_options): Autodetect value for debug statement
2759 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2760 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2762 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2763 markers. Integrate source bind into debug stmt expand loop.
2764 (pass_expand::execute): Check debug marker limit. Avoid deep
2765 TER and expand debug locations for debug bind insns only.
2766 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2768 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2769 * final.c (reemit_insn_block_notes): Take current block from
2770 nonbind markers. Declare note where it's first set.
2771 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2772 begin stmt markers if enabled.
2773 (notice_source_line): Handle nonbind markers. Fail if their
2774 location is unknown or that of builtins.
2775 (rest_of_handle_final): Convert begin stmt markers to notes if
2776 var-tracking didn't run.
2777 (rest_of_clean_state): Skip begin stmt markers.
2778 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2780 * function.c (allocate_struct_function): Set begin_stmt_markers.
2781 * function.h (struct function): Add debug_marker_count counter
2782 and debug_nonbind_markers flag.
2783 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2784 * gimple-low.c (lower_function_body): Adjust
2785 debug_nonbind_markers.
2786 (lower_stmt): Drop or skip gimple debug stmts.
2787 (lower_try_catch): Skip debug stmts.
2788 * gimple.c (gimple_build_debug_begin_stmt): New.
2789 (gimple_copy): Increment debug_marker_count if copying one.
2790 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2791 * gimplify.c (rexpr_location): New.
2792 (rexpr_has_location): New.
2793 (warn_switch_unreachable_r): Handle gimple debug stmts.
2794 (shortcut_cond_r): Call expr_location.
2796 (find_goto_label): New.
2797 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2799 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2801 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2802 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2803 (LANG_HOOKS_INITIALIZER): ... this.
2804 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2805 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2807 (update_ebb_live_info): Skip debug insn markers.
2808 * lra.c (debug_insn_static_data): Rename to...
2809 (debug_bind_static_data): ... this.
2810 (debug_marker_static_data): New.
2811 (lra_set_insn_recog_data): Select one of the above depending
2813 (lra_update_isn_regno_info): Don't assume debug insns have
2815 (push_insns): Skip debug insns.
2816 * lto-streamer-in.c (input_function): Drop debug stmts
2817 depending on active options. Adjust debug_nonbind_markers.
2818 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2819 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2820 begin stmt marker notes.
2821 (print_insn): Likewise.
2822 * recog.c (extract_insn): Recognize rtl for debug markers.
2823 * rtl.def (DEBUG_MARKER): New.
2824 * tree-inline.c: Include params.h.
2825 (remap_gimple_stmt): Handle nonbind markers.
2826 (maybe_move_debug_stmts_to_successors): Likewise.
2827 (copy_debug_stmt): Likewise.
2828 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2829 markers regardless of no side effects.
2830 (tsi_link_before): Don't update container's side effects when adding
2831 a begin stmt marker.
2832 (tsi_link_after): Likewise.
2833 (expr_first): Skip begin stmt markers.
2834 (expr_last): Likewise.
2835 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2836 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2837 Disregard nonbind markers.
2838 * tree.c (make_node_stat): Don't set side effects for begin stmt
2840 (build1_stat): Likewise.
2841 * tree.def (DEBUG_BEGIN_STMT): New.
2842 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2843 * var-tracking.c (delete_debug_insns): Renamed to...
2844 (delete_vta_debug_insns): ... this.
2845 (reemit_marker_as_note): New.
2846 (vt_initialize): Reemit markers.
2847 (delete_vta_debug_insns): Likewise.
2848 (vt_debug_insns_local): Reemit or delete markers.
2849 (variable_tracking_main_1): Likewise.
2850 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2851 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2852 (gimple_debug_nonbind_marker_p): New.
2853 (gimple_build_debug_bind): Adjust.
2854 (gimple_build_debug_begin_stmt): New.
2855 * doc/invoke.texi (max-debug-marker-count): New param.
2856 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2857 (debug_parameter_ref, debug_marker): New.
2858 (NOTE_INSN_BEGIN_STMT): New.
2859 (DEBUG_INSN): Describe begin stmt markers.
2861 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2862 without debug insns, we wouldn't, but clean up debug insns
2863 after a control flow insn nevertheless.
2865 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2866 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2867 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2868 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2869 (rtl_tidy_fallthru_edge): Likewise.
2870 (rtl_verify_fallthru): Likewise.
2871 (rtl_verify_bb_layout): Likewise.
2872 (skip_insns_after_block): Likewise.
2873 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2874 * dwarf2out.c: Include print-rtl.h.
2875 (dwarf2out_next_real_insn): New.
2876 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2877 Dump debug binds in asm comments.
2878 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2879 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2880 callers to use gsi_start_nondebug_bb instead.
2881 (gsi_after_labels): Skip gimple debug stmts.
2882 (gsi_start_nondebug): New.
2883 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2884 (proper_loop_form_for_interchange): Adjust.
2885 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2886 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2887 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2888 (collect_fallthrough_labels): Likewise.
2889 (should_warn_for_implicit_fallthrough): Likewise.
2890 (warn_implicit_fallthrough_r): Likewise.
2891 (expand_FALLTHROUGH_r): Likewise.
2892 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2893 (graphite_copy_stmts_from_block): Skip nonbind markers.
2894 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2895 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2896 * jump.c (clean_barriers): Skip debug insns.
2897 * omp-expand.c (expand_parallel_call): Skip debug insns.
2898 (expand_task_call): Likewise.
2899 (remove_exit_barrier): Likewise.
2900 (expand_omp_taskreg): Likewise.
2901 (expand_omp_for_init_counts): Likewise.
2902 (expand_omp_for_generic): Likewise.
2903 (expand_omp_for_static_nochunk): Likewise.
2904 (expand_omp_for_static_chunk): Likewise.
2905 (expand_omp_simd): Likewise.
2906 (expand_omp_taskloop_for_outer): Likewise.
2907 (expand_omp_taskloop_for_inner): Likewise.
2908 (expand_oacc_for): Likewise.
2909 (expand_omp_sections): Likewise.
2910 (expand_omp_single): Likewise.
2911 (expand_omp_synch): Likewise.
2912 (expand_omp_atomic_load): Likewise.
2913 (expand_omp_atomic_store): Likewise.
2914 (expand_omp_atomic_fetch_op): Likewise.
2915 (expand_omp_atomic_pipeline): Likewise.
2916 (expand_omp_atomic_mutex): Likewise.
2917 (expand_omp_target): Likewise.
2918 (grid_expand_omp_for_loop): Likewise.
2919 (grid_expand_target_grid_body): Likewise.
2920 (build_omp_regions_1): Likewise.
2921 * omp-low.c (check_combined_parallel): Skip debug stmts.
2922 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2923 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2924 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2926 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2927 (make_edges): Likewise.
2928 (cleanup_dead_labels): Likewise.
2929 (gimple_can_merge_blocks_p): Likewise.
2930 (stmt_starts_bb_p): Likewise.
2931 (gimple_block_label): Likewise.
2932 (gimple_redirect_edge_and_branch): Likewise.
2933 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2935 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2937 * tree-pretty-print (print_declaration): Omit initializer in slim
2939 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2941 (eliminate_unnecessary_stmts): Stabilize block removal order.
2942 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2943 * var-tracking.c (get_first_insn): New.
2944 (vt_emit_notes): Call it.
2945 (vt_initialize): Walk any insns before the first BB.
2946 (delete_debug_insns): Likewise.
2948 * gimple.h (enum gimple_debug_subcode): Add
2949 GIMPLE_DEBUG_BEGIN_STMT.
2950 (gimple_debug_begin_stmt_p): New.
2951 (gimple_debug_nonbind_marker_p): New.
2952 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2953 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2954 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2955 * insn-notes.def (BEGIN_STMT): New.
2956 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2957 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2958 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2959 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2960 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2961 (INSN_DEBUG_MARKER_KIND): New.
2962 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2963 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2964 (INSN_VAR_LOCATION_PTR): New.
2965 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2967 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2968 deps only if debug bind insns are enabled.
2969 (pass_expand::execute): Avoid deep TER and expand
2970 debug locations for debug bind insns only.
2971 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2972 debug stmts special handling down to debug bind stmts.
2973 * combine.c (try_combine): Narrow debug insns special handling
2974 down to debug bind insns.
2975 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2976 Narrow debug insns preexisting special handling down to debug
2978 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2979 handling down to debug bind insns.
2980 * function.c (instantiate_virtual_regs): Skip debug markers,
2981 adjust handling of debug binds.
2982 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2983 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2984 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2985 insns to debug bind insns.
2986 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2987 special handling of debug stmts to debug bind stmts.
2988 * ipa-split.c (split_function): Likewise.
2989 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2990 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2992 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2993 * regrename.c (build_def_use): Likewise.
2994 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2995 (pass_cprop_hardreg): Narrow special casing of debug insns to
2997 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2998 * reload1.c (reload): Likewise.
2999 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3000 casing of debug stmts to debug bind stmts.
3001 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3002 * ssa-iterators.h (num_imm_uses): Likewise.
3003 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3004 debug stmts to debug bind stmts.
3005 * tree-inline.c (tree_function_versioning): Narrow special casing
3006 of debug stmts to debug bind stmts.
3007 * tree-loop-distribution.c (generate_loops_for_partition):
3008 Narrow special casing of debug stmts to debug bind stmts.
3009 * tree-sra.c (analyze_access_subtree): Narrow special casing
3010 of debug stmts to debug bind stmts.
3011 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3012 stmts to debug bind stmts.
3013 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3014 casing of debug stmts to debug bind stmts.
3015 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3016 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3017 casing of debug stmts to debug bind stmts.
3018 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3020 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3021 debug stmts to debug bind stmts.
3022 (gimple_replace_ssa_lhs): Likewise.
3023 (insert_debug_temp_for_var_def): Likewise.
3024 (insert_debug_temps_for_defs): Likewise.
3025 (reset_debug_uses): Likewise.
3026 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3027 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3028 (adjust_debug_stmts): Likewise.
3029 (adjust_phi_and_debug_stmts): Likewise.
3030 (vect_do_peeling): Likewise.
3031 * tree-vect-loop.c (vect_transform_loop): Likewise.
3032 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3033 * var-tracking.c (adjust_mems): Narrow special casing of debug
3034 insns to debug bind insns.
3035 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3036 (compute_bb_dataflow, vt_find_locations): Likewise.
3037 (vt_expand_loc, emit_notes_for_changes): Likewise.
3038 (vt_init_cfa_base): Likewise.
3039 (vt_emit_notes): Likewise.
3040 (vt_initialize): Likewise.
3041 (vt_finalize): Likewise.
3043 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3044 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3045 (next_nonnote_nondebug_insn_bb): New.
3046 (prev_nonnote_nondebug_insn_bb): New.
3047 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3048 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3049 (prev_nonnote_nondebug_insn_bb): Declare.
3050 (next_nonnote_nondebug_insn_bb): Declare.
3051 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3052 * cfgrtl.c (get_last_bb_insn): Likewise.
3053 * lra.c (push_insns): Likewise.
3055 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3058 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3060 * selftest.c (selftest::test_runner::test_runner): New ctor.
3061 (selftest::test_runner::~test_runner): New dtor.
3062 * selftest.h (class selftest::test_runner): New class.
3064 2017-12-11 Carl Love <cel@us.ibm.com>
3066 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3067 vec_extract_fp32_from_shortl]): Add #defines.
3068 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3069 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3070 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3071 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3072 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3073 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3074 * doc/extend.texi: Add documentation for the added builtins.
3076 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3078 PR rtl-optimization/80693
3079 PR rtl-optimization/81019
3080 PR rtl-optimization/81020
3081 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3082 are not mentioned in i3. Place the REG_UNUSED note on i2,
3083 possibly modified to REG_DEAD, if it did not originate in i3.
3085 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3087 * recog.c (store_data_bypass_p_1): New function.
3088 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3089 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3091 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3093 PR rtl-optimization/83361
3094 * ifcvt.c (if_convert): Call fixup_partitions.
3096 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3098 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3099 early folding of splat_u{8,16,32}.
3101 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3103 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3104 output_operand_lossage first argument with capital letter.
3105 (aarch64_override_options): Don't start error and sorry first argument
3106 with capital letter.
3108 2017-12-11 Andi Kleen <ak@linux.intel.com>
3110 PR gcov-profile/83355
3111 * auto-profile.c (string_table::get_index_by_decl): Don't
3112 recurse when abstract origin points to itself.
3114 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3116 PR tree-optimization/83320
3117 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3118 (prune_datarefs_not_in_loop): Ditto.
3120 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3122 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3124 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3126 PR tree-optimization/83337
3127 * gimple-loop-interchange.cc (compute_access_stride): Handle
3128 bitfield DRs properly.
3130 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3132 PR tree-optimization/83338
3133 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3134 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3135 vector of unsigned integers to vector of signed integers.
3137 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3139 PR rtl-optimization/83317
3140 * lra-constraints.c (process_address_1): Add insn code check.
3142 2017-12-08 Michael Matz <matz@suse.de>
3144 Fix PR tree-optimization/83323
3145 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3146 head-controlled loops and loop BBs.
3147 * common.opt (funroll-and-jam): Remove, instead ...
3148 (floop-unroll-and-jam): ... reuse this option.
3149 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3150 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3151 (-floop-unroll-and-jam): ... this option.
3153 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3155 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3156 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3157 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3159 2017-12-08 Richard Biener <rguenther@suse.de>
3162 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3163 handle accesses outside of zero-sized vars.
3165 2017-12-08 Martin Jambor <mjambor@suse.cz>
3167 PR tree-optimization/83141
3168 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3169 test for MEM_REFs implicitely changing types with padding. Remove
3171 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3173 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3175 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3177 * config/avr/avr.c (avr_attribute_table): Likewise.
3178 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3179 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3180 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3181 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3182 of struct attribute_spec.
3183 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3185 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3188 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3190 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3193 * profile-count.c (profile_count::from_gcov_type): Move from
3194 profile-count.h; handle overflow.
3195 * profile-count.h (profile_count::from_gcov_type): Move offline.
3197 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3199 PR rtl-optimization/83304
3200 * combine.c (move_deaths): If we do not know where a register died,
3203 2017-12-08 Richard Biener <rguenther@suse.de>
3205 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3206 Provide -fopt-info-loop feedback when we interchange in a nest.
3208 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3210 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3211 for armv6 ARM CPU IDs.
3213 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3215 * common/config/arm/arm-common.c: Include <algorithm>.
3216 (INCLUDE_VECTOR): Define.
3217 (compare_opt_names): New function.
3218 (arm_rewrite_selected_arch): Only strip out extensions that can be
3219 expressed through -mfpu. Sort the remaining extensions
3222 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3224 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3225 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3226 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3227 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3229 2017-06-08 Tristan Gingold <gindold@adacore.com>
3232 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3233 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3235 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3237 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3238 of struct attribute_spec.
3240 2017-12-08 Julia Koval <julia.koval@intel.com>
3242 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3243 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3244 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3245 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3246 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3247 _mm_maskz_dpwssds_epi32): Ditto.
3249 2017-12-08 Richard Biener <rguenther@suse.de>
3251 PR tree-optimization/81303
3252 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3253 conditions try to create a comparison vector type matching
3254 the data vector type.
3255 (vectorizable_condition): Adjust.
3256 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3257 Leave invariant conditions alone in case we can vectorize those.
3259 2017-12-08 Julia Koval <julia.koval@intel.com>
3261 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3262 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3263 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3264 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3265 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3267 2017-12-08 Julia Koval <julia.koval@intel.com>
3269 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3270 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3271 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3272 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3273 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3274 _mm_maskz_dpbusds_epi32): New intrinsics.
3276 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3278 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3280 (ld<bh>io_signed): Likewise.
3281 (st<bhw>io): Likewise.
3282 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3283 SMALL_INT12 constant integer operand.
3285 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3287 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3289 (nios2_large_constant_allowed): ...this. Adjust uses.
3290 (nios2_plus_symbolic_constant_p): Rename to...
3291 (nios2_plus_large_constant_p): ...this. Adjust uses.
3292 (nios2_legitimate_address_p): Correct CONST_INT handling.
3293 (nios2_symbolic_memory_operand_p): Rename to...
3294 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3295 (nios2_large_constant_p): Check for large constant integers too.
3296 (nios2_split_large_constant): Handle constant integers.
3297 (nios2_split_symbolic_memory_operand): Rename to...
3298 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3299 (nios2_legitimize_constant_address): Handle constant integers.
3300 (r0rel_constant_p): Handle small constant integers.
3301 (nios2_print_operand_address): Handle r0-relative integer addresses.
3302 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3303 * config/nios2/nios2.md: Adjust for renamed functions.
3305 2017-12-07 Andrew Waterman <andrew@sifive.com>
3307 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3308 (TARGET_HAVE_SRODATA_SECTION): New define.
3309 (riscv_select_section): New function.
3311 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3312 Alexander Monakov <amonakov@ispras.ru>
3313 Jakub Jelinek <jakub@redhat.com>
3316 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3318 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3320 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3321 a bool success value. Don't call output_operand_lossage here.
3322 (aarch64_print_ldpstp_address): Return a bool success value.
3323 (aarch64_print_operand_address): Call output_addr_const if
3324 aarch64_print_address_internal fails.
3325 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3326 'y'; call output_operand_lossage instead. Call output_operand_lossage
3327 if aarch64_print_ldpstp_address fails.
3329 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3331 * tree-vector-builder.h
3332 (tree_vector_builder::binary_encoded_nelts): Declare.
3333 * tree-vector-builder.c
3334 (tree_vector_builder::binary_encoded_nelts): New function.
3335 * fold-const.c (negate_expr_p): Likewise.
3336 (operand_equal_p, fold_checksum_tree): Likewise.
3337 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3338 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3339 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3340 (uniform_vector_p): Likewise.
3341 * varasm.c (const_hash_1, compare_constant): Likewise.
3342 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3343 (valid_lattice_transition): Operate directly on the VECTOR_CST
3345 * ipa-icf.c: Include tree-vector-builder.h.
3346 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3347 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3349 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3351 * tree.c (build_vector): Delete.
3352 * tree.h (build_vector): Make static and move into the self-testing
3355 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3357 * vector-builder.h (vector_builder::derived): New const overload.
3358 (vector_builder::elt): New function.
3359 * tree-vector-builder.h (tree_vector_builder::type): New function.
3360 (tree_vector_builder::apply_step): Declare.
3361 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3363 * gimple-fold.h (tree_vector_builder): Declare.
3364 (gimple_build_vector): Take a tree_vector_builder instead of a
3365 type and vector of elements.
3366 * gimple-fold.c (gimple_build_vector): Likewise.
3367 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3369 (get_initial_defs_for_reduction): Likewise.
3370 (vectorizable_induction): Likewise.
3372 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3374 * tree-vector-builder.h
3375 (tree_vector_builder::new_binary_operation): Declare.
3376 * tree-vector-builder.c
3377 (tree_vector_builder::new_binary_operation): New function.
3378 * fold-const.c (fold_relational_const): Use it.
3379 (const_binop): Likewise. Check that both input vectors have
3380 the same number of elements, thus excluding things like WIDEN_SUM.
3381 Check whether it is possible to operate directly on the encodings
3384 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3386 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3387 new_unary_operation, operating only on the encoded elements.
3388 (const_unop): Likewise.
3389 (exact_inverse): Likewise.
3390 (distributes_over_addition_p): New function.
3391 (const_binop): Use tree_vector_builder and new_unary_operation
3392 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3393 on the encoded elements unless the encoding is strided and the
3394 operation does not distribute over addition.
3395 (fold_convert_const): Use tree_vector_builder and
3396 new_unary_operation. Operate only on the encoded elements
3397 for truncating integer conversions, or for non-stepped encodings.
3399 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3401 * config/sparc/sparc.c: Include tree-vector-builder.h.
3402 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3403 * expmed.c: Include tree-vector-builder.h.
3404 (make_tree): Use tree_vector_builder instead of build_vector.
3405 * fold-const.c: Include tree-vector-builder.h.
3406 (const_binop): Use tree_vector_builder instead of build_vector.
3407 (const_unop): Likewise.
3408 (native_interpret_vector): Likewise.
3409 (fold_vec_perm): Likewise.
3410 (fold_ternary_loc): Likewise.
3411 * gimple-fold.c: Include tree-vector-builder.h.
3412 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3414 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3415 (simplify_vector_constructor): Use tree_vector_builder instead
3417 * tree-vect-generic.c: Include tree-vector-builder.h.
3418 (add_rshift): Use tree_vector_builder instead of build_vector.
3419 (expand_vector_divmod): Likewise.
3420 (optimize_vector_constructor): Likewise.
3421 * tree-vect-loop.c: Include tree-vector-builder.h.
3422 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3423 of build_vector. Explicitly use a stepped encoding for
3425 * tree-vect-slp.c: Include tree-vector-builder.h.
3426 (vect_get_constant_vectors): Use tree_vector_builder instead
3428 (vect_transform_slp_perm_load): Likewise.
3429 (vect_schedule_slp_instance): Likewise.
3430 * tree-vect-stmts.c: Include tree-vector-builder.h.
3431 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3432 (vect_gen_perm_mask_any): Likewise.
3433 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3434 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3436 (build_vector_from_val): Likewise. Explicitly use a duplicate
3439 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3441 * doc/generic.texi (VECTOR_CST): Describe new representation of
3443 * vector-builder.h: New file.
3444 * tree-vector-builder.h: Likewise.
3445 * tree-vector-builder.c: Likewise.
3446 * Makefile.in (OBJS): Add tree-vector-builder.o.
3447 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3448 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3449 (tree_vector): Change the number of elements to
3450 vector_cst_encoded_nelts.
3451 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3452 (VECTOR_CST_ELTS): Delete.
3453 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3454 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3455 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3456 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3457 (VECTOR_CST_ENCODED_ELT): Likewise.
3458 (vector_cst_encoded_nelts): New function.
3459 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3460 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3461 (vector_cst_int_elt, vector_cst_elt): Declare.
3462 * tree.c: Include tree-vector-builder.h.
3463 (tree_code_size): Abort if passed VECTOR_CST.
3464 (tree_size): Update for new VECTOR_CST layout.
3465 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3466 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3467 (build_vector): Use tree_vector_builder.
3468 (vector_cst_int_elt, vector_cst_elt): New functions.
3469 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3470 encoded elements and then create the vector in the canonical form.
3471 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3472 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3473 (tree_c_tests): Call test_vector_cst_patterns.
3474 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3476 (hash_tree): Likewise.
3477 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3478 (streamer_write_tree_header): Likewise.
3479 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3480 (streamer_alloc_tree): Likewise. Update call to make_vector.
3481 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3483 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3485 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3486 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3487 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3489 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3490 Richard Biener <rguenther@suse.de>
3492 PR tree-optimization/81303
3493 * Makefile.in (gimple-loop-interchange.o): New object file.
3494 * common.opt (floop-interchange): Reuse the option from graphite.
3495 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3496 -floop-interchange and mention it for -O3.
3497 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3498 * gimple-loop-interchange.cc: New file.
3499 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3500 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3501 * passes.def (pass_linterchange): New pass.
3502 * timevar.def (TV_LINTERCHANGE): New time var.
3503 * tree-pass.h (make_pass_linterchange): New declaration.
3504 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3505 interchange. Record IV before/after increment in new parameters.
3506 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3507 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3509 (check_reduction_path): ...New function here.
3510 * tree-vectorizer.h (check_reduction_path): New declaration.
3512 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3515 PR rtl-optimization/80818
3516 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3517 always early clobbered.
3518 * lra-lives.c (process_bb_lives): Check input hard regs for early
3519 clobbered non-operand hard reg.
3521 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3524 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3525 types_compatible_p, just that TYPE_MODE is the same.
3527 2017-12-07 Martin Sebor <msebor@redhat.com>
3530 * attribs.c (empty_attribute_table): Initialize new member of
3531 struct attribute_spec.
3532 (decl_attributes): Add argument. Handle mutually exclusive
3533 combinations of attributes.
3534 (selftests::test_attribute_exclusions): New function.
3535 (selftests::attribute_c_tests): Ditto.
3536 * attribs.h (decl_attributes): Add default argument.
3537 * selftest.h (attribute_c_tests): Declare.
3538 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3539 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3541 * doc/extend.texi (Common Function Attributes): Update const and pure.
3542 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3543 of struct attribute_spec.
3544 * config/arc/arc.c (arc_attribute_table): Same.
3545 * config/arm/arm.c (arm_attribute_table): Same.
3546 * config/avr/avr.c ( avr_attribute_table): Same.
3547 * config/bfin/bfin.c (bfin_attribute_table): Same.
3548 * config/cr16/cr16.c (cr16_attribute_table): Same.
3549 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3550 * config/h8300/h8300.c (h8300_attribute_table): Same.
3551 * config/i386/i386.c (ix86_attribute_table): Same.
3552 * config/ia64/ia64.c (ia64_attribute_table): Same.
3553 * config/m32c/m32c.c (m32c_attribute_table): Same.
3554 * config/m32r/m32r.c (m32r_attribute_table): Same.
3555 * config/m68k/m68k.c (m68k_attribute_table): Same.
3556 * config/mcore/mcore.c (mcore_attribute_table): Same.
3557 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3558 * config/mips/mips.c (mips_attribute_table): Same.
3559 * config/msp430/msp430.c (msp430_attribute_table): Same.
3560 * config/nds32/nds32.c (nds32_attribute_table): Same.
3561 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3562 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3563 * config/rl78/rl78.c (rl78__attribute_table): Same.
3564 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3565 * onfig/rx/rx.c (rx_attribute_table): Same.
3566 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3567 * config/sh/sh.c (sh_attribute_table): Same.
3568 * config/sparc/sparc.c (sparc_attribute_table): Same.
3569 * config/spu/spu.c (spu_attribute_table): Same.
3570 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3571 * config/v850/v850.c (v850_attribute_table): Same.
3572 * config/visium/visium.c (visium_attribute_table): Same.
3574 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3577 * config/arm/arm.c (INCLUDE_STRING): Define.
3578 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3579 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3582 2017-12-07 Michael Matz <matz@suse.de>
3584 Add unroll and jam pass
3586 * gimple-loop-jam.c: New file.
3587 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3588 * common.opt (funroll-and-jam): New option.
3589 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3590 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3591 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3592 * passes.def: Add pass_loop_jam.
3593 * timevar.def (TV_LOOP_JAM): Add.
3594 * tree-pass.h (make_pass_loop_jam): Declare.
3595 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3596 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3597 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3598 to flow_loop_tree_node_add.
3599 (duplicate_subloops, copy_loops_to): Append to sibling list.
3600 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3601 * doc/invoke.texi (-funroll-and-jam): Document new option.
3602 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3604 2017-12-07 Richard Biener <rguenther@suse.de>
3606 PR tree-optimization/83296
3607 PR tree-optimization/67769
3608 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3609 flow sensitive info in an unrelated BB.
3610 (value_replacement): Use reset_flow_sensitive_info.
3611 (minmax_replacement): Reset flow sensitive info on the def
3612 we move. Do not reset flow sensitive info in the whole BB
3613 we move the stmt to.
3614 (abs_replacement): Likewise.
3616 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3619 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3620 rs6000_cpu to the given -mcpu=, or to the default processor.
3622 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3624 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3625 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3626 instead of rs6000_cpu_attr.
3627 (rs6000_adjust_cost): Ditto.
3628 (is_microcoded_insn): Ditto.
3629 (rs6000_adjust_priority): Ditto.
3630 (rs6000_issue_rate): Ditto.
3631 (rs6000_use_sched_lookahead): Ditto.
3632 (rs6000_use_sched_lookahead_guard): Ditto.
3633 (rs6000_sched_reorder): Ditto.
3634 (force_new_group): Ditto.
3635 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3636 (group_ending_nop): Ditto.
3638 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3640 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3641 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3642 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3643 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3645 (rs6000_reassociation_width): Ditto.
3646 (rs6000_emit_epilogue): Ditto.
3647 (rs6000_adjust_cost): Ditto.
3648 (is_microcoded_insn): Ditto.
3649 (is_cracked_insn): Ditto.
3650 (rs6000_adjust_priority): Ditto.
3651 (rs6000_sched_reorder): Ditto.
3652 (rs6000_sched_reorder2): Ditto.
3653 (insn_must_be_first_in_group): Ditto.
3654 (insn_must_be_last_in_group): Ditto.
3655 (rs6000_register_move_cost): Ditto.
3656 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3659 2017-12-07 Julia Koval <julia.koval@intel.com>
3661 * config.gcc: Add vaesintrin.h.
3662 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3664 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3665 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3667 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3668 * config/i386/immintrin.h: Include vaesintrin.h.
3669 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3670 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3671 _mm_aesdec_epi128): New intrinsics.
3673 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3675 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3676 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3677 c-family/c-spellcheck.cc.
3678 (best_macro_match::best_macro_match): Likewise.
3680 (struct edit_distance_traits<cpp_hashnode *>): Move to
3681 c-family/c-spellcheck.h.
3682 (class best_macro_match): Likewise.
3684 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3686 PR tree-optimization/83293
3687 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3688 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3689 might insert into empty bb.
3692 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3693 simplify for plus with :c added, and pointer_plus without that.
3694 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3695 with undefined overflow and the conversion is not widening,
3696 perform negation in utype and only convert to type afterwards.
3697 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3698 simplify for plus with :c added, and pointer_plus without that.
3699 If type is integral with undefined overflow and the conversion is
3700 not widening, perform minus in utype and only convert to type
3701 afterwards. Move the last pointer_diff_expr simplify into the
3704 2017-12-06 Martin Sebor <msebor@redhat.com>
3706 PR tree-optimization/82646
3707 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3708 strncpy, not maxlen.
3710 2017-12-06 Martin Sebor <msebor@redhat.com>
3712 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3715 PR tree-optimization/83075
3716 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3717 strncat/strncpy don't change length of source string.
3719 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3722 2017-11-29 Martin Aberg <maberg@gaisler.com>
3724 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3725 to prevent b2bst errata sequence.
3726 (sqrtdf2_fix): Likewise.
3728 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3730 PR tree-optimization/81945
3731 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3732 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3733 to dest_cfun has orig_loop_num set, either remap it to the new
3734 loop number if the loop got moved too, or clear it.
3736 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3738 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3740 (thunderx2t99_nothing): New insn reservation.
3741 (thunderx2t99_mrs): New insn reservation.
3742 (thunderx2t99_multiple): New insn reservation.
3743 (thunderx2t99_alu_basi): Add bfx to reservation.
3744 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3746 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3749 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3751 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3753 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3754 tree-ssa-pre.c::remove_dead_inserted_code.
3755 * tree-ssa-dce.h: New file.
3756 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3757 (remove_dead_inserted_code): Move and rename to function
3758 tree-ssa-dce.c::simple_dce_from_worklist.
3759 (pass_pre::execute): Update use.
3761 2017-12-05 Richard Biener <rguenther@suse.de>
3763 PR tree-optimization/83277
3764 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3765 to code-gen liveout vars.
3767 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3769 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3770 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3771 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3774 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3776 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3778 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3779 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3782 2017-12-05 Richard Biener <rguenther@suse.de>
3784 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3785 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3786 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3787 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3789 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3791 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3793 2017-12-05 Martin Liska <mliska@suse.cz>
3794 Jakub Jelinek <jakub@redhat.com>
3796 * doc/invoke.texi: Document the options.
3797 * flag-types.h (enum sanitize_code): Add
3798 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3799 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3800 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3801 * opts.c: Define new sanitizer options.
3802 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3803 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3805 2017-12-05 Julia Koval <julia.koval@intel.com>
3807 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3808 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3809 (ix86_handle_option): Handle -mavx512vnni.
3810 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3811 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3812 * config/i386/i386-c (__AVX512VNNI__): New.
3813 * config/i386/i386.c (ix86_target_string): Handle new option.
3814 (ix86_valid_target_attribute_inner_p): Handle new option.
3815 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3816 * config/i386/i386.opt (mavx512vnni): New option.
3818 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3821 * athlon.md: Disable for generic.
3822 * haswell.md: Enable for generic.
3823 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3824 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3826 (ix86_adjust_cost): Move generic to haswell path.
3828 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3830 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3831 instead of MEM_P in a couple more places. Fix formatting issues.
3833 2017-12-04 Jim Wilson <jimw@sifive.com>
3835 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3836 instead of GP_REG_LAST-1.
3837 (riscv_adjust_libcall_cfi_prologue): Likewise.
3838 (riscv_adjust_libcall_cri_epilogue): Likewise.
3839 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3842 2017-12-04 Luis Machado <luis.machado@linaro.org>
3844 * ipa-pure-const.c (check_decl): Add missing newline.
3845 (state_from_flags): Likewise.
3847 2017-12-04 Jeff Law <law@redhat.com>
3849 PR tree-optimizatin/78496
3850 * gimple-ssa-evrp-analyze.h
3851 (evrp_range_analyzer::get_vr_values): Simplify.
3852 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3853 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3854 and gimple-ssa-evrp-analyze.h.
3855 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3856 (simplify_stmt_for_jump_threading): Copy a blob of code from
3857 tree-vrp.c to use ranges to simplify statements.
3858 (dom_opt_dom_walker::before_dom_children): Call
3859 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3860 (dom_opt_dom_walker::after_dom_children): Similarly for
3861 evrp_range_analyzer::leave.
3862 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3865 * gimple-ssa-evrp-analyze.c
3866 (evrp_range_analyzer::extract_range_from_stmt): Always use
3867 vr_values::update_value_range so preexisting range info is
3868 medged with new range info, even if the new range is VR_VARYING.
3870 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3872 * combine.c: Adjust comment.
3873 (use_crosses_set_p): Delete.
3874 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3875 (try_combine): Ditto.
3877 2017-12-04 Richard Biener <rguenther@suse.de>
3879 PR tree-optimization/83255
3880 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3881 Re-add zero-iteration check.
3883 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3885 PR rtl-optimization/83245
3886 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3887 hard registers as earlyclobber, also if not in an asm.
3889 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3893 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3896 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3897 rs6000_cpu based on cpu_index, not tune_index.
3899 2017-12-04 Richard Biener <rguenther@suse.de>
3901 PR tree-optimization/83238
3902 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3903 code match comment, rejecting invalid SESE regions.
3905 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3907 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3908 require base operand is a REG_POINTER prior to reload on targets
3909 with non-equivalent space registers.
3911 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3913 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3914 (update_specialized_profile): Fix updating of counts.
3915 (perhaps_add_new_callers): Likewise.
3917 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3920 * x86-tune.def: Remove obsolette FIXMEs.
3921 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3922 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3923 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3925 (X86_TUNE_PAD_RETURNS): Disable for generic.
3927 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3929 PR tree-optimization/83170
3930 PR tree-optimization/83241
3931 * gimple-ssa-store-merging.c
3932 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3933 gimple_vuse (ins_stmt) in case it has changed.
3934 (imm_store_chain_info::output_merged_store): Likewise.
3936 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3940 * tree-cfg.c (pass_warn_function_return::execute): Handle
3941 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3942 with BUILTINS_LOCATION.
3946 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3947 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3948 instead of DECL_MODE.
3950 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3951 last argument to rtx pointer.
3952 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3953 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3954 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3955 depending on the chosen ISAs.
3956 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3957 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3958 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3960 * config/i386/sse.md (mov<mode>_internal): Likewise.
3961 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3963 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3965 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3967 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3969 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3970 parameters from prototype.
3971 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3972 parameters. Don't print a comment.
3973 (emit_fusion_gpr_load): Adjust.
3974 (emit_fusion_load_store): Adjust.
3975 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3976 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3977 comment on the second line.
3979 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3982 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3983 rs6000_cpu based on cpu_index, not tune_index.
3985 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3987 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3988 which_alternative instead of which_alternative + 1.
3989 (output_asm_insn): Print an extra tab if the template is short.
3991 2017-12-01 Jim Wilson <jimw@sifive.com>
3993 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3995 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3996 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3998 * doc/tm.texi: Regenerate.
4000 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4003 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4004 whether we can allocate pseudos before trying to fix an address.
4005 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4006 memory address is indexed or indirect.
4007 (floatuns_<mode>si2_hw2): Likewise.
4009 2017-12-01 Jason Merrill <jason@redhat.com>
4011 * Makefile.in (TAGS): Add c-family/*.cc.
4013 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4015 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4016 (call_value_insn): Likewise.
4017 (sibcall_insn): Likewise.
4018 (sibcall_value_insn): Likewise.
4019 (movsi_aarch64): Likewise.
4020 (movdi_aarch64): Likewise.
4021 (add_losym_): Likewise.
4022 (ldr_got_small_): Likewise.
4023 (ldr_got_small_sidi): Likewise.
4024 (ldr_got_small_28k_): Likewise.
4025 (ldr_got_small_28k_sidi): Likewise.
4026 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4027 Move output_addr_const to symbolic case. Add error check.
4029 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4031 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4032 * config/s390/s390.md: Disable MVC merging peephole if it would
4033 disable operand forwarding.
4034 (new peephole2): Split MVCs if it would turn them into up to 2
4037 2017-12-01 Richard Biener <rguenther@suse.de>
4039 PR tree-optimization/83232
4040 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4041 detection of same access. Instead of breaking the group here
4042 do not consider the duplicate. Add comment explaining real fix.
4044 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4046 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4048 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4050 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4052 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4054 * function.h (struct function): Remove cilk_frame_decl,
4055 is_cilk_function and calls_cilk_spawn fields.
4056 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4058 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4059 cilk_elemental field.
4060 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4061 * target.def: Adjust comment.
4062 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4063 Don't test cilk_elemental.
4065 PR tree-optimization/83233
4066 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4067 bswap_stat name for the struct.
4070 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4071 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4072 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4073 SWITCH_BREAK_LABEL_P set on the label.
4074 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4075 added for default case if it was missing and not all cases covered.
4076 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4077 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4079 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4080 to avoid -Wimplicit-fallthrough warning.
4081 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4082 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4084 PR tree-optimization/83221
4085 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4087 (init_reassoc): Formatting fix.
4090 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4091 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4095 * tree-cfg.c: Include asan.h.
4096 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4098 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4100 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4102 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4104 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4105 vpcmpeqd instruction.
4107 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4109 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4110 registers type for 128bit mode.
4112 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4114 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4116 * gengtype-state.c (read_state_pair): Likewise.
4117 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4118 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4119 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4120 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4121 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4122 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4123 * builtins.c (fold_builtin_3): Likewise.
4124 * graphite-scop-detection.c
4125 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4126 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4129 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4132 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4133 parameter. Return true if ignore_decl_rtl_set_p is true and other
4134 conditions are satisfied.
4135 * asan.h (asan_protect_global): Add new parameter.
4136 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4137 to asan_protect_global calls.
4139 2017-11-30 Jim Wilson <jimw@sifive.com>
4141 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4142 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4143 -mstrict-align, add info on default value. Delete redundant lines for
4144 -mabi. Add missing -mexplicit-relocs docs.
4146 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4148 * config/arc/arc.md (trap): New pattern.
4150 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4152 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4153 instruction to end into a delay slot.
4154 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4155 can be placed into a delay slot against reg_note.
4157 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4159 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4160 labels number of usages.
4162 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4164 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4166 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4168 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4171 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4172 multiplication by power of 2 constant into two shifts + comparison.
4174 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4177 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4179 2017-11-30 Richard Biener <rguenther@suse.de>
4181 PR tree-optimization/83202
4182 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4183 allow_peel argument and guard peeling.
4184 (canonicalize_loop_induction_variables): Likewise.
4185 (canonicalize_induction_variables): Pass false.
4186 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4187 peeling from cunrolli.
4189 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4191 * combine.c (try_combine): Print a message to dump file whenever
4192 I0, I1, or I2 cannot be combined into I3.
4194 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4196 PR rtl-optimization/83156
4197 PR rtl-optimization/82621
4198 * combine.c (try_combine): Don't split an I2 if one of the dests is
4199 set again before I3. Allow unused dests.
4201 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4203 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4205 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4207 PR rtl-optimization/80818
4208 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4209 recursively. Use insn code for clobber.
4210 (lra_set_insn_recog_data): Pass the new arg to
4211 collect_non_operand_hard_regs.
4212 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4214 (lra_update_insn_regno_info): Pass insn to
4215 add_regs_to_insn_regno_info.
4217 2017-11-29 Jim Wilson <jimw@sifive.com>
4218 Andrew Waterman <andrew@sifive.com>
4220 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4221 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4222 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4223 use SINGLE_SHIFT_COST.
4224 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4225 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4227 2017-11-29 Julia Koval <julia.koval@intel.com>
4229 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4230 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4231 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4232 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4233 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4234 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4235 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4236 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4237 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4238 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4239 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4240 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4241 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4242 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4243 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4244 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4245 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4246 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4247 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4248 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4249 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4250 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4251 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4252 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4253 __builtin_ia32_vpshldv_v2di_mask,
4254 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4255 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4256 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4258 2017-11-29 Julia Koval <julia.koval@intel.com>
4260 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4261 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4262 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4263 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4264 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4265 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4266 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4267 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4268 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4269 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4270 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4271 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4272 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4273 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4274 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4275 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4276 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4277 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4278 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4279 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4280 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4282 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4284 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4285 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4288 2017-11-29 Martin Aberg <maberg@gaisler.com>
4290 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4291 to prevent b2bst errata sequence.
4292 (sqrtdf2_fix): Likewise.
4294 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4296 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4297 (div_sqrt_insn_p): New function.
4298 (sparc_do_work_around_errata): Insert NOP instructions to
4299 prevent sequences that could trigger the TN-0013 errata for
4300 certain LEON3 processors.
4301 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4302 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4303 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4304 (in_branch_delay): Prevent div and sqrt in delay slot if
4306 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4308 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4310 * config/sparc/sparc.c (atomic_insn_p): New function.
4311 (sparc_do_work_around_errata): Insert NOP instructions to
4312 prevent sequences that could trigger the TN-0010 errata for
4314 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4315 instruction referable in atomic_insns_p.
4317 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4319 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4320 (atomic_compare_and_swap_leon3_1): Likewise.
4323 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4325 * config/sparc/sparc.c (fpop_insn_p): New function.
4326 (sparc_do_work_around_errata): Insert NOP instructions to
4327 prevent sequences that could trigger the TN-0012 errata for
4329 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4330 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4331 (in_branch_annul_delay): Prevent floating-point instructions
4332 in delay slot of annulled integer branch.
4334 2017-11-29 Richard Biener <rguenther@suse.de>
4336 PR tree-optimization/83202
4337 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4339 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4340 nodes vectorized to the same stmts multiple times.
4341 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4342 (vect_analyze_slp_instance): Adjust.
4343 (scalar_stmts_to_slp_tree_map_t): New typedef.
4344 (vect_schedule_slp_instance): Add a map recording the SLP node
4345 representing the vectorized stmts for a set of scalar stmts.
4346 Avoid code-generating redundancies.
4347 (vect_schedule_slp): Allocate map and pass it down.
4349 2017-11-29 Nathan Sidwell <nathan@acm.org>
4352 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4355 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4357 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4358 cases for printing LDP/STP memory addresses.
4359 (aarch64_print_address_internal): Renamed from
4360 aarch64_print_operand_address, added parameter, add Pmode check.
4361 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4362 (aarch64_print_operand_address): Indirect to
4363 aarch64_print_address_internal.
4364 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4365 'y' operand output specifier.
4367 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4370 * tree.c (build_simple_mem_ref_loc): Handle
4371 get_addr_base_and_unit_offset returning a MEM_REF.
4374 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4377 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4380 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4382 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4384 (layout::move_to_column): Likewise.
4386 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4389 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4390 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4392 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4393 Martin Jambor <mjambor@suse.cz>
4396 * tree.h (expr_type_first_operand_type_p): Declare
4397 * tree.c (expr_type_first_operand_type_p): New function.
4398 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4399 (ipa_value_from_jfunc): Adjust declaration.
4400 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4401 Use it as result type for arithmetics, unless it is NULL in which case
4402 be more conservative.
4403 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4404 ipa_get_jf_pass_through_result.
4405 (propagate_vals_across_pass_through): Likewise.
4406 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4407 is to propagate_vals_across_pass_through.
4408 (propagate_constants_across_call): Pass PARM_TYPE to
4409 propagate_scalar_across_jump_function.
4410 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4411 ipa_value_from_jfunc.
4412 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4413 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4414 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4415 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4416 target_type, pass it to ipa_value_from_jfunc.
4417 (update_indirect_edges_after_inlining): Pass parameter type to
4418 try_make_edge_direct_simple_call.
4420 2017-11-28 Jeff Law <law@redhat.com>
4422 * gimple-ssa-evrp-analyze.c
4423 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4424 refine ranges if scev_initialized_p returns true.
4425 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4427 2017-11-28 Julia Koval <julia.koval@intel.com>
4429 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4430 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4431 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4432 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4433 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4434 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4435 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4436 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4437 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4438 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4439 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4440 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4441 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4442 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4443 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4444 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4445 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4446 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4447 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4448 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4449 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4450 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4452 2017-11-28 Julia Koval <julia.koval@intel.com>
4454 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4455 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4456 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4457 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4458 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4459 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4460 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4461 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4462 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4463 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4464 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4465 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4466 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4467 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4468 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4469 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4470 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4471 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4472 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4473 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4474 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4475 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4476 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4477 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4478 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4479 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4480 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4481 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4482 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4483 __builtin_ia32_vpshld_v2di_mask): New builtins.
4484 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4485 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4487 2017-11-28 Richard Biener <rguenther@suse.de>
4489 PR tree-optimization/80776
4490 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4492 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4494 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4495 If the incoming edge is an effective fallthru because the other
4496 edge only reaches a __builtin_unreachable () then record ranges
4497 derived from the controlling condition in SSA info.
4498 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4499 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4501 2017-11-28 Olivier Hainque <hainque@adacore.com>
4503 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4504 on mingw build hosts.
4506 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4509 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4511 2017-11-28 Richard Biener <rguenther@suse.de>
4514 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4515 copies generated from memcpy use a character array as reference
4518 2017-11-28 Julia Koval <julia.koval@intel.com>
4519 Sebastian Peryt <sebastian.peryt@intel.com>
4521 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4522 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4523 cilk-common.o, cilk.h, cilk-common.c): Remove.
4525 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4526 * builtins.c (is_builtin_name): Remove cilkplus condition.
4527 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4528 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4529 cilk-builtins.def, cilkplus.def): Remove.
4530 * cif-code.def (CILK_SPAWN): Remove.
4531 * cilk-builtins.def: Delete.
4532 * cilk-common.c: Ditto.
4534 * cilkplus.def: Ditto.
4535 * config/darwin.h (fcilkplus): Delete.
4536 * cppbuiltin.c: Ditto.
4537 * doc/extend.texi: Remove cilkplus doc.
4538 * doc/generic.texi: Ditto.
4539 * doc/invoke.texi: Ditto.
4540 * doc/passes.texi: Ditto.
4541 * gcc.c (fcilkplus): Remove.
4542 * gengtype.c (cilk.h): Remove.
4543 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4545 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4547 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4549 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4550 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4551 cilkplus conditions.
4552 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4553 inline_read_section): Ditto.
4554 * ipa-inline-analysis.c (cilk.h): Remove.
4555 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4556 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4557 append_linker_options): Remove condition for fcilkplus.
4558 * lto/lto-lang.c (cilk.h): Remove.
4559 (lto_init): Remove condition for fcilkplus.
4560 * omp-expand.c (expand_cilk_for_call): Delete.
4561 (expand_omp_taskreg, expand_omp_for_static_chunk,
4562 expand_omp_for): Remove cilkplus
4564 (expand_cilk_for): Delete.
4565 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4566 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4567 execute_lower_omp, diagnose_sb_0): Ditto.
4568 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4569 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4570 * tree-nested.c: Ditto.
4571 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4572 (dump_generic_node): Ditto.
4573 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4574 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4576 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4578 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4580 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4581 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4582 for vector mode and !TARGET_SIMD.
4584 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4586 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4588 * tree.h (SWITCH_LABELS): Remove.
4589 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4590 assert SWITCH_BODY is non-NULL.
4591 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4593 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4595 PR tree-optimization/80788
4596 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4597 has TREE_OVERFLOW set, call drop_tree_overflow.
4599 2017-11-28 Richard Biener <rguenther@suse.de>
4601 PR tree-optimization/83158
4602 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4604 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4607 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4608 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4610 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4612 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4613 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4615 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4618 * varasm.c (compare_constant): Take the mode of the constants into
4619 account when comparing floating point constants.
4621 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4623 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4624 from explicit instantiation of debug_helper.
4625 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4627 2017-11-27 Richard Biener <rguenther@suse.de>
4629 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4632 2017-11-27 Richard Biener <rguenther@suse.de>
4634 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4635 (type_hash_canon): Also clear int_cst_hash_table entry for
4637 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4639 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4641 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4642 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4643 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4645 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4647 * hash-map.h (gt_cleare_cache): Avoid UB.
4649 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4651 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4652 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4653 (decide_unroll_runtime_iterations): Likewise.
4654 (decide_unroll_stupid): Likewise.
4656 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4659 * config/i386/i386.md: Add a loop with incssp.
4661 2017-11-27 Martin Jambor <mjambor@suse.cz>
4663 PR tree-optimization/81248
4664 * tree-sra.c (splice_param_accesses): Remove size check.
4665 (decide_one_param_reduction): Fix size check.
4666 * gimple-pretty-print.c (dump_profile): Silence warning.
4667 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4669 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4672 * dbxout.c (lastlineno): New variable.
4673 (dbx_debug_hooks): Use dbxout_switch_text_section as
4674 switch_text_section debug hook.
4675 (dbxout_function_end): Switch to current_function_section
4676 rather than function_section. If crtl->has_bb_partition,
4677 output just one N_FUN, depending on in_cold_section_p.
4678 (dbxout_source_line): Remember last lineno in lastlineno.
4679 (dbxout_switch_text_section): New function.
4680 (dbxout_function_decl): Adjust dbxout_block caller.
4681 (dbx_block_with_cold_children): New function.
4682 (dbxout_block): Return true if any LBRAC/RBRAC have been
4683 emitted. Use dbx_block_with_cold_children at depth == 0
4684 in second partition. Add PARENT_BLOCKNUM argument, pass
4685 it optionally adjusted to children. Output LBRAC/RBRAC
4686 around recursive call only if the block is in the current
4687 partition, if not and anything was output, emit empty
4689 * final.c (final_scan_insn): Compute cold_function_name
4690 before calling switch_text_section debug hook. Call
4691 that hook even if dwarf2out_do_frame if not emitting
4695 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4696 TREE_READONLY decls.
4698 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4700 PR rtl-optimization/82488
4701 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4703 2017-11-26 Julia Koval <julia.koval@intel.com>
4705 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4707 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4710 2017-11-26 Julia Koval <julia.koval@intel.com>
4712 * config/i386/driver-i386.c (host_detect_local_cpu):
4713 Detect skylake-avx512.
4715 2017-11-26 Julia Koval <julia.koval@intel.com>
4717 * config.gcc: Add -march=cannonlake.
4718 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4719 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4720 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4721 (PTA_CANNONLAKE): New.
4722 (processor_target_table): Add cannonlake.
4723 (ix86_option_override_internal): Ditto.
4724 (fold_builtin_cpu): Ditto.
4725 (get_builtin_code_for_version): Handle cannonlake.
4726 (M_INTEL_COREI7_CANNONLAKE): New.
4727 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4728 * doc/invoke.texi: Add -march=cannonlake.
4730 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4732 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4733 (try_init_one_plugin): Alternative implementation for MinGW.
4734 * Makefile.in (plugin_implib): New.
4735 (gengtype-lex.c): Fix broken AIX workaround.
4736 * configure: Regenerate.
4737 * doc/plugins.texi: Document support for MinGW.
4739 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4741 PR rtl-optimization/81553
4742 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4743 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4744 is a shift where C1 has different mode than the whole shift, use C1's
4745 mode for MULT rather than the shift's mode.
4748 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4749 builtins not enabled in the currently selected ISA.
4751 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4753 PR tree-optimization/71026
4754 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4755 (insert_reciprocals): Change to insert reciprocals before a division
4756 by a square and to insert the square of a reciprocal.
4757 (execute_cse_reciprocals_1): Change to consider division by a square.
4758 (register_division_in): Add importance parameter.
4760 2017-11-24 Richard Biener <rguenther@suse.de>
4762 PR tree-optimization/82402
4763 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4764 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4766 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4768 * match.pd (0-ptr): New transformation.
4770 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4773 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4774 when skipping checking.
4776 2017-11-24 Richard Biener <rguenther@suse.de>
4778 PR tree-optimization/83128
4779 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4780 (vn_reference_lookup_3): Likewise.
4782 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4785 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4786 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4789 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4790 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4791 check for the latter with gimple_call_builtin_p. Do not handle
4792 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4794 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4796 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4799 2017-11-23 Julia Koval <julia.koval@intel.com>
4801 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4802 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4803 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4804 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4805 _mm512_maskz_expandloadu_epi16): New intrinsics.
4806 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4807 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4808 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4809 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4810 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4811 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4812 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4813 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4814 _mm256_maskz_expandloadu_epi8): New intrinsics.
4815 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4816 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4817 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4818 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4819 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4820 * config/i386/sse.md (VI248_VLBW): New iterator.
4821 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4823 2017-11-23 Julia Koval <julia.koval@intel.com>
4825 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4826 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4827 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4828 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4829 _mm512_mask_compressstoreu_epi16): New.
4830 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4831 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4832 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4833 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4834 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4835 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4836 _mm256_mask_compressstoreu_epi8): New.
4837 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4838 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4839 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4840 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4841 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4842 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4843 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4844 __builtin_ia32_compresshi128_mask,
4845 __builtin_ia32_compressstoreuqi512_mask,
4846 __builtin_ia32_compressstoreuhi512_mask,
4847 __builtin_ia32_compressstoreuqi256_mask,
4848 __builtin_ia32_compressstoreuqi128_mask,
4849 __builtin_ia32_compressstoreuhi256_mask,
4850 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4851 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4853 (ix86_expand_special_args_builtin): Handle new types.
4854 (s4fma_expand): Handle new builtin array.
4855 * config/i386/immintrin.h: Include new headers.
4856 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4857 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4859 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4861 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4862 fixes. Declare temp and g variables at the top in order to avoid
4863 {} in most of the cases.
4865 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4867 * match.pd (ptr-0): New transformation.
4869 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4871 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4872 (struct addr_mode_cost_table): New.
4873 (struct tune_params): Add field addr_mode_costs.
4874 * config/arm/arm.c (generic_addr_mode_costs): New.
4875 (arm_slowmul_tune): Initialise addr_mode_costs field.
4876 (arm_fastmul_tune): Likewise.
4877 (arm_strongarm_tune): Likewise.
4878 (arm_xscale_tune): Likewise.
4879 (arm_9e_tune): Likewise.
4880 (arm_marvell_pj4_tune): Likewise.
4881 (arm_v6t2_tune): Likewise.
4882 (arm_cortex_tune): Likewise.
4883 (arm_cortex_a8_tune): Likewise.
4884 (arm_cortex_a7_tune): Likewise.
4885 (arm_cortex_a15_tune): Likewise.
4886 (arm_cortex_a35_tune): Likewise.
4887 (arm_cortex_a53_tune): Likewise.
4888 (arm_cortex_a57_tune): Likewise.
4889 (arm_exynosm1_tune): Likewise.
4890 (arm_xgene1_tune): Likewise.
4891 (arm_cortex_a5_tune): Likewise.
4892 (arm_cortex_a9_tune): Likewise.
4893 (arm_cortex_a12_tune): Likewise.
4894 (arm_cortex_a73_tune): Likewise.
4895 (arm_v7m_tune): Likewise.
4896 (arm_cortex_m7_tune): Likewise.
4897 (arm_v6m_tune): Likewise.
4898 (arm_fa726te_tune): Likewise.
4899 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4901 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4903 * config/arm/arm.c (arm_mem_costs): New function.
4904 (arm_rtx_costs_internal): Use arm_mem_costs.
4906 2017-11-23 Mark Wielaard <mark@klomp.org>
4908 * dwarf2out.c (init_sections_and_labels): Use generation to create
4909 unique ranges_section_label and ranges_base_label. Return generation.
4910 (output_rnglists): Add generation argument. Use generation to create
4911 unique ranges labels.
4912 (dwarf2out_finish): Get generation from init_sections_and_labels
4913 and pass generation to output_rnglists.
4915 2017-11-23 Mike Stump <mikestump@comcast.net>
4916 Eric Botcazou <ebotcazou@adacore.com>
4918 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4919 * cfgloop.h (struct loop): Add unroll field.
4920 * function.h (struct function): Add has_unroll bitfield.
4921 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4922 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4923 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4924 (pass_rtl_unroll_loops::gate): Likewise.
4925 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4926 for which loop->unroll==1.
4927 (decide_unroll_constant_iterations): Use note for consistency and
4928 take loop->unroll into account. Return early if loop->unroll is set.
4929 Fix thinko in existing test.
4930 (decide_unroll_runtime_iterations): Use note for consistency and
4931 take loop->unroll into account.
4932 (decide_unroll_stupid): Likewise.
4933 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4934 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4935 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4937 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4938 (print_loop): Print loop->unroll if set.
4939 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4940 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4941 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4943 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4944 loop->unroll is set and smaller than the trip count. Otherwise bypass
4945 entirely the heuristics if loop->unroll is set. Remove dead note.
4946 Fix off-by-one bug in other note.
4947 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4948 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4950 (tree_unroll_loops_completely): Make static.
4951 (pass_complete_unroll::execute): Use correct type for variable.
4952 (pass_complete_unrolli::execute): Fix formatting.
4953 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4955 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4957 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4958 enable when TARGET_PREFER_AVX128 is set.
4960 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4962 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4964 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4965 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4967 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4969 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4970 frequencies. Fix estimation of aggregate parameters.
4972 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4974 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4977 2017-11-23 Tom de Vries <tom@codesourcery.com>
4979 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4981 2017-11-23 Tom de Vries <tom@codesourcery.com>
4983 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4986 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4989 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4990 sibcall_value_pcrel_fdpic): Use local variable instead of
4992 (calli_tbr_rel): Add missing operand 2.
4993 (call_valuei_tbr_rel): Add missing operand 3.
4995 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4998 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4999 bitpos/bitsize covering the whole destination, use store_expr only if
5000 the complex mode is the same. Otherwise, use expand_normal and if
5001 it returns CONCAT, subreg each part separately instead of trying to
5002 subreg the whole result.
5004 2017-11-23 Richard Biener <rguenther@suse.de>
5006 PR tree-optimization/23094
5007 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5008 come here from walking over backedges in the first iteration.
5009 (vn_reference_lookup_3): Skip clobbers that store the same value.
5011 2017-11-23 Richard Biener <rguenther@suse.de>
5013 PR tree-optimization/81403
5014 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5015 a block we need a leader relative to.
5016 (phi_translate_1): For nary processing require a leader from
5017 get_representative_for given we run expression simplification
5018 using match-and-simplify. Remove previous fix.
5020 2017-11-22 Jeff Law <law@redhat.com>
5022 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5023 Use new method allocate_value_range rather than accessing the
5024 vrp_value_range_pool data member directly.
5025 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5026 to use extract_range_from_stmt method to avoid need for
5027 extract_range_from_assignment method.
5028 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5029 method rather than setting values_propgated data member directly.
5030 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5031 and values propagated data members and extract_range_from_assignment
5032 method. Reorder private data members to conform to standards.
5033 Add new methods set_lattice_propagation_complete and
5034 allocate_value_range.
5036 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5038 PR rtl-optimization/83030
5039 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5040 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5041 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5042 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5043 CROSSING_JUMP_P flag is already set before setting it.
5044 * cfgrtl.c (fixup_partition_crossing): Likewise.
5045 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5048 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5050 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5051 Handle the case where both arguments are using gen_const_vec_series.
5053 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5056 * pretty-print.c (pp_format): Move quoting implementation to
5057 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5058 to pass address of "quote" local.
5059 (pp_begin_quote): New function.
5060 (pp_end_quote): New function.
5061 * pretty-print.h (printer_fn): Convert penultimate param from bool
5063 (pp_begin_quote): New decl.
5064 (pp_end_quote): New decl.
5065 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5066 param from bool to bool *.
5067 * tree-diagnostic.h (default_tree_printer): Likewise.
5069 2017-11-22 Jeff Law <law@redhat.com>
5071 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5072 Initialize vr_values.
5073 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5074 vr_values via vr_values class instance rather than delegators.
5075 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5076 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5077 (evrp_range_analyzer::push_value_range): Likewise.
5078 (evrp_range_analyzer::pop_value_range): Likewise.
5079 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5080 most delegators. Those remaining are exposed as public interfaces.
5081 Make vr_values a pointer and private.
5082 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5084 (evrp_range_analyzer::get_vr_value): New method.
5085 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5086 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5087 (class evrp_dom_walker): Attach evrp_folder class, initialize
5088 it in the ctor. Remove temporary delegators.
5089 (evrp_dom_walker::before_dom_children): Call methods in attached
5090 evrp_range_analyzer class via class instance pointer. Use
5091 free value_range_constant_singleton to remove need for
5092 op_with_constant_singleton_value delegator method. Do not
5093 create a vrp_prop class instance for every call! Narrow
5094 scope of a couple variables.
5095 (evrp_dom_walker::cleanup): Call methods in attached
5096 evrp_range_analyzer class via class instance pointer.
5097 * vr-values.h (class vr_values): Privatize many methods and
5100 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5102 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5103 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5104 * expr.c (expand_expr_real_2): Likewise.
5105 * fold-const.c (const_unop): Likewise.
5106 * optabs-tree.c (optab_for_tree_code): Likewise.
5107 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5108 * tree-inline.c (estimate_operator_cost): Likewise.
5109 * tree-pretty-print.c (dump_generic_node): Likewise.
5110 (op_code_prio): Likewise.
5111 (op_symbol_code): Likewise.
5112 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5113 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5114 * internal-fn.c (direct_internal_fn_optab): New function.
5115 (direct_internal_fn_array, direct_internal_fn_supported_p
5116 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5117 * fold-const-call.c (fold_const_reduction): New function.
5118 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5120 * tree-vect-loop.c: Include internal-fn.h.
5121 (reduction_code_for_scalar_code): Rename to...
5122 (reduction_fn_for_scalar_code): ...this and return an internal
5124 (vect_model_reduction_cost): Take an internal_fn rather than
5126 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5128 (vectorizable_reduction): Use internal functions rather than tree
5129 codes for the reduction operation. Update calls to the functions
5131 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5132 Use calls to internal functions rather than REDUC tree codes.
5133 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5135 2017-11-22 Olivier Hainque <hainque@adacore.com>
5137 * config/vxworks.c (vxworks_override_options): Pick default
5138 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5139 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5140 DWARF_GNAT_ENCODINGS_DEFAULT.
5141 * config/vxworksae.h: Likewise.
5143 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5145 PR tree-optimization/83104
5146 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5149 2017-11-22 Marek Polacek <polacek@redhat.com>
5150 H.J. Lu <hongjiu.lu@intel.com>
5151 Jason Merrill <jason@redhat.com>
5156 * calls.c (initialize_argument_information): Call
5157 warn_parameter_passing_abi target hook.
5158 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5159 argument onto stack.
5160 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5161 * common.opt: Update -fabi-version description.
5162 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5163 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5165 (ix86_is_empty_record): New function.
5166 (ix86_warn_parameter_passing_abi): New function.
5167 (TARGET_EMPTY_RECORD_P): Redefine.
5168 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5169 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5170 * doc/tm.texi: Regenerated.
5171 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5172 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5173 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5174 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5175 instead of int_size_in_bytes.
5176 * expr.c (copy_blkmode_to_reg): Likewise.
5177 * function.c (aggregate_value_p): Return 0 for empty types.
5178 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5180 (locate_and_pad_parm): Call arg size_in_bytes instead
5182 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5183 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5184 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5186 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5187 (std_gimplify_va_arg_expr): Skip empty records. Call
5188 arg_size_in_bytes instead size_in_bytes.
5189 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5190 * tree-core.h (tree_type_common): Add empty_flag.
5191 (tree_decl_common): Update comments.
5192 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5194 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5195 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5197 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5198 * tree.c (default_is_empty_type): New function.
5199 (default_is_empty_record): New function.
5200 (arg_int_size_in_bytes): New function.
5201 (arg_size_in_bytes): New function.
5202 (get_ultimate_context): New function.
5203 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5204 TRANSLATION_UNIT_WARN_EMPTY_P.
5205 (default_is_empty_record, arg_int_size_in_bytes,
5206 arg_size_in_bytes, get_ultimate_context): Declare.
5208 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5210 * config/arm/arm.c (cmse_clear_registers): New function.
5211 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5212 code by call to cmse_clear_registers.
5213 (cmse_nonsecure_entry_clear_before_return): Likewise.
5215 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5217 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5218 (vdot_s32, vdotq_s32): New.
5219 (vdot_lane_u32, vdotq_lane_u32): New.
5220 (vdot_lane_s32, vdotq_lane_s32): New.
5223 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5226 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5227 for unsigned values with fewer HWIs than the precision.
5228 (test_overflow): New function.
5229 (wide_int_cc_tests): Call it.
5231 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5233 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5234 has the same mode class as Pmode.
5236 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5238 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5239 padding_bits_to_clear_ptr.
5240 (cmse_nonsecure_entry_clear_before_return): Likewise.
5242 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5244 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5245 auto_sbitap instead of integer bitfield to control register needing
5248 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5250 PR tree-optimization/83044
5251 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5252 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5253 that need the upper bound. Subtract offset from
5254 get_addr_base_and_unit_offset only if positive and subtract it
5255 before division by eltsize rather than after it.
5258 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5259 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5260 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5264 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5265 expand_binop with *mul_widen_optab, make sure at least one of the
5266 operands doesn't have VOIDmode.
5269 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5271 PR rtl-optimization/82044
5272 PR tree-optimization/82042
5273 * dse.c (record_store): Check for overflow.
5274 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5275 add_wild_read instead of clear_rhs_from_active_local_stores on
5276 overflow and log it into dump_file.
5278 2017-11-22 Richard Biener <rguenther@suse.de>
5280 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5281 fake edges to exit when looking for a place to insert.
5282 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5284 (insert_into_preds_of_block): Commit edge insertion immediately,
5285 assert that doesn't require new BBs.
5286 (fini_pre): Release expressions.
5287 (pass_pre::execute): Shuffle things around a bit, if the fn
5288 is too large do not compute AVAIL either as this is really the
5291 2017-11-22 Richard Biener <rguenther@suse.de>
5293 PR tree-optimization/83089
5294 * tree-if-conv.c (pass_if_conversion::execute): If anything
5295 changed reset SCEV and free the number of iteration estimates.
5297 2017-11-21 Martin Sebor <msebor@redhat.com>
5299 PR tree-optimization/82945
5300 * calls.h (warn_nonstring_bound): Remove unused function.
5302 2017-11-21 Martin Sebor <msebor@redhat.com>
5304 PR tree-optimization/82945
5305 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5306 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5307 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5309 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5310 * calls.h (get_attr_nonstring_decl): Declare new function.
5311 * doc/extend.texi (attribute nonstring): Update.
5312 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5313 get_attr_nonstring_decl and handle it.
5314 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5315 detection of nul-termination.
5316 (strlen_to_stridx): Change to a pointer.
5317 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5318 (pass_strlen::execute): Same.
5320 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5322 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5323 for the new option -mprefer-vector-width=[none|128|256|512].
5324 * config/i386/i386.c (ix86_target_string): remove old style options
5325 -mprefer-avx256 and make -mprefer-avx128 as alias.
5326 (ix86_option_override_internal): Apply defaults for the
5327 -mprefer-vector-width=[128|256] option.
5328 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5329 Implement macros to work with -mprefer-vector-width=.
5330 * config/i386/i386.opt: Implemented option
5331 -mprefer-vector-width=[none|128|256|512].
5332 * doc/invoke.texi: Documentation for
5333 -mprefer-vector-width=[none|128|256|512].
5335 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5337 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5339 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5341 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5343 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5345 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5347 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5348 * expr.c (expand_expr_real_2): Likewise.
5349 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5350 fold_binary_loc): Likewise.
5351 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5352 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5353 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5354 MINUS_EXPR transformations.
5355 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5356 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5357 * tree-inline.c (estimate_operator_cost): Likewise.
5358 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5359 op_symbol_code): Likewise.
5360 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5361 * vr-values.c (extract_range_from_binary_expr): Likewise.
5362 * varasm.c (initializer_constant_valid_p_1): Likewise.
5363 * tree.def: New tree code POINTER_DIFF_EXPR.
5365 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5367 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5368 integer suffix to movbe mnemonic.
5369 (*bswaphi2_movbe): Ditto.
5370 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5372 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5375 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5376 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5377 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5379 2017-11-21 Martin Liska <mliska@suse.cz>
5381 * tree-inline.c (expand_call_inline): Remove not needed
5384 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5385 Jakub Jelinek <jakub@redhat.com>
5388 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5389 Remove static keyword from f variable.
5391 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5393 PR tree-optimization/83086
5394 * gimple-ssa-store-merging.c
5395 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5396 rather than n.base_addr.
5398 2017-11-21 Martin Liska <mliska@suse.cz>
5400 PR rtl-optimization/82044
5401 PR tree-optimization/82042
5402 * dse.c (check_mem_read_rtx): Check for overflow.
5404 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5406 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5409 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5411 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5412 and its corresponding call arg location note.
5414 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5415 Andrew Burgess <andrew.burgess@embecosm.com>
5417 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5419 (arc_return_slot_offset): Likewise.
5420 (arc_eh_return_address_location): New declaration.
5421 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5422 (MUST_SAVE_REGISTER): Add exception handler case.
5423 (MUST_SAVE_RETURN_ADDR): Likewise.
5424 (arc_frame_pointer_required): Likewise.
5425 (arc_frame_pointer_needed): New function.
5426 (arc_compute_frame_size): Changed.
5427 (arc_expand_prologue): Likewise.
5428 (arc_expand_epilogue): Likewise.
5429 (arc_initial_elimination_offset): Likewise.
5430 (arc_return_slot_offset): Delete.
5431 (arc_eh_return_address_location): New function.
5432 (arc_builtin_setjmp_frame_value): Likewise.
5433 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5434 (EH_RETURN_STACKADJ_RTX): Define.
5435 (EH_RETURN_HANDLER_RTX): Likewise.
5436 * config/arc/arc.md (eh_return): Delete.
5438 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5440 * print-tree.h (debug_vec_tree): Remove prototype.
5441 * gdbinit.in (pvt): Remove macro.
5443 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5445 PR tree-optimization/83047
5446 * gimple-ssa-store-merging.c
5447 (imm_store_chain_info::output_merged_store): If the loads with the
5448 same vuse are in different basic blocks, for load_gsi pick a load
5449 location that is dominated by the other loads.
5452 * config/i386/i386.c (ix86_memmodel_check): Start
5453 -Winvalid-memory-model diagnostics with lowercase letter.
5456 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5457 set in_cold_section_p to the partition containing loc_list->first.
5458 When seeing loc_list->last_before_switch node, update secname and
5459 perform range_across_switch second partition handling only after that.
5462 * run-rtl-passes.c: Include debug.h.
5463 (run_rtl_passes): Call debug_hooks->assembly_start.
5464 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5468 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5469 OPTAB_DIRECT in calls to expand_simple_binop.
5471 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5474 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5476 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5479 * spellcheck.h (best_match::blithely_get_best_candidate): New
5482 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5484 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5485 with lower case letter.
5487 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5489 * config/i386/i386.md (bswaphi2): New expander.
5490 (*bswaphi2_movbe): New insn pattern.
5491 (bswaphi -> rorhi pepehole2): New peephole pattern.
5493 2017-11-20 Jeff Law <law@redhat.com>
5495 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5496 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5497 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5498 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5499 gimple-ssa-evrp-analyze.h.
5501 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5502 set BB_VISITED here.
5503 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5505 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5507 * tree-predcom.c: Add general comment on Store-Store chains.
5508 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5510 (get_chain_last_ref_at): Rename into...
5511 (get_chain_last_write_at): ...this.
5512 (get_chain_last_write_before_load): New function.
5513 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5514 CT_STORE_STORE when write reference is added.
5515 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5516 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5517 (initialize_root_vars_store_elim_1): Ditto.
5518 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5519 definition is created.
5520 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5521 chain by replacing it with dominant stored value.
5523 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5525 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5527 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5529 * vr-values.c (extract_range_from_binary_expr): Use a full range
5532 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5534 * config/arm/arm.md (R4_REGNUM): Define constant.
5535 (nonsecure_call_internal): Remove r4 clobber.
5536 (nonsecure_call_value_internal): Likewise.
5537 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5538 clobber and resequence match_operands.
5539 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5540 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5541 (nonsecure_call_value_reg_thumb2): Likewise.
5543 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5545 PR tree-optimization/78821
5546 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5547 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5548 to the address of the base rather than the base itself.
5549 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5550 (find_bswap_or_nop_finalize): New function.
5551 (find_bswap_or_nop): Use it.
5552 (bswap_replace): Return a tree rather than bool, change first
5553 argument from gimple * to gimple_stmt_iterator, allow inserting
5554 into an empty sequence, allow ins_stmt to be NULL - then emit
5555 all stmts into gsi. Fix up MEM_REF address gimplification.
5556 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5557 (struct store_immediate_info): Add N and INS_STMT non-static
5559 (store_immediate_info::store_immediate_info): Initialize them
5560 from newly added ctor args.
5561 (merged_store_group::apply_stores): Formatting fixes. Sort by
5563 (stmts_may_clobber_ref_p): For stores call also
5564 refs_anti_dependent_p.
5565 (gather_bswap_load_refs): New function.
5566 (imm_store_chain_info::try_coalesce_bswap): New method.
5567 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5568 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5569 (imm_store_chain_info::output_merged_store): Fail if number of
5570 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5571 and NOP_EXPR rhs_code.
5572 (pass_store_merging::process_store): Compute n and ins_stmt, if
5573 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5574 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5576 (pass_store_merging::execute): Calculate dominators.
5578 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5579 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5580 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5581 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5582 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5583 class pass_optimize_bswap, bswap_replace,
5584 pass_optimize_bswap::execute): Moved to ...
5585 * gimple-ssa-store-merging.c: ... this file.
5586 Include optabs-tree.h.
5587 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5588 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5589 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5590 anonymous namespace, remove static keywords.
5591 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5592 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5594 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5597 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5599 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5601 * vec.h (debug_helper): New function.
5602 (DEFINE_DEBUG_VEC): New macro.
5603 * hash-set.h (debug_helper): New function.
5604 (DEFINE_DEBUG_HASH_SET): New macro.
5605 * cfg.c (debug_slim (edge)): New function.
5606 Call DEFINE_DEBUG_VEC for edges.
5607 Call DEFINE_DEBUG_HASH_SET for edges.
5608 * cfghooks.c (debug_slim (basic_block)): New function.
5609 Call DEFINE_DEBUG_VEC for basic blocks.
5610 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5611 * print-tree.c (debug_slim): New function to handle trees.
5612 Call DEFINE_DEBUG_VEC for trees.
5613 Call DEFINE_DEBUG_HASH_SET for trees.
5614 (debug (vec<tree, va_gc>) &): Remove.
5615 (debug (<vec<tree, va_gc>) *): Remove.
5616 * print-rtl.c (debug_slim): New function to handle const_rtx.
5617 Call DEFINE_DEBUG_VEC for rtx_def.
5618 Call DEFINE_DEBUG_VEC for rtx_insn.
5619 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5620 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5621 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5622 (debug (vec<rtx_insn *> *ptr): Remove.
5623 (debug_insn_vector): Remove.
5624 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5626 2017-11-20 Tom de Vries <tom@codesourcery.com>
5628 PR rtl-optimization/82020
5629 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5630 IF_THEN_ELSE condition.
5632 2017-11-19 Jeff Law <law@redhat.com>
5634 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5635 of degenerates resulting from ignoring an edge.
5637 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5640 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5642 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5645 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5646 for uninitialied counts.
5648 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5651 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5653 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5656 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5658 2017-11-19 Tom de Vries <tom@codesourcery.com>
5660 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5663 2017-11-19 Tom de Vries <tom@codesourcery.com>
5665 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5668 2017-11-19 Tom de Vries <tom@codesourcery.com>
5670 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5673 2017-11-19 Tom de Vries <tom@codesourcery.com>
5675 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5677 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5679 2017-11-19 Tom de Vries <tom@codesourcery.com>
5681 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5683 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5684 semicolon after MCORE_EXPORT_NAME call.
5686 2017-11-19 Tom de Vries <tom@codesourcery.com>
5689 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5690 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5692 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5694 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5695 (cgraph_edge::clone): Cleanup updating of profile.
5696 * ipa-cp.c (update_profiling_info): Likewise.
5697 * ipa-inline-transform.c (inline_transform): Likewise.
5698 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5699 * ipa-split.c (execute_split_functions): Do not split when function
5701 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5702 * profile-count.c (profile_count::dump): Add global0.
5703 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5705 (profile_count::to_sreal_scale): Likewise.
5706 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5707 (profile_count::combine_with_ipa_count): New function.
5708 * profile-count.h (profile_guessed_global0adjusted): New.
5709 (profile_count::adjusted_zero): New.
5710 (profile_count::global0adjusted): New.
5711 (profile_count::combine_with_ipa_count): New.
5712 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5713 correct profile of return block of split functions.
5714 (copy_cfg_body): Remove unused profile_count.
5715 (copy_body): Likewise.
5716 (expand_call_inline): Update.
5717 (tree_function_versioning): Update.
5719 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5721 * hash-set.h (hash_set::empty): New.
5722 * tree-ssa-threadbackward.h: Delete.
5723 * tree-ssa-threadbackward.c (class thread_jumps): New.
5724 Move max_threaded_paths into class.
5725 (fsm_find_thread_path): Remove arguments that are now in class.
5726 (profitable_jump_thread_path): Rename to...
5727 (thread_jumps::profitable_jump_thread_path): ...this.
5728 (convert_and_register_jump_thread_path): Rename to...
5729 (thread_jumps::convert_and_register_current_path): ...this.
5730 (check_subpath_and_update_thread_path): Rename to...
5731 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5732 (register_jump_thread_path_if_profitable): Rename to...
5733 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5734 (handle_phi): Rename to...
5735 (thread_jumps::handle_phi): ...this.
5736 (handle_assignment): Rename to...
5737 (thread_jumps::handle_assignment): ...this.
5738 (fsm_find_control_statement_thread_paths): Rename to...
5739 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5740 (find_jump_threads_backwards): Rename to...
5741 (thread_jumps::find_jump_threads_backwards): ...this.
5742 Initialize path local data.
5743 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5744 from within thread_jumps class.
5745 (pass_early_thread_jumps::execute): Same.
5747 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5749 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5751 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5753 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5754 to canonical location.
5756 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5759 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5761 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5763 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5765 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5766 rcrt1.o%s/grcrt1.o%s for -static-pie.
5768 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5770 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5771 ix86_shift_rotate_cost): Break out from ...
5772 (ix86_rtx_costs): ... here.
5773 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5776 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5778 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5779 when entry block was promoted unlikely.
5780 (estimate_bb_frequencies): Increase frequency scale.
5781 * profile-count.h (profile_count): Export precision info.
5783 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5785 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5786 disturbin profile of entry block.
5788 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5790 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5793 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5795 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5796 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5798 2017-11-17 Jeff Law <law@redhat.com>
5800 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5801 from evrp_dom_walker class. Various methods moved into new class.
5802 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5803 (evrp_range_analyzer::enter): New method.
5804 (evrp_range_analyzer::leave): New method.
5805 (evrp_dom_walker): Remove delegators no longer needed by this class.
5806 Replace vr_values data member with evrp_range_analyzer
5808 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5809 method extracted from evrp_dom_walker::before_dom_children.
5810 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5811 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5813 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5814 Add private copy constructor and move assignment operators.
5815 Privatize methods and class data where trivially possible.
5816 (evrp_dom_walker::cleanup): New function, extracted from
5817 execute_early_vrp. Simplify access to class data.
5819 * vr-values.h (get_output_for_vrp): Prototype.
5820 * vr-values.c (get_output_for_vrp): New function extracted from
5821 vrp_visit_assignment_or_call and extract_range_from_stmt.
5822 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5824 2017-11-17 Luis Machado <luis.machado@linaro.org>
5826 * config/aarch64/aarch64.c
5827 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5828 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5829 tune_params::AUTOPREFETCHER_WEAK.
5831 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5834 * config/arm/arm.c (arm_valid_target_attribute_rec):
5835 Parse "arch=" and "+<ext>".
5836 (arm_valid_target_attribute_tree): Re-init global options.
5837 (arm_option_override): Make non-static.
5838 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5839 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5840 (__ARM_FEATURE_CRC32): Support undef.
5841 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5842 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5844 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5846 * gdbinit.in (break-on-diagnostic): New command.
5848 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5850 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5851 return to indirect jump for EH return if control-flow
5852 protection is enabled. Change explicit 'false' argument in
5853 pro_epilogue_adjust_stack with a value of
5855 * config/i386/i386.md (simple_return_indirect_internal):
5856 Remove SImode restriction to support 64-bit.
5858 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5860 * combine.c (added_notes_insn): New.
5861 (try_combine): Handle added_notes_insn like added_links_insn.
5862 Rewrite return value code.
5863 (distribute_notes): Set added_notes_insn to the earliest insn we added
5866 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5868 PR rtl-optimization/82621
5869 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5870 dest of one of those SETs is unused.
5872 2017-11-17 Richard Biener <rguenther@suse.de>
5875 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5876 * tree-pretty-print.c (dump_generic_node): Handle
5877 annot_expr_parallel_kind.
5878 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5879 * gimplify.c (gimple_boolify): Likewise.
5881 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5883 * config.gcc (extra_headers): Add cet.h for x86 targets.
5884 * config/i386/cet.h: New file.
5885 * doc/install.texi: Add --enable-cet/--disable-cet.
5887 2017-11-17 Richard Biener <rguenther@suse.de>
5889 PR tree-optimization/83017
5890 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5891 (gen_parallel_loop): Properly count iterations.
5892 (parallelize_loops): Handle loop->can_be_parallel independent
5893 of flag_loop_parallelize_all. Make static profitability test match
5895 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5896 * invoke.texi (parloops-min-per-thread): Document.
5898 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5900 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5901 upstreaming review comments.
5903 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5905 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5906 with fast unaligned access.
5907 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5909 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5911 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5912 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5914 2017-11-17 Richard Biener <rguenther@suse.de>
5916 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5917 folding of references.
5919 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5922 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5923 of replacing call to strncmp with corresponding call to strcmp when
5926 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5928 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5929 option prefer-avx256 for skylake-avx512 configuration.
5930 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5931 (get_builtin_code_for_version): Ditto.
5933 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5934 Monk Chiang <sh.chiang04@gmail.com>
5936 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5937 (FIXED_REGISTERS): Reserve more register numbers.
5938 (CALL_USED_REGISTERS): Likewise.
5939 (REG_ALLOC_ORDER): Likewise.
5940 (REG_CLASS_CONTENTS): Likewise.
5941 (REGISTER_NAMES): Likewise.
5943 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5944 Kito Cheng <kito.cheng@gmail.com>
5946 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5948 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5950 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5952 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5953 parameter list for vec_splats.
5955 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5957 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5959 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5961 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5964 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5965 with the required version of automake 1.11.6.
5967 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5969 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5971 (DU_C2_3_power9): Correct reservation combinations.
5972 (FP_DIV_power9, VEC_DIV_power9): New.
5973 (power9-alu): Split out rotate/shift...
5974 (power9-rot): ...to here, correct dispatch resource.
5975 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5977 (power9-fp): Correct latency.
5978 (power9-sdiv): Add div/sqrt resource.
5979 (power9-ddiv): Correct latency, add div/sqrt resource.
5980 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5981 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5983 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5985 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5987 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5988 switch statement mapping KF built-ins to TF built-ins if we don't
5989 have the proper ISA 3.0 assembler support.
5991 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5993 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5994 (lower_emutls_function_body): Do not compute it.
5996 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5998 * ipa-split.c (split_bb_info): Turn time to sreal.
5999 (split_point): Likewise.
6000 (dump_split_point): Likewise.
6001 (fine_split_points): Likewise.
6002 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6004 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6006 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6009 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6011 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6013 (expensive_function_p): Remove useless assert.
6014 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6016 2017-11-16 Martin Sebor <msebor@redhat.com>
6018 PR tree-optimization/82588
6019 PR tree-optimization/82583
6020 * tree-vrp.c (check_array_ref): Handle flexible array members,
6021 string literals, and inner indices.
6022 (search_for_addr_array): Add detail to diagnostics.
6024 2017-11-16 Nathan Sidwell <nathan@acm.org>
6028 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6029 (SET_DECL_ASSEMBLER_NAME): Forward to
6030 overwrite_decl_assembler_name.
6031 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6032 (overwrite_decl_assembler_name): Declare.
6033 * tree.c (overwrite_decl_assembler_name): New.
6034 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6035 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6036 (LANG_HOOKS_INITIALIZER): Add it.
6037 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6038 * langhooks.c (lhd_set_decl_assembler_name): Use
6039 SET_DECL_ASSEMBLER_NAME.
6040 (lhd_overwrite_decl_assembler_name): Default implementation.
6042 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6043 Jackson Woodruff <jackson.woodruff@arm.com>
6045 PR tree-optimization/71026
6046 * match.pd: Canonicalize constant multiplies in division.
6048 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6050 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6051 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6052 Remove OPT_fomit_frame_pointer entry.
6053 * common/config/arc/arc-common.c: Likewise.
6054 * common/config/arm/arm-common.c: Likewise.
6055 * common/config/avr/avr-common.c: Likewise.
6056 * common/config/c6x/c6x-common.c: Likewise.
6057 * common/config/cr16/cr16-common.c: Likewise.
6058 * common/config/cris/cris-common.c: Likewise.
6059 * common/config/epiphany/epiphany-common.c: Likewise.
6060 * common/config/fr30/fr30-common.c: Likewise.
6061 * common/config/frv/frv-common.c: Likewise.
6062 * common/config/ia64/ia64-common.c: Likewise.
6063 * common/config/iq2000/iq2000-common.c: Likewise.
6064 * common/config/lm32/lm32-common.c: Likewise.
6065 * common/config/m32r/m32r-common.c: Likewise.
6066 * common/config/mcore/mcore-common.c: Likewise.
6067 * common/config/microblaze/microblaze-common.c: Likewise.
6068 * common/config/mips/mips-common.c: Likewise.
6069 * common/config/mmix/mmix-common.c: Likewise.
6070 * common/config/mn10300/mn10300-common.c: Likewise.
6071 * common/config/nios2/nios2-common.c: Likewise.
6072 * common/config/pa/pa-common.c: Likewise.
6073 * common/config/pdp11/pdp11-common.c: Likewise.
6074 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6075 * common/config/riscv/riscv-common.c: Likewise.
6076 * common/config/rs6000/rs6000-common.c: Likewise.
6077 * common/config/rx/rx-common.c: Likewise.
6078 * common/config/s390/s390-common.c: Likewise.
6079 * common/config/sh/sh-common.c: Likewise.
6080 * common/config/sparc/sparc-common.c: Likewise.
6081 * common/config/tilegx/tilegx-common.c: Likewise.
6082 * common/config/tilepro/tilepro-common.c: Likewise.
6083 * common/config/v850/v850-common.c: Likewise.
6084 * common/config/visium/visium-common.c: Likewise.
6085 * common/config/xstormy16/xstormy16-common.c: Likewise.
6086 * common/config/xtensa/xtensa-common.c: Likewise.
6087 * invoke.texi (-fomit-frame-pointer): Update documentation.
6089 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6091 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6093 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6095 * cfg.c (scale_bbs_frequencies_int,
6096 cale_bbs_frequencies_gcov_type): Remove.
6097 * cfg.h (scale_bbs_frequencies_int,
6098 cale_bbs_frequencies_gcov_type): Remove.
6100 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6102 * tree-ssa-loop-manip.c
6103 (scale_dominated_blocks_in_loop): Update to profile counts.
6104 (tree_transform_and_unroll_loop): Likewise.
6106 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6108 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6109 scale_bbs_frequencies_int.
6111 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6113 * final.c (compute_alignments): Use counts rather than frequencies.
6115 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6117 * cfgloopanal.c: Include sreal.h
6118 (average_num_loop_insns): Use counts and sreal for accounting.
6120 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6122 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6125 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6126 Kito Cheng <kito.cheng@gmail.com>
6128 * config/nds32/constraints.md: Provide more constraints.
6129 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6130 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6131 support constraints usage.
6133 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6135 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6137 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6138 Kito Cheng <kito.cheng@gmail.com>
6140 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6141 * config/nds32/nds32.opt: Refine the layout.
6142 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6143 TARGET_EXT_STRING): Support new options.
6144 * config/nds32/nds32.h: Likewise.
6145 * config/nds32/nds32.md: Likewise.
6146 * config/nds32/nds32-predicates.c: Likewise.
6147 * config/nds32/constraints.md: Likewise.
6148 * common/config/nds32/nds32-common.c: Likewise.
6150 2017-11-16 Julia Koval <julia.koval@intel.com>
6153 * config/i386/gfniintrin.h: Add sse check.
6154 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6156 2017-11-16 Julia Koval <julia.koval@intel.com>
6158 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6159 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6160 (ix86_handle_option): Handle -mavx512vbmi2.
6161 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6162 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6163 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6164 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6165 (ix86_valid_target_attribute_inner_p): Ditto.
6166 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6167 * config/i386/i386.opt (mavx512vbmi2): New option.
6168 * doc/invoke.texi: Add new option.
6170 2017-11-16 Julia Koval <julia.koval@intel.com>
6172 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6173 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6174 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6175 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6176 _mm512_gf2p8mul_epi8): New intrinsics.
6177 * config/i386/i386-builtin-types.def
6178 (V64QI_FTYPE_V64QI_V64QI): New type.
6179 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6180 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6181 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6182 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6183 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6184 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6186 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6188 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6189 explicitly as a stream of bytes.
6191 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6193 * config/rs6000/altivec.h (vec_xst_be): New #define.
6194 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6195 and externalize from *altivec_vperm_<mode>_internal.
6196 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6198 (XL_BE_V8HI): Likewise.
6199 (XL_BE_V4SI): Likewise.
6200 (XL_BE_V4SI): Likewise.
6201 (XL_BE_V2DI): Likewise.
6202 (XL_BE_V4SF): Likewise.
6203 (XL_BE_V2DF): Likewise.
6204 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6205 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6206 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6207 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6208 VSX_BUILTIN_VEC_XST_BE.
6209 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6210 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6212 (altivec_init_builtins): Replace conditional calls to def_builtin
6213 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6214 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6215 with unconditional calls. Remove calls to def_builtin for
6216 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6217 __builtin_vec_xst_be.
6218 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6219 to define_expand, and add alternate RTL generation for P8.
6220 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6221 vsx_ld_elemrev_v8hi.
6222 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6223 add alternate RTL generation for P8.
6224 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6225 vsx_ld_elemrev_v16qi.
6226 (vsx_st_elemrev_v8hi): Convert define_insn
6227 to define_expand, and add alternate RTL generation for P8.
6228 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6229 vsx_st_elemrev_v8hi.
6230 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6231 add alternate RTL generation for P8.
6232 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6233 vsx_st_elemrev_v16qi.
6235 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6238 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6239 TARGET_AVX512ER check.
6240 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6241 neither -mzeroupper nor -mno-zeroupper is used and
6242 TARGET_EMIT_VZEROUPPER is set.
6243 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6244 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6246 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6248 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6249 folding of vector compares.
6250 (fold_build_vec_cmp): New helper function.
6251 (fold_compare_helper): New helper function.
6252 (builtin_function_type): Add compare builtins to the list of functions
6253 having unsigned arguments. Cosmetic updates to comment indentation.
6254 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6255 the not+eq combination.
6257 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6259 PR tree-optimization/82726
6260 PR tree-optimization/70754
6261 * tree-predcom.c (order_drefs_by_pos): New function.
6262 (combine_chains): Move code setting has_max_use_after to...
6263 (try_combine_chains): ...here. New parameter. Sort combined chains
6264 according to position information.
6265 (tree_predictive_commoning_loop): Update call to above function.
6266 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6268 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6270 PR tree-optimization/82726
6272 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6274 PR tree-optimization/70754
6275 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6276 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6277 combined stmt before it if not NULL.
6278 (combine_chains): Process refs reversely and compute dominance point
6282 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6284 PR tree-optimization/79663
6285 * tree-predcom.c (combine_chains): Process refs in reverse order
6286 only for ZERO length chains, and add explaining comment.
6288 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6290 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6291 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6292 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6294 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6296 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6298 2017-11-15 Martin Liska <mliska@suse.cz>
6300 * tree-cfg.c (pass_warn_function_return::execute):
6301 Compare warn_return_type for greater than zero.
6303 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6307 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6308 to return true on Xeon and not on Xeon Phi.
6309 (ix86_check_avx256_register): Changed to ...
6310 (ix86_check_avx_upper_register): ... this. Add extra check for
6311 VALID_AVX512F_REG_OR_XI_MODE.
6312 (ix86_avx_u128_mode_needed): Changed
6313 ix86_check_avx256_register to ix86_check_avx_upper_register.
6314 (ix86_check_avx256_stores): Changed to ...
6315 (ix86_check_avx_upper_stores): ... this. Changed
6316 ix86_check_avx256_register to ix86_check_avx_upper_register.
6317 (ix86_avx_u128_mode_after): Changed
6318 avx_reg256_found to avx_upper_reg_found. Changed
6319 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6320 (ix86_avx_u128_mode_entry): Changed
6321 ix86_check_avx256_register to ix86_check_avx_upper_register.
6322 (ix86_avx_u128_mode_exit): Ditto.
6323 * config/i386/i386.h: (host_detect_local_cpu): New define.
6325 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6327 * config/arm/xgene1.md (xgene1): Split into automatons
6328 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6329 (xgene1_f_load): Adjust reservations and/or types.
6330 (xgene1_f_store): Likewise.
6331 (xgene1_load_pair): Likewise.
6332 (xgene1_store_pair): Likewise.
6333 (xgene1_fp_load1): Likewise.
6334 (xgene1_load1): Likewise.
6335 (xgene1_store1): Likewise.
6336 (xgene1_move): Likewise.
6337 (xgene1_alu): Likewise.
6338 (xgene1_simd): Likewise.
6339 (xgene1_bfm): Likewise.
6340 (xgene1_neon_load1): Likewise.
6341 (xgene1_neon_store1): Likewise.
6342 (xgene1_neon_logic): Likewise.
6343 (xgene1_neon_st1): Likewise.
6344 (xgene1_neon_ld1r): Likewise.
6345 (xgene1_alu_cond): Added.
6346 (xgene1_shift_reg): Likwise.
6347 (xgene1_bfx): Likewise.
6348 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6350 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6353 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6355 (can_widen_mult_without_libcall): New function.
6356 (expand_mul_overflow): If only checking unsigned mul overflow,
6357 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6358 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6359 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6360 (expand_DIVMOD): Formatting fix.
6361 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6362 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6363 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6365 PR tree-optimization/82977
6366 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6367 constructed temporary to strlen_to_stridx.put.
6369 2017-11-15 Martin Liska <mliska@suse.cz>
6371 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6372 * configure: Regenerate.
6374 2017-11-15 Martin Liska <mliska@suse.cz>
6377 * config/sh/sh-mem.cc: Use proper probability for
6380 2017-11-14 Jeff Law <law@redhat.com>
6382 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6383 the red zone for stack_clash_protection_final_dynamic_probe targets
6384 when the total dynamic stack size is zero bytes.
6386 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6387 blocks is post order.
6389 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6391 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6392 * final.c (rest_of_clean_state): Set it for the
6393 -fcompare-debug dump.
6394 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6395 class when TDF_COMPARE_DEBUG is set.
6397 * dwarf2out.c (gen_producer_string): Discard
6400 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6403 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6404 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6405 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6406 (__TGMATH_REAL_2_3): Remove macros.
6407 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6408 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6410 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6412 (remquo): Define using __TGMATH_REAL_3.
6414 2017-11-14 Jeff Law <law@redhat.com>
6416 * vr-values.c: New file with contents extracted from tree-vrp.c.
6417 * Makefile.in (OBJS): Add vr-values.o
6418 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6419 (set_value_range, set_and_canonicalize_value_range): Likewise.
6420 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6421 (value_range_constant_singleton, symbolic_range_p): Likewise.
6422 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6423 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6424 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6425 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6426 (find_case_label_range, find_case_label_index): Likewise.
6427 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6428 (range_int_cst_singleton_p, value_inside_range): Likewise.
6429 (get_single_symbol): Likewise.
6430 (switch_update): Move structure definition here.
6431 (to_remove_edges, to_update_switch_stmts): Provide externs.
6432 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6433 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6434 (vrp_val_is_min, set_value_range): Likewise.
6435 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6436 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6437 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6438 (range_is_nonnull, range_int_cst_p): Likewwise.
6439 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6440 (get_single_symbol, operand_less_p): Likewise
6441 (compare_values_warnv, compare_values): Likewise.
6442 (value_inside_range, value_range_constant_singleton): Likewise.
6443 (zero_nonzero_bitgs_from_vr): Likewise.
6444 (extract_range_from_binary_expr_1): Likewise.
6445 (overflow_comparison_p): Likewise.
6446 (to_remove_edges, to_update_switch_stmts): Likewise.
6447 (find_case_label-index, find_case_label_range): Likewise.
6448 (switch_update, set_value_range_to_nonnegative): Remove.
6449 (set_value_range_to_truthvalue): Likewise.
6450 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6451 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6452 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6453 (find_case_label_ranges, test_for_singularity): Likewise.
6454 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6455 (x_vr_values): Move to its remaining use site.
6457 2017-11-10 Jeff Law <law@redhat.com>
6459 * vr-values.h (VR_INITIALIZER): Move #define here.
6460 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6461 * Makefile.in (OBJS): Add tree-evrp.o
6462 * tree-vrp.h (assert_info): Move structure definition here.
6463 (set_value_range_to_varying): Prototype.
6464 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6465 (infer_value_range, register_edge_assert_for): Likewise.
6466 (stmt_interesting_for_vrp): Likewise.
6467 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6468 (set_value_range_to_varying): No longer static.
6469 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6470 (infer_value_range, register_edge_assert_for): Likewise.
6472 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6474 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6475 generate the XXBRD instruction.
6477 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6478 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6479 as being compatible if -mabi=ieeelongdouble.
6480 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6481 to setup float128 built-ins with hardware support.
6482 (BU_FLOAT128_HW_2): Likewise.
6483 (BU_FLOAT128_HW_3): Likewise.
6484 (BU_FLOAT128_HW_VSX_1): Likewise.
6485 (BU_FLOAT128_HW_VSX_2): Likewise.
6486 (scalar_extract_expq): Change float128 built-in functions to
6487 accommodate having both KFmode and TFmode functions. Use the
6488 KFmode variant as the default.
6489 (scalar_extract_sigq): Likewise.
6490 (scalar_test_neg_qp): Likewise.
6491 (scalar_insert_exp_q): Likewise.
6492 (scalar_insert_exp_qp): Likewise.
6493 (scalar_test_data_class_qp): Likewise.
6494 (sqrtf128_round_to_odd): Delete processing the round to odd
6495 built-in functions as special built-in functions, and define them
6496 as float128 built-ins. Use the KFmode variant as the default.
6497 (truncf128_round_to_odd): Likewise.
6498 (addf128_round_to_odd): Likewise.
6499 (subf128_round_to_odd): Likewise.
6500 (mulf128_round_to_odd): Likewise.
6501 (divf128_round_to_odd): Likewise.
6502 (fmaf128_round_to_odd): Likewise.
6503 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6504 support for KFmode and TFmode xststdcqp calls.
6505 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6506 point, switch the built-in handlers for the get/set float128
6507 exponent, get float128 mantissa, float128 test built-ins, and the
6508 float128 round to odd built-in functions. Eliminate creating the
6509 float128 round to odd built-in functions as special built-ins.
6510 (rs6000_init_builtins): Eliminate special creation of the float128
6511 round to odd built-in functions.
6512 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6513 function insns to support both TFmode and KFmode variants.
6514 (xsxsigqp_<mode>): Likewise.
6515 (xsiexpqpf_<mode>): Likewise.
6516 (xsiexpqp_<mode>): Likewise.
6517 (xststdcqp_<mode>): Likewise.
6518 (xststdcnegqp_<mode>): Likewise.
6519 (xststdcqp_<mode>): Likewise.
6521 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6523 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6524 unused path_in_freq_ptr parameter.
6525 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6527 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6529 * ipa-inline.c (edge_badness): Dump sreal frequency.
6530 (compute_inlined_call_time): Match natural implementaiton ...
6531 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6532 forgotten division by CGRAPH_FREQ_BASE.
6534 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6536 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6537 Solaris 11. Update comment.
6538 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6540 * config/sol2.h (STARTFILE_SPEC): Likewise.
6541 * configure: Regenerate.
6543 2017-11-14 Carl Love <cel@us.ibm.com>
6545 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6546 le_ and be_ prefixes to swap* variables. Remove
6547 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6550 2017-11-14 Jason Merrill <jason@redhat.com>
6552 Support GTY((cache)) on hash_map.
6553 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6554 (ggc_cache_remove): Override it instead of ggc_mx.
6555 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6556 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6557 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6558 (simple_cache_map_traits): Override maybe_mx.
6559 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6560 (hash_map): Friend gt_cleare_cache.
6561 (gt_cleare_cache): New.
6562 * tree.h (tree_cache_traits): New hash_map traits class.
6563 (tree_cache_map): New typedef.
6565 2017-11-14 Richard Biener <rguenther@suse.de>
6567 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6568 paramter and handling.
6569 (cleanup_control_flow_bb): Likewise.
6570 (cleanup_control_flow_pre): New helper performing a DFS walk
6571 to call cleanup_control_flow_bb in PRE order.
6572 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6573 via cleanup_control_flow_pre.
6575 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6577 * config/aarch64/aarch64-simd.md
6578 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6579 (*aarch64_simd_bsl<mode>_alt): Likewise.
6580 (aarch64_simd_bsldi_internal): New.
6581 (aarch64_simd_bsldi_alt): Likewise.
6583 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6585 * tracer.c (better_p): Do not compare frequencies.
6586 * reg-stack.c (better_edge): Likewise.
6587 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6590 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6592 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6593 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6594 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6595 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6596 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6597 * predict.c (maybe_hot_frequency_p): Inline to ...
6598 (maybe_hot_count_p): ... here; rewrite to counts.
6599 (counts_to_freqs): Rename to ...
6600 (update_max_bb_count): ... this one.
6601 (expensive_function_p): Use counts.
6602 (estimate_bb_frequencies): Update.
6603 (rebuild_frequencies): Update.
6604 * predict.h (counts_to_freqs): Rename to ...
6605 (update_max_bb_count): ... this one.
6606 * profile.c (compute_branch_probabilities): Add debug info
6607 * tree-inline.c (expand_call_inline): Update debug info.
6608 (optimize_inline_calls): Use update_max_bb_count..
6609 (tree_function_versioning): Use update_max_bb_count..
6610 * value-prof.c (gimple_value_profile_transformations):
6611 Do not use update_max_bb_count.
6613 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6615 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6616 always use frequencies.
6618 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6620 * bb-reorder.c: Remove frequencies from comments.
6621 (better_edge_p): Use profile counts.
6622 (find_traces): Dump profile counts.
6623 (rotate_loop): Use profile counts.
6624 (find_traces_1_round): Likewise.
6625 (connect_better_edge_p): Use counts instead of probabilities for
6627 (copy_bb_p): Drop early check for non-0 frequency.
6628 (sanitize_hot_paths): Update comments.
6630 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6632 * ipa-split.c (struct split_point): Add count.
6633 (consider_split): Do not compute incoming frequency; compute incoming
6634 count and store it to split_point.
6635 (split_function): Set count of the call to split part correctly.
6637 2017-11-13 Carl Love <cel@us.ibm.com>
6639 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6641 2017-11-13 Tom Tromey <tom@tromey.com>
6643 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6645 2017-11-13 Carl Love <cel@us.ibm.com>
6647 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6648 Add support for builtins:
6649 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6650 vector {un,}signed {char,int,short},
6651 vector {un,}signed {char,int,short}) arguments.
6652 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6653 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6654 Add BU_P9V_AV_2 expansions for the builtins.
6655 * config/rs6000/altivec.h (vec_first_match_index,
6656 vec_first_mismatch_index, vec_first_match_or_eos_index,
6657 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6658 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6659 new extern declaration.
6660 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6661 * config/rs6000/vsx.md (first_match_index_<mode>,
6662 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6663 first_mismatch_or_eos_index_<mode>): Add define expand.
6664 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6665 * doc/extend.texi: Update the built-in documenation file for the new
6668 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6670 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6671 into the min/max operations for _Float<N> and _Float<N>X types.
6673 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6676 * dwarf2out.c (do_eh_frame): New static variable.
6677 (dwarf2out_begin_prologue): Set it.
6678 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6680 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6682 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6684 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6686 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6689 * debug.h (dwarf2out_do_eh_frame): Declare.
6690 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6691 (dwarf2out_do_frame): Use it.
6692 (dwarf2out_do_cfi_asm): Likewise.
6693 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6694 (dwarf2out_assembly_start): Likewise.
6695 (dwarf2out_begin_prologue): Fix comment.
6696 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6697 if the target needs either debug or unwind DWARF2 info.
6698 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6699 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6700 -ftrapping-math, -ftrapv and -fwrapv.
6702 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6704 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6705 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6706 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6707 (estimate_edge_size_and_time): Likewise.
6708 (ipa_merge_fn_summary_after_inlining): Likewise.
6709 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6710 (compute_uninlined_call_time): Use sreal_frequency.
6711 (compute_inlined_call_time): Likewise.
6712 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6713 * profile-count.c: Include sreal.h.
6714 (profile_count::to_sreal_scale): New.
6715 * profile-count.h: Forward declare sreal.
6716 (profile_count::to_sreal_scale): Declare.
6718 2017-11-13 Nathan Sidwell <nathan@acm.org>
6720 * diagnostic.c (maybe_line_and_column): New.
6721 (diagnostic_get_location_text): Use it.
6722 (diagnostic_report_current_module): Likewise.
6723 (test_diagnostic_get_location_text): Add tests.
6725 2017-11-13 Luis Machado <luis.machado@linaro.org>
6727 * doc/md.texi (Specifying processor pipeline description): Fix
6728 incorrect latency for the div instruction example.
6730 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6732 PR tree-optimization/78821
6733 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6734 that bit_not_p is the same.
6735 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6736 (split_group): Count precisely bit_not_p bits in each statement.
6737 (invert_op): New function.
6738 (imm_store_chain_info::output_merged_store): Use invert_op to
6739 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6740 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6742 2017-11-13 Martin Liska <mliska@suse.cz>
6744 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6745 (struct source_info): Likewise.
6746 (add_branch_counts): Likewise.
6747 (add_line_counts): Likewise.
6748 (function_summary): Likewise.
6749 (output_intermediate_line): Likewise.
6750 (generate_results): Likewise.
6752 2017-11-13 Martin Liska <mliska@suse.cz>
6754 * gcov.c (struct block_info): Remove typedef for block_t.
6755 (struct line_info): Likewise.
6756 (line_info::has_block): Likewise.
6757 (EXIT_BLOCK): Likewise.
6758 (unblock): Likewise.
6759 (circuit): Likewise.
6760 (get_cycles_count): Likewise.
6761 (process_file): Likewise.
6762 (read_graph_file): Likewise.
6763 (solve_flow_graph): Likewise.
6764 (find_exception_blocks): Likewise.
6765 (add_line_counts): Likewise.
6766 (accumulate_line_info): Likewise.
6767 (output_line_details): Likewise.
6769 2017-11-13 Martin Liska <mliska@suse.cz>
6771 * gcov.c (struct arc_info): Remove typedef for arc_t.
6772 (struct line_info): Likewise.
6773 (add_branch_counts): Likewise.
6774 (output_branch_count): Likewise.
6775 (function_info::~function_info): Likewise.
6776 (circuit): Likewise.
6777 (output_intermediate_line): Likewise.
6778 (read_graph_file): Likewise.
6779 (solve_flow_graph): Likewise.
6780 (find_exception_blocks): Likewise.
6781 (add_line_counts): Likewise.
6782 (accumulate_line_info): Likewise.
6783 (output_line_details): Likewise.
6784 (output_function_details): Likewise.
6786 2017-11-13 Martin Liska <mliska@suse.cz>
6788 * gcov.c (struct function_info): Remove typedef for function_t.
6789 (struct source_info): Likewise.
6790 (source_info::get_functions_at_location): Likewise.
6791 (solve_flow_graph): Likewise.
6792 (find_exception_blocks): Likewise.
6793 (add_line_counts): Likewise.
6794 (output_intermediate_file): Likewise.
6795 (process_file): Likewise.
6796 (generate_results): Likewise.
6797 (release_structures): Likewise.
6798 (read_graph_file): Likewise.
6799 (read_count_file): Likewise.
6800 (accumulate_line_counts): Likewise.
6801 (output_lines): Likewise.
6803 2017-11-13 Martin Liska <mliska@suse.cz>
6805 * gcov.c (function_info::function_info): Remove num_counts
6806 and add vector<gcov_type>.
6807 (function_info::~function_info): Use the vector.
6808 (process_file): Likewise.
6809 (read_graph_file): Likewise.
6810 (read_count_file): Likewise.
6811 (solve_flow_graph): Likewise.
6813 2017-11-13 Martin Liska <mliska@suse.cz>
6815 * gcov.c (function_info::is_artificial): New function.
6816 (process_file): Erase all artificial early.
6817 (generate_results): Skip as all artificial are already
6820 2017-11-13 Martin Liska <mliska@suse.cz>
6822 * gcov.c (read_graph_file): Store to global vector of functions.
6823 (read_count_file): Iterate the vector.
6824 (process_file): Likewise.
6825 (generate_results): Likewise.
6826 (release_structures): Likewise.
6828 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6830 PR tree-optimization/82954
6831 * gimple-ssa-store-merging.c
6832 (imm_store_chain_info::coalesce_immediate_stores): If
6833 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6835 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6837 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6838 Upddate call to ENDIAN_LANE_N.
6839 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6840 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6841 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6842 and use aarch64_endian_lane_rtx.
6843 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6845 2017-11-12 Tom de Vries <tom@codesourcery.com>
6847 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6849 2017-11-12 Tom de Vries <tom@codesourcery.com>
6851 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6853 2017-11-12 Tom de Vries <tom@codesourcery.com>
6855 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6857 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6858 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6859 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6861 2017-11-11 Martin Sebor <msebor@redhat.com>
6864 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6867 * prefic.c (translate_name): Replace strncpy with memcpy to
6868 avoid -Wstringop-truncation.
6870 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6872 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6874 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6876 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6878 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6880 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6883 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6885 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6886 enable generating XXBRH if the value is in a vector register.
6887 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6888 value is in a vector register.
6889 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6890 register to register bswap64's instead of doing the GPR sequence
6891 used on previous machines.
6892 (bswapdi2_xxbrd): New insn.
6893 (bswapdi2_reg): Disallow on ISA 3.0.
6894 (register to register bswap64 splitter): Do not split the insn on
6895 ISA 3.0 systems that use XXBRD.
6897 2017-11-10 Martin Sebor <msebor@redhat.com>
6900 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6902 (find_subframework_file): Same.
6904 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6906 * auto-profile.c (afdo_indirect_call): Drop frequency.
6907 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6908 (cgraph_node::create_edge): Drop frequency argument.
6909 (cgraph_node::create_indirect_edge): Drop frequency argument.
6910 (cgraph_edge::make_speculative): Drop frequency arguments.
6911 (cgraph_edge::resolve_speculation): Do not update frequencies
6912 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6913 (cgraph_node::dump): Check consistency in IPA mode.
6914 (cgraph_edge::maybe_hot_p): Use IPA counter.
6915 (cgraph_edge::verify_count_and_frequency): Rename to ...
6916 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6917 (cgraph_node::verify_node): Update.
6918 * cgraph.h (struct cgraph_edge): Drop frequency.
6919 (cgraph_edge::frequency): New function.
6920 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6922 (cgraph_edge::rebuild_edges): Likewise.
6923 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6924 (duplicate_thunk_for_node): Do not pass frequency.
6925 (cgraph_node::create_clone): Scale only counts.
6926 (cgraph_node::create_virtual_clone): Do not pass frequency.
6927 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6928 (cgraph_node::create_version_clone): Do not pass frequency.
6929 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6930 (cgraph_node::expand_thunk): Do not pass frequency.
6931 (cgraph_node::create_wrapper): Do not pass frequency.
6932 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6934 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6935 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6936 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6937 (gather_caller_stats): Use frequency function.
6938 (ipcp_cloning_candidate_p): Use frequency function.
6939 (ipcp_propagate_stage): Use frequency function.
6940 (get_info_about_necessary_edges): Use frequency function.
6941 (update_profiling_info): Update only IPA profile.
6942 (update_specialized_profile): Use frequency functoin.
6943 (perhaps_add_new_callers): Update only IPA profile.
6944 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6945 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6946 (dump_ipa_call_summary): Use frequency function.
6947 (estimate_edge_size_and_time): Use frequency function.
6948 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6949 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6950 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6951 (update_noncloned_counts): ... ths one; scale counts only.
6952 (clone_inlined_nodes): Do not scale frequency.
6953 (inline_call): Do not pass frequency.
6954 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6955 (compute_inlined_call_time): Use IPA profile.
6956 (want_inline_small_function_p): Use IPA profile.
6957 (want_inline_self_recursive_call_p): Use IPA profile.
6958 (edge_badness): Use IPA profile.
6959 (lookup_recursive_calls): Use IPA profile.
6960 (recursive_inlining): Do not pass frequency.
6961 (resolve_noninline_speculation): Do not update frequency.
6962 (inline_small_functions): Collect max of IPA profile.
6963 (dump_overall_stats): Dump IPA porfile.
6964 (dump_inline_stats): Dump IPA porfile.
6965 (ipa_inline): Collect IPA stats.
6966 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6967 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6968 (ipa_propagate_frequency): Use frequency function.
6969 (ipa_profile): Cleanup.
6970 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6971 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6972 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6973 (input_node): Do not stream frequency.
6974 (input_edge): Do not stream frequency.
6975 (merge_profile_summaries): Scale only IPA profiles.
6976 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6977 * predict.c (drop_profile): Do not recompute frequency.
6978 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6979 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6980 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6981 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6982 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6983 * tree-inline.c (copy_bb): Do not scale frequency.
6984 (expand_call_inline): Do not scale frequency.
6985 (tree_function_versioning): Do not scale frequency.
6986 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6988 2017-11-10 Julia Koval <julia.koval@intel.com>
6990 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6991 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6992 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6993 (_mm256_maskz_gf2p8affine_epi64_epi8)
6994 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6995 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6996 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6997 (__builtin_ia32_vgf2p8affineqb_v32qi)
6998 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6999 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7001 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7005 (arm_option_override): Refactor.
7006 (arm_option_reconfigure_globals): New.
7007 (arm_options_perform_arch_sanity_checks): New.
7008 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7010 (arm_options_perform_arch_sanity_checks): Likewise
7012 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7014 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7015 (power9-qpmul): New.
7016 * rs6000/rs6000.md ("type" attr): Add qmul.
7017 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7018 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7019 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7021 2017-11-10 Martin Sebor <msebor@redhat.com>
7024 * builtins.c (compute_objsize): Handle arrays that
7025 compute_builtin_object_size likes to fail for. Make extern.
7026 * builtins.h (compute_objsize): Declare.
7027 (check_strncpy_sizes): New function.
7028 (expand_builtin_strncpy): Call check_strncpy_sizes.
7029 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7030 -Wstringop-truncation.
7031 (gimple_fold_builtin_strncat): Same.
7032 * gimple.c (gimple_build_call_from_tree): Set call location.
7033 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7034 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7035 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7036 (handle_builtin_strlen): Use strlen_to_stridx.
7037 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7039 Use strlen_to_stridx.
7040 (pass_strlen::execute): Release strlen_to_stridx.
7041 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7042 (-Wstringop-truncation): Document new option.
7044 2017-11-10 Martin Liska <mliska@suse.cz>
7046 PR gcov-profile/82702
7047 * gcov.c (main): Handle intermediate files in a different
7049 (get_gcov_intermediate_filename): New function.
7050 (output_gcov_file): Remove support of intermediate files.
7051 (generate_results): Allocate intermediate file.
7052 (release_structures): Clean-up properly fn_end.
7053 (output_intermediate_file): Start iterating with line 1.
7055 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7057 PR tree-optimization/82929
7058 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7059 ops_swapped_p non-static data member.
7060 (store_immediate_info::store_immediate_info): Clear it.
7061 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7062 ops set ops_swapped_p.
7063 (count_multiple_uses): Handle ops_swapped_p.
7065 2017-11-10 Martin Liska <mliska@suse.cz>
7067 * coverage.c (coverage_init): Stream information about
7068 support of has_unexecuted_blocks.
7069 * doc/gcov.texi: Document that.
7070 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7071 * gcov.c (read_graph_file): Likewise.
7072 (output_line_beginning): Fix a small issue with
7075 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7077 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7078 reference of trivial component.
7080 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7083 * gimple-ssa-store-merging.c
7084 (pass_store_merging::terminate_all_aliasing_chains): For
7085 gimple_store_p stmts also call refs_output_dependent_p.
7087 PR rtl-optimization/82913
7088 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7091 2017-11-09 Jeff Law <law@redhat.com>
7093 * vr-values.h: New file with vr_values class.
7094 * tree-vrp.c: Include vr-values.h
7095 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7096 data objects into the vr_values class.
7097 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7098 (get_value_range): Make it a member function within vr_values class.
7099 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7100 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7101 (op_with_constant_singleton_value_range): Likewise.
7102 (extract_range_for_var_from_comparison_expr): Likewise.
7103 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7104 (extract_range_from_binary_expr): Likewise.
7105 (extract_range_from_unary_expr): Likewise.
7106 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7107 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7108 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7109 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7110 (compare_name_with_value, compare_names): Likewise.
7111 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7112 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7113 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7114 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7115 (extract_range_from_phi_node): Likewise.
7116 (simplify_truth_ops_using_ranges): Likewise.
7117 (simplify_div_or_mod_using_ranges): Likewise.
7118 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7119 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7120 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7121 (simplify_float_conversion_using_ranges): Likewise.
7122 (simplify_internal_call_using_ranges): Likewise.
7123 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7124 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7125 poitner via x_vr_values for calls into gimple folder.
7126 (vrp_initialize_lattice): Make this the vr_values ctor.
7127 (vrp_free_lattice): Make this the vr_values dtor.
7128 (set_vr_value): New function.
7129 (class vrp_prop): Add vr_values data member. Add various member
7130 functions as well as member functions that delegate to vr_values.
7131 (check_array_ref): Make a member function within vrp_prop class.
7132 (search_for_addr_array, vrp_initialize): Likewise.
7133 (vrp_finalize): Likewise. Revamp to avoid direct access to
7134 vr_value, values_propagated, etc.
7135 (check_array_bounds): Extract vrp_prop class instance pointer from
7136 walk info structure. Use it to call member functions.
7137 (check_all_array_refs): Make a member function within vrp_prop class.
7138 Smuggle class instance pointer via walk info structure.
7139 (x_vr_values): New local static.
7140 (vrp_valueize): Use x_vr_values to get class instance.
7141 (vr_valueize_1): Likewise.
7142 (class vrp_folder): Add vr_values data member. Add various member
7143 functions as well as member functions that delegate to vr_values.
7144 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7145 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7146 class instance from vr_values. Use it to call member functions.
7147 (vrp_dom_walker): Add vr_values data member.
7148 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7149 instance via x_vr_values.
7150 (identify_jump_threads): Accept vr_values as argument. Store
7151 it into the walker structure.
7152 (evrp_dom_walker): Add vr_values class data member. Add various
7154 (evrp_dom_walker::try_find_new_range): Use vr_values data
7155 member to access the memory allocator.
7156 (evrp_dom_walker::before_dom_children): Store vr_values class
7157 instance into the vrp_folder class.
7158 (evrp_dom_walker::push_value_range): Rework to avoid direct
7159 access to num_vr_values and vr_value.
7160 (evrp_dom_walker::pop_value_range): Likewise.
7161 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7162 Use vr_values to get to dump_all_value_ranges member function.
7163 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7164 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7165 Pass vr_values class instance down to identify_jump_threads.
7166 Remove call to vrp_free_lattice.
7167 (debug_all_value_ranges): Remove.
7169 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7170 (vrp_folder): Likewise.
7172 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7173 Get it from the existing bitmap instead.
7174 (vrp_intersect_ranges_1): Likewise.
7176 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7178 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7180 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7181 set bit_not_p to it.
7182 (imm_store_chain_info::coalesce_immediate_stores): Break group
7183 if bit_not_p is different.
7184 (count_multiple_uses, split_group,
7185 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7186 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7187 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7188 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7189 to store_immediate_info ctor.
7191 2017-11-09 Jim Wilson <jimw@sifive.com>
7193 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7194 (scan_prog_file): Likewise.
7196 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7198 * bb-reorder.c (max_entry_frequency): Remove.
7199 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7200 connect_traces, push_to_next_round_p): Remove prototypes.
7201 (find_traces_1_round): Use counts only.
7202 (push_to_next_round_p): Likewise.
7203 (find_traces): Likewise.
7204 (rotate_loop): Likewise.
7205 (find_traces_1_round): Likewise.
7206 (connect_traces): Likewise.
7207 (edge_order): Likewise.
7209 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7211 * config/arm/arm.c (output_return_instruction): Add comments to
7212 indicate requirement for cmse_nonsecure_entry return to account
7213 for the size of clearing instruction output here.
7214 (thumb_exit): Likewise.
7215 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7216 return in hardfloat mode.
7218 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7220 * config/rs6000/rs6000.c (machine_function): Add a bool,
7221 "toc_is_wrapped_separately".
7222 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7223 if it wasn't explicitly set or unset, we are optimizing for speed, and
7224 doing separate shrink-wrapping.
7225 (rs6000_get_separate_components): Enable the TOC component if
7226 saving the TOC register in the prologue.
7227 (rs6000_components_for_bb): Handle the TOC component.
7228 (rs6000_emit_prologue_components): Store the TOC register where needed.
7229 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7230 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7232 2017-11-09 Martin Jambor <mjambor@suse.cz>
7234 * ipa-param-manipulation.c: New file.
7235 * ipa-param-manipulation.h: Likewise.
7236 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7237 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7238 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7239 (ipa_parm_adjustment): Likewise.
7240 (ipa_parm_adjustment_vec): Likewise.
7241 (ipa_get_vector_of_formal_parms): Moved declaration to
7242 ipa-param-manipulation.h.
7243 (ipa_get_vector_of_formal_parm_types): Likewise.
7244 (ipa_modify_formal_parameters): Likewise.
7245 (ipa_modify_call_arguments): Likewise.
7246 (ipa_combine_adjustments): Likewise.
7247 (ipa_dump_param_adjustments): Likewise.
7248 (ipa_modify_expr): Likewise.
7249 (ipa_get_adjustment_candidate): Likewise.
7250 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7251 ipa-param-manipulation.c.
7252 (ipa_get_vector_of_formal_parm_types): Likewise.
7253 (ipa_modify_formal_parameters): Likewise.
7254 (ipa_modify_call_arguments): Likewise.
7255 (ipa_modify_expr): Likewise.
7256 (get_ssa_base_param): Likewise.
7257 (ipa_get_adjustment_candidate): Likewise.
7258 (index_in_adjustments_multiple_times_p): Likewise.
7259 (ipa_combine_adjustments): Likewise.
7260 (ipa_dump_param_adjustments): Likewise.
7261 * tree-sra.c: Also include ipa-param-manipulation.h
7262 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7265 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7266 Alan Hayward <alan.hayward@arm.com>
7267 David Sherwood <david.sherwood@arm.com>
7269 * doc/sourcebuild.texi (vect_masked_store): Document.
7271 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7272 Alan Hayward <alan.hayward@arm.com>
7273 David Sherwood <david.sherwood@arm.com>
7275 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7277 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7278 Alan Hayward <alan.hayward@arm.com>
7279 David Sherwood <david.sherwood@arm.com>
7281 * doc/sourcebuild.texi (vect_variable_length): Document.
7283 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7284 Alan Hayward <alan.hayward@arm.com>
7285 David Sherwood <david.sherwood@arm.com>
7287 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7289 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7290 Alan Hayward <alan.hayward@arm.com>
7291 David Sherwood <david.sherwood@arm.com>
7293 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7295 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7296 Alan Hayward <alan.hayward@arm.com>
7297 David Sherwood <david.sherwood@arm.com>
7299 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7300 previously undocumented selectors.
7301 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7303 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7305 * doc/rtl.texi (const_vector): Say that elements can be
7306 const_wide_ints too.
7307 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7308 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7309 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7310 * optabs.c (expand_vector_broadcast): Likewise.
7312 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7313 Alan Hayward <alan.hayward@arm.com>
7314 David Sherwood <david.sherwood@arm.com>
7316 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7317 scaled index even if the unscaled address was invalid.
7318 Don't increase the complexity of using a scale in that case.
7320 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7321 Alan Hayward <alan.hayward@arm.com>
7322 David Sherwood <david.sherwood@arm.com>
7324 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7325 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7326 UNITS_PER_WORD bytes.
7327 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7328 based on REGMODE_NATURAL_SIZE of the inner mode.
7329 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7330 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7331 * expmed.c (lowpart_bit_field_p): Divide the value up into
7332 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7333 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7334 whether something is likely to occupy more than one register.
7336 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7339 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7340 profile_count::adjust_for_ipa_scaling.
7341 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7342 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7344 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7346 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7348 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7349 (split_group): Add total_orig and total_new arguments, estimate the
7350 number of statements related to the store group without store merging
7351 and with store merging.
7352 (imm_store_chain_info::output_merged_store): Adjust split_group
7353 callers, punt if estimated number of statements with store merging
7354 is not smaller than estimated number of statements without it.
7356 (handled_load): Remove has_single_use checks.
7357 (pass_store_merging::process_store): Likewise.
7359 2017-11-09 Richard Biener <rguenther@suse.de>
7361 PR tree-optimization/82902
7362 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7364 2017-11-09 Martin Liska <mliska@suse.cz>
7367 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7370 2017-11-09 Martin Liska <mliska@suse.cz>
7372 PR tree-optimization/82669
7373 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7375 2017-11-09 Martin Liska <mliska@suse.cz>
7377 PR gcov-profile/48463
7378 * coverage.c (coverage_begin_function): Output also end locus
7379 of a function and information whether the function is
7381 * gcov-dump.c (tag_function): Parse and print the information.
7382 * gcov.c (INCLUDE_MAP): Add include.
7383 (INCLUDE_SET): Likewise.
7384 (struct line_info): Move earlier in the source file because
7385 of vector<line_info> in function_info structure.
7386 (line_info::line_info): Likewise.
7387 (line_info::has_block): Likewise.
7388 (struct source_info): Add new member index.
7389 (source_info::get_functions_at_location): New function.
7390 (function_info::group_line_p): New function.
7391 (output_intermediate_line): New function.
7392 (output_intermediate_file): Use the mentioned function.
7393 (struct function_start): New.
7394 (struct function_start_pair_hash): Likewise.
7395 (process_file): Add code that identifies group functions.
7396 Assign lines either to global or function scope.
7397 (generate_results): Skip artificial functions.
7398 (find_source): Assign index for each source file.
7399 (read_graph_file): Read new flag artificial and end_line.
7400 (add_line_counts): Assign it either to global of function scope.
7401 (accumulate_line_counts): Isolate core of the function to
7402 accumulate_line_info and call it for both function and global
7404 (accumulate_line_info): New function.
7405 (output_line_beginning): Fix GNU coding style.
7406 (print_source_line): New function.
7407 (output_line_details): Likewise.
7408 (output_function_details): Likewise.
7409 (output_lines): Iterate both source (global) scope and function
7411 (struct function_line_start_cmp): New class.
7412 * doc/gcov.texi: Reflect changes in documentation.
7414 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7417 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7418 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7419 and similarly for not instead of neg.
7421 2017-11-08 Andi Kleen <ak@linux.intel.com>
7423 * config/i386/i386.opt: Add -mforce-indirect-call.
7424 * config/i386/predicates.md: Check for flag_force_indirect_call.
7425 * doc/invoke.texi: Document -mforce-indirect-call
7427 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7429 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7431 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7432 config/riscv/riscv.c (predict.h): New include.
7433 (riscv_slow_unaligned_access_p): No longer static.
7434 (riscv_block_move_straight): Add require.
7435 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7437 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7440 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7441 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7442 nonimmediate_operand predicate for operand 1 instead of
7445 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7447 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7449 * config/aarch64/constraints.md (Uml): New constraint.
7450 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7453 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7455 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7456 of two vec_duplicates into a vec_concat.
7458 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7460 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7461 Simplify vec_merge of vec_duplicate and vec_concat.
7462 * config/aarch64/constraints.md (Utq): New constraint.
7463 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7466 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7468 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7469 Simplify vec_merge of vec_duplicate and const_vector.
7470 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7472 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7473 mode iterator. Update predicate on operand 1 to
7474 handle non-const_vec constants. Delete constraints.
7475 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7477 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7479 PR tree-optimization/78821
7480 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7482 (store_operand_info::store_operand_info): Initialize it to false.
7483 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7484 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7485 store in the group, and if chain_info is non-NULL, to ignore altogether
7487 (compatible_load_p): Fail if bit_not_p does not match.
7488 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7489 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7490 (pass_store_merging::process_store): Adjust
7491 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7492 call terminate_all_aliasing_chains newly when adding a store into
7493 a chain with non-NULL chain_info.
7495 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7497 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7499 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7501 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7503 (aarch64_layout_frame): Initialise emit_frame_chain.
7504 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7505 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7507 2017-11-08 Martin Liska <mliska@suse.cz>
7509 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7510 of gimple_call_internal_p.
7512 2017-11-07 Tom de Vries <tom@codesourcery.com>
7514 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7516 2017-11-07 Tom de Vries <tom@codesourcery.com>
7518 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7521 2017-11-08 Martin Liska <mliska@suse.cz>
7524 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7526 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7528 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7530 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7533 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7534 (*cmp<mode>_ccz_1): New insn with $k alternative.
7537 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7538 RTX_COMM_COMPARE as commutative as well.
7539 (ix86_binary_operator_ok): Formatting fix.
7540 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7541 *<code><mode>3<mask_name><round_saeonly_name>,
7542 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7543 *<s>mul<mode>3_highpart<mask_name>,
7544 *vec_widen_umult_even_v16si<mask_name>,
7545 *vec_widen_umult_even_v8si<mask_name>,
7546 *vec_widen_umult_even_v4si<mask_name>,
7547 *vec_widen_smult_even_v16si<mask_name>,
7548 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7549 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7550 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7551 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7552 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7553 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7554 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7555 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7556 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7557 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7558 ix86_binary_operator_ok. Formatting fixes.
7559 (*<plusminus_insn><mode>3<mask_name><round_name>,
7560 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7563 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7565 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7566 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7567 eq and ne if TARGET_ISEL.
7568 (cmp): New code_iterator.
7569 (UNS, UNSU_, UNSIK): New code_attrs.
7570 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7571 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7573 ("eq<mode>3"): ... this.
7574 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7576 ("ne<mode>3"): ... this.
7578 2017-11-07 Julia Koval <julia.koval@intel.com>
7581 * common/config/i386/i386-common.c
7582 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7583 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7584 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7585 * config/i386/i386.opt: Ditto.
7586 * config/i386/i386.c (ix86_target_string): Ditto.
7587 (ix86_option_override_internal): Ditto.
7588 (ix86_init_mpx_builtins): Move MPX to args2.
7589 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7590 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7591 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7592 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7593 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7594 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7595 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7597 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7600 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7601 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7602 (zero-extendsidi peephole2): Remove peephole.
7604 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7607 * stor-layout.c: Include attribs.h.
7608 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7609 explicit lookup of "aligned" attribute.
7611 2017-11-07 Andrew Waterman <andrew@sifive.com>
7613 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7614 (riscv_expand_block_move): Likewise.
7615 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7617 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7618 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7619 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7620 (riscv_adjust_block_mem): Likewise.
7621 (riscv_block_move_loop): Likewise.
7622 (riscv_expand_block_move): Likewise.
7623 * config/riscv/riscv.md (movmemsi): New pattern.
7625 2017-11-07 Michael Clark <michaeljclark@mac.com>
7627 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7628 (MUSL_DYNAMIC_LINKER): Likewise.
7630 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7632 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7633 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7636 2017-11-07 Richard Biener <rguenther@suse.de>
7638 * match.pd: Fix build.
7640 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7641 Jackson Woodruff <jackson.woodruff@arm.com>
7643 PR tree-optimization/71026
7644 * match.pd: Canonicalize negate in division.
7646 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7649 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7651 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7653 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7654 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7656 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7658 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7659 non-scalar integral types.
7660 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7661 (-(A-B), -(~A)): New transformations.
7663 2017-11-07 Tom de Vries <tom@codesourcery.com>
7665 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7666 semicolon after "do {} while (0)".
7667 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7668 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7669 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7670 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7671 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7673 2017-11-07 Tom de Vries <tom@codesourcery.com>
7675 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7676 after "do {} while (0)".
7677 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7678 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7679 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7680 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7681 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7683 2017-11-07 Tom de Vries <tom@codesourcery.com>
7686 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7688 (arm_rtx_costs_internal): Add missing semicolon after
7689 HANDLE_NARROW_SHIFT_ARITH call.
7691 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7693 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7694 disable isel if it was not set explicitly.
7696 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7698 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7699 (add_type_attribute) likewise.
7701 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7703 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7704 of struct ix86_frame.
7705 (ix86_initial_elimination_offset): Likewise.
7706 (ix86_expand_split_stack_prologue): Likewise.
7708 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7710 * tree-vrp.h (enum value_range_type): Update stale comment.
7712 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7713 Alan Hayward <alan.hayward@arm.com>
7714 David Sherwood <david.sherwood@arm.com>
7716 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7717 (aarch64_expand_vec_perm_const): Take the number of units too.
7718 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7719 (aarch64_expand_vec_perm_const): Likewise.
7720 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7721 (vec_perm<mode>): Update accordingly.
7723 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7724 Alan Hayward <alan.hayward@arm.com>
7725 David Sherwood <david.sherwood@arm.com>
7727 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7728 Take the number of units too.
7729 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7730 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7731 but check for a vector mode before rather than after the call.
7732 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7733 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7734 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7735 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7736 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7737 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7738 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7739 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7740 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7741 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7742 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7743 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7744 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7745 (aarch64_sqdmull2_n<mode>): Update accordingly.
7747 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7748 Alan Hayward <alan.hayward@arm.com>
7749 David Sherwood <david.sherwood@arm.com>
7751 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7752 the number of units too.
7753 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7754 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7755 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7756 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7757 (vec_store_lanesxi<mode>): Update accordingly.
7759 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7760 Alan Hayward <alan.hayward@arm.com>
7761 David Sherwood <david.sherwood@arm.com>
7763 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7764 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7765 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7766 of units rather than the mode.
7767 * config/aarch64/iterators.md (nunits): New mode attribute.
7768 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7769 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7770 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7771 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7772 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7773 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7774 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7775 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7776 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7777 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7778 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7779 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7780 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7781 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7782 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7783 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7784 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7785 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7786 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7787 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7788 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7789 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7790 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7791 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7792 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7793 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7794 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7795 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7796 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7797 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7798 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7799 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7800 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7801 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7802 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7803 (aarch64_simd_vec_setv2di): Likewise.
7805 2017-11-06 Carl Love <cel@us.ibm.com>
7807 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7809 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7810 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7812 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7814 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7815 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7816 Add power 8 macro expansions.
7817 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7818 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7819 power 8 instructions. (VSX_XXBR): Add iterator.
7821 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7823 * config/arm/arm.md (predicable_short_it): Change default to "no",
7824 improve documentation, remove uses that are identical to the default.
7825 (enabled_for_depr_it): Rename to enabled_for_short_it.
7826 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7827 Remove default uses.
7828 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7829 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7830 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7831 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7833 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7836 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7837 float128 helper macros, which are no longer used after deleting
7838 the old 'q' built-in functions, and moving the round to odd
7839 built-in functions to being special built-in functions.
7840 (BU_FLOAT128_2): Likewise.
7841 (BU_FLOAT128_1_HW): Likewise.
7842 (BU_FLOAT128_2_HW): Likewise.
7843 (BU_FLOAT128_3_HW): Likewise.
7844 (FABSQ): Delete old 'q' built-in functions.
7845 (COPYSIGNQ): Likewise.
7846 (SQRTF128_ODD): Move round to odd built-in functions to be
7847 special built-in functions, so that we can handle
7848 -mabi=ieeelongdouble.
7849 (TRUNCF128_ODD): Likewise.
7850 (ADDF128_ODD): Likewise.
7851 (SUBF128_ODD): Likewise.
7852 (MULF128_ODD): Likewise.
7853 (DIVF128_ODD): Likewise.
7854 (FMAF128_ODD): Likewise.
7855 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7856 built-in names to 'f128'.
7857 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7858 old 'q' built-in functions, as the machine independent code for
7859 'f128' built-in functions handles this.
7860 (rs6000_expand_builtin): Add expansion for float128 round to odd
7861 functions, keying off on -mabi=ieeelongdouble of whether to use
7862 the KFmode or TFmode variant.
7863 (rs6000_init_builtins): Initialize the _Float128 round to odd
7865 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7866 _Float128 'q' built-in functions are now mapped into the new
7867 'f128' built-in functions.
7869 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7871 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7872 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7874 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7877 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7878 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7879 * toplev.c: Include "ipa-fnsummary.h".
7880 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7882 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7884 PR tree-optimization/82838
7885 * gimple-ssa-store-merging.c
7886 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7887 on a separate gimple_seq which is then appended to seq.
7889 2017-11-06 Jeff Law <law@redhat.com>
7892 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7893 (get_probe_interval): New functions.
7894 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7895 (ix86_adjust_stack_and_probe): Likewise.
7896 (output_adjust_stack_and_probe): Likewise.
7897 (ix86_emit_probe_stack_range): Likewise.
7898 (ix86_expand_prologue): Likewise.
7900 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7902 PR tree-optimization/82816
7903 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7904 if the modes of the two types are the same.
7905 (convert_plusminus_to_widen): Likewise.
7907 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7909 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7911 (usadv16qi): New define_expand.
7912 (usadv8hi): New define_expand.
7914 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7917 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7919 (inline_transform): Likewise.
7920 * predict.c (counts_to_freqs): Remove useless conditional.
7921 * profile-count.h (profile_count::apply_scale): Move sanity check.
7922 * tree-inline.c (copy_bb): Always scale.
7923 (copy_cfg_body): Likewise.
7925 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7928 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7930 (*cmp_ite0): Add enabled_for_depr_it attribute.
7931 (*cmp_ite1): Likewise.
7933 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7935 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7938 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7940 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7941 return the value by pointer.
7942 (extract_range_from_multiplicative_op_1): Update accordingly.
7943 Return as soon as an operation fails.
7945 2017-11-05 Tom de Vries <tom@codesourcery.com>
7948 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7949 (DEF_SANITIZER_BUILTIN): ... here.
7950 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7951 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7953 2017-11-05 Tom de Vries <tom@codesourcery.com>
7956 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7958 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7959 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7960 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7962 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7963 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7964 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7966 2017-11-05 Tom de Vries <tom@codesourcery.com>
7969 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7972 2017-11-04 Michael Clark <michaeljclark@mac.com>
7974 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7975 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7977 (*addsi3_extended): Likewise.
7978 (*addsi3_extended2): Likewise.
7979 (<optab>si3): Likewise.
7980 (<optab>di3): Likewise.
7981 (<optab><mode>3): Likewise.
7982 (<*optabe>si3_internal): Likewise.
7983 (zero_extendqi<SUPERQI:mode>2): Likewise.
7984 (*add<mode>hi3): Likewise.
7985 (*xor<mode>hi3): Likewise.
7986 (<optab>di3): Likewise.
7987 (*<optab>si3_extend): Likewise.
7988 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7989 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7990 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7992 2017-11-04 Andrew Waterman <andrew@sifive.com>
7994 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7995 TARGET_STRICT_ALIGN based upon -mtune argument.
7997 2017-11-04 Andrew Waterman <andrew@sifive.com>
7999 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8001 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8003 * config/i386/i386.c (choose_basereg): Use optional scratch
8004 register and add assertion.
8005 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8006 needed, and don't allocate stack.
8007 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8008 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8009 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8011 2017-11-03 Jeff Law <law@redhat.com>
8013 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8014 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8015 to probe at the start of a noreturn function.
8017 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8019 PR tree-optimization/78821
8020 * gimple-ssa-store-merging.c: Update the file comment.
8021 (MAX_STORE_ALIAS_CHECKS): Define.
8022 (struct store_operand_info): New type.
8023 (store_operand_info::store_operand_info): New constructor.
8024 (struct store_immediate_info): Add rhs_code and ops data members.
8025 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8026 and op1r arguments to the ctor, initialize corresponding data members.
8027 (struct merged_store_group): Add load_align_base and load_align
8029 (merged_store_group::merged_store_group): Initialize them.
8030 (merged_store_group::do_merge): Update them.
8031 (merged_store_group::apply_stores): Pick the constant for
8032 encode_tree_to_bitpos from one of the two operands, or skip
8033 encode_tree_to_bitpos if neither operand is a constant.
8034 (class pass_store_merging): Add process_store method decl. Remove
8035 bool argument from terminate_all_aliasing_chains method decl.
8036 (pass_store_merging::terminate_all_aliasing_chains): Remove
8037 var_offset_p argument and corresponding handling.
8038 (stmts_may_clobber_ref_p): New function.
8039 (compatible_load_p): New function.
8040 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8041 if there is overlap and rhs_code is not INTEGER_CST. For
8042 non-overlapping stores terminate group if rhs is not mergeable.
8043 (get_alias_type_for_stmts): Change first argument from
8044 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8045 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8046 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8048 (get_location_for_stmts): Change first argument from
8049 auto_vec<gimple *> & to vec<gimple *> &.
8050 (struct split_store): Remove orig_stmts data member, add orig_stores.
8051 (split_store::split_store): Create orig_stores rather than orig_stmts.
8052 (find_constituent_stmts): Renamed to ...
8053 (find_constituent_stores): ... this. Change second argument from
8054 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8055 to info structures rather than the statements.
8056 (split_group): Rename ALLOW_UNALIGNED argument to
8057 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8058 it. Adjust find_constituent_stores caller.
8059 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8060 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8061 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8062 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8063 (mem_valid_for_store_merging): New function.
8064 (handled_load): New function.
8065 (pass_store_merging::process_store): New method.
8066 (pass_store_merging::execute): Use process_store method. Adjust
8067 terminate_all_aliasing_chains caller.
8069 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8071 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8072 Return true for more constants, symbols and label references.
8073 (aarch64_valid_floating_const): Remove unused function.
8075 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8078 * config/aarch64/aarch64.c (aarch64_layout_frame):
8079 Undo forcing of LR at bottom of frame.
8081 2017-11-03 Jeff Law <law@redhat.com>
8084 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8085 for int_registers_saved.
8087 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8088 extracted from tree-ssa-dom.c.
8089 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8090 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8091 (record_equivalences_from_incoming_edge): Add additional argument
8092 to single_pred_edge_ignoring_loop_edges call.
8093 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8094 (uncprop_dom_walker::before_dom_children): Add additional argument
8095 to single_pred_edge_ignoring_loop_edges call.
8096 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8097 single_pred_edge_ignoring_loop_edges rather than open coding.
8098 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8100 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8102 * match.pd (-(-A)): Rewrite.
8104 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8106 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8107 (rs6000_emit_int_cmove): New declaration.
8108 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8109 (rs6000_emit_sISEL): Delete.
8110 (rs6000_emit_int_cmove): Make non-static.
8111 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8112 instead of rs6000_emit_sISEL.
8114 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8116 * asan.c (create_cond_insert_point): Maintain profile.
8117 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8119 * basic-block.h (struct basic_block_def): Remove frequency.
8120 (EDGE_FREQUENCY): Use to_frequency
8121 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8123 (find_traces): Update to use to_frequency.
8124 (find_traces_1_round): Likewise; use only IPA counts.
8125 (bb_to_key): Likewise.
8126 (connect_traces): Use IPA counts only.
8127 (copy_bb_p): Update to use to_frequency.
8128 (fix_up_crossing_landing_pad): Likewise.
8129 (sanitize_hot_paths): Likewise.
8130 * bt-load.c (basic_block_freq): Likewise.
8131 * cfg.c (init_flow): Set count_max to uninitialized.
8132 (check_bb_profile): Remove frequencies; check counts.
8133 (dump_bb_info): Do not dump frequencies.
8134 (update_bb_profile_for_threading): Update counts only.
8135 (scale_bbs_frequencies_int): Likewise.
8136 (MAX_SAFE_MULTIPLIER): Remove.
8137 (scale_bbs_frequencies_gcov_type): Update counts only.
8138 (scale_bbs_frequencies_profile_count): Update counts only.
8139 (scale_bbs_frequencies): Update counts only.
8140 * cfg.h (struct control_flow_graph): Add count-max.
8141 (update_bb_profile_for_threading): Update prototype.
8142 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8143 (find_many_sub_basic_blocks): Likewise.
8144 * cfgcleanup.c (try_forward_edges): Likewise.
8145 (try_crossjump_to_edge): Likewise.
8146 * cfgexpand.c (expand_gimple_cond): Likewise.
8147 (expand_gimple_tailcall): Likewise.
8148 (construct_init_block): Likewise.
8149 (construct_exit_block): Likewise.
8150 * cfghooks.c (verify_flow_info): Check consistency of counts.
8151 (dump_bb_for_graph): Do not dump frequencies.
8152 (split_block_1): Do not update frequencies.
8153 (split_edge): Do not update frequencies.
8154 (make_forwarder_block): Do not update frequencies.
8155 (duplicate_block): Do not update frequencies.
8156 (account_profile_record): Do not update frequencies.
8157 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8158 for global heuristics.
8159 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8160 (expected_loop_iterations_unbounded): Use counts only.
8161 * cfgloopmanip.c (scale_loop_profile): Simplify.
8162 (create_empty_loop_on_edge): Simplify
8164 (duplicate_loop_to_header_edge): Simplify
8165 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8166 (update_br_prob_note): Take care of removing note when profile
8168 (relink_block_chain): Do not dump frequency.
8169 (rtl_account_profile_record): Use to_frequency.
8170 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8171 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8172 (cgraph_update_edges_for_call_stmt_node): Likewise.
8173 (cgraph_edge::verify_count_and_frequency): Update.
8174 (cgraph_node::verify_node): Temporarily disable frequency verification.
8175 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8176 to_cgraph_frequency.
8177 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8178 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8180 (cgraph_node::expand_thunk): Update profile.
8181 * except.c (dw2_build_landing_pads): Do not update frequency.
8182 * final.c (compute_alignments): Use to_frequency.
8183 (dump_basic_block_info): Do not dump frequency.
8184 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8185 (dump_gimple_bb_header): Do not dump frequency.
8186 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8188 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8189 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8190 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8191 (init_before_recovery): Do not update frequency.
8192 (sched_create_recovery_edges): Do not update frequency.
8193 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8194 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8195 (ipa_cp_c_finalize): Set max_count to uninitialized.
8196 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8197 (param_change_prob): Use counts.
8198 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8200 * ipa-split.c (consider_split): Use to_frequency.
8201 (split_function): Use to_frequency.
8202 * ira-build.c (loop_compare_func): Likewise.
8203 (mark_loops_for_removal): Likewise.
8204 (mark_all_loops_for_removal): Likewise.
8205 * loop-doloop.c (doloop_modify): Do not update frequency.
8206 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8208 * lto-streamer-in.c (input_function): Update count_max.
8209 * omp-expand.c (expand_omp_taskreg): Update count_max.
8210 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8211 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8212 (maybe_hot_count_p): Use ipa counts only.
8213 (maybe_hot_bb_p): Simplify.
8214 (maybe_hot_edge_p): Simplify.
8215 (probably_never_executed): Do not take frequency argument.
8216 (probably_never_executed_bb_p): Do not pass frequency.
8217 (probably_never_executed_edge_p): Likewise.
8218 (combine_predictions_for_bb): Check that profile is nonzero.
8219 (propagate_freq): Do not set frequency.
8220 (drop_profile): Simplify.
8221 (counts_to_freqs): Simplify.
8222 (expensive_function_p): Use to_frequency.
8223 (propagate_unlikely_bbs_forward): Simplify.
8224 (determine_unlikely_bbs): Simplify.
8225 (estimate_bb_frequencies): Add hack to silence graphite issues.
8226 (compute_function_frequency): Use ipa counts.
8227 (pass_profile::execute): Update.
8228 (rebuild_frequencies): Use counts only.
8229 (force_edge_cold): Use counts only.
8230 * profile-count.c (profile_count::dump): Dump new count types.
8231 (profile_count::differs_from_p): Check compatiblity.
8232 (profile_count::to_frequency): New function.
8233 (profile_count::to_cgraph_frequency): New function.
8234 * profile-count.h (struct function): Declare.
8235 (enum profile_quality): Add profile_guessed_local and
8236 profile_guessed_global0.
8237 (class profile_proability): Decrease number of bits to 29;
8238 update from_reg_br_prob_note and to_reg_br_prob_note.
8239 (class profile_count: Update comment; decrease number of bits
8240 to 61. Check compatibility.
8241 (profile_count::compatible_p): New private member function.
8242 (profile_count::ipa_p): New member function.
8243 (profile_count::operator<): Handle global zero correctly.
8244 (profile_count::operator>): Handle global zero correctly.
8245 (profile_count::operator<=): Handle global zero correctly.
8246 (profile_count::operator>=): Handle global zero correctly.
8247 (profile_count::nonzero_p): New member function.
8248 (profile_count::force_nonzero): New member function.
8249 (profile_count::max): New member function.
8250 (profile_count::apply_scale): Handle IPA scalling.
8251 (profile_count::guessed_local): New member function.
8252 (profile_count::global0): New member function.
8253 (profile_count::ipa): New member function.
8254 (profile_count::to_frequency): Declare.
8255 (profile_count::to_cgraph_frequency): Declare.
8256 * profile.c (OVERLAP_BASE): Delete.
8257 (compute_frequency_overlap): Delete.
8258 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8259 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8260 * sched-ebb.c (rank): Use counts only.
8261 * shrink-wrap.c (handle_simple_exit): Use counts only.
8262 (try_shrink_wrapping): Use counts only.
8263 (place_prologue_for_one_component): Use counts only.
8264 * tracer.c (find_best_predecessor): Use to_frequency.
8265 (find_trace): Use to_frequency.
8266 (tail_duplicate): Use to_frequency.
8267 * trans-mem.c (expand_transaction): Do not update frequency.
8268 * tree-call-cdce.c: Do not update frequency.
8269 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8270 (gimple_merge_blocks): Likewise.
8271 (gimple_split_edge): Likewise.
8272 (gimple_duplicate_sese_region): Likewise.
8273 (gimple_duplicate_sese_tail): Likewise.
8274 (move_sese_region_to_fn): Likewise.
8275 (gimple_account_profile_record): Likewise.
8276 (insert_cond_bb): Likewise.
8277 * tree-complex.c (expand_complex_div_wide): Likewise.
8278 * tree-eh.c (lower_resx): Update profile.
8279 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8281 (initialize_cfun): Do not initialize frequencies
8282 (freqs_to_counts): Delete.
8283 (copy_cfg_body): Ignore count parameter.
8284 (copy_body): Update.
8285 (expand_call_inline): Update count_max.
8286 (optimize_inline_calls): Update count_max.
8287 (tree_function_versioning): Update count_max.
8288 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8289 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8291 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8292 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8293 (try_peel_loop): Likewise.
8294 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8296 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8297 (tree_transform_and_unroll_loop): Do not use frequencies
8298 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8299 Use reliable prediction only.
8300 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8301 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8302 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8303 probability scaling.
8304 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8305 not update frequency
8306 (any_remaining_duplicated_blocks): Likewise.
8307 (update_profile): Likewise.
8308 (estimated_freqs_path): Delete.
8309 (freqs_to_counts_path): Delete.
8310 (clear_counts_path): Delete.
8311 (ssa_fix_duplicate_block_edges): Likewise.
8312 (duplicate_thread_path): Likewise.
8313 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8314 * tree-tailcall.c (decrease_profile): Do not update frequency.
8315 (eliminate_tail_call): Likewise.
8316 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8317 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8318 (optimize_mask_stores): Likewise.
8319 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8320 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8321 (ubsan_expand_ptr_ifn): Update profile.
8322 * value-prof.c (gimple_ic): Simplify.
8323 * value-prof.h (gimple_ic): Update prototype.
8324 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8325 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8327 (want_inline_self_recursive_call_p): Likewise.
8328 (resolve_noninline_speculation): Only cummulate defined counts.
8329 (inline_small_functions): Use nonzero_p.
8330 (ipa_inline): Do not access freed node.
8332 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8334 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8335 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8337 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8339 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8340 non-legitimate address.
8342 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8344 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8345 (*lt0_<mode>di, *lt0_<mode>si): New.
8347 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8349 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8350 TARGET_PAIRED_FLOAT.
8352 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8353 Jim Wilson <jim.wilson@linaro.org>
8355 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8356 * config/aarch64/aarch64-tune.md: Regenerated.
8357 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8358 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8360 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8362 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8363 (arc_expand_prologue): Restore blink for millicode.
8364 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8366 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8369 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8370 gen_vec_duplicate after forcing the scalar into a register.
8372 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8374 * combine (try_combine): Print the insns input to try_combine to the
8377 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8380 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8381 Remove second argument from aarch64_process_target_attr call.
8382 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8384 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8386 (aarch64_handle_attr_arch): Remove second argument.
8387 (aarch64_handle_attr_cpu): Ditto.
8388 (aarch64_handle_attr_tune): Ditto.
8389 (aarch64_handle_attr_isa_flags): Ditto.
8390 (aarch64_process_one_target_attr): Ditto.
8391 (aarch64_process_target_attr): Ditto.
8392 (aarch64_option_valid_attribute_p): Remove second argument.
8393 on aarch64_process_target_attr call.
8395 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8397 * diagnostic.c: Include "selftest-diagnostic.h".
8398 (selftest::assert_location_text): New function.
8399 (selftest::test_diagnostic_get_location_text): New function.
8400 (selftest::diagnostic_c_tests): Call it.
8402 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8404 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8405 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8406 (class selftest::test_diagnostic_context): Move to...
8407 * selftest-diagnostic.c: New file.
8408 * selftest-diagnostic.h: New file.
8410 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8412 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8413 offset range for FT32B.
8414 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8415 * config/ft32/ft32.md: Add TARGET_NOPM.
8416 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8417 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8419 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8421 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8423 2017-11-02 Jeff Law <law@redhat.com>
8425 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8426 virtual keyword on FINAL OVERRIDE members.
8428 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8429 virtuals together. Add virtual destructor.
8430 (substitute_and_fold_engine): Similarly.
8432 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8434 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8436 2017-11-02 Richard Biener <rguenther@suse.de>
8438 PR tree-optimization/82795
8439 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8441 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8443 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8445 (gcc_GAS_CHECK_FEATURE): Remove.
8446 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8447 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8448 * configure: Regenerate.
8450 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8452 * config/arc/arc.c (hwloop_optimize): Account for empty
8455 2017-11-02 Richard Biener <rguenther@suse.de>
8458 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8459 Truncate ARRAY_REF index and element size.
8461 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8463 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8465 2017-11-01 Jeff Law <law@redhat.com>
8467 * tree-ssa-ccp.c (ccp_folder): New class derived from
8468 substitute_and_fold_engine.
8469 (ccp_folder::get_value): New member function.
8470 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8471 (ccp_fold_stmt): Remove prototype.
8472 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8473 * tree-ssa-copy.c (copy_folder): New class derived from
8474 substitute_and_fold_engine.
8475 (copy_folder::get_value): Renamed from get_value.
8476 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8477 * tree-vrp.c (vrp_folder): New class derived from
8478 substitute_and_fold_engine.
8479 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8480 (vrp_folder::get_value): New member function.
8481 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8482 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8483 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8484 provide a class interface to folder/substitute routines.
8485 (ssa_prop_fold_stmt_fn): Remove typedef.
8486 (ssa_prop_get_value_fn): Likewise.
8487 (subsitute_and_fold): Remove prototype.
8488 (replace_uses_in): Likewise.
8489 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8490 Renamed from replace_uses_in. Call the virtual member function
8491 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8492 (substitute_and_fold_dom_walker): Remove initialization of
8493 data member entries for calbacks. Add substitute_and_fold_engine
8494 member and initialize it.
8495 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8496 member functions for get_value, replace_phi_args_in c
8497 replace_uses_in, and fold_stmt calls.
8498 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8499 substitute_and_fold. Remove assert. Update ctor call.
8501 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8502 (ssa_prop_visit_phi_fn): Likewise.
8503 (class ssa_propagation_engine): New class to provide an interface
8505 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8507 (ssa_prop_visit_phi): Likewise.
8508 (ssa_propagation_engine::simulate_stmt): Moved into class.
8509 Call visit_phi/visit_stmt from the class rather than via
8510 file scoped static variables.
8511 (ssa_propagation_engine::simulate_block): Moved into class.
8512 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8513 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8514 set file scoped statics for the visit_stmt/visit_phi callbacks.
8515 * tree-complex.c (complex_propagate): New class derived from
8516 ssa_propagation_engine.
8517 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8518 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8519 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8521 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8522 ssa_propagation_engine.
8523 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8524 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8525 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8526 * tree-ssa-copy.c (copy_prop): New class derived from
8527 ssa_propagation_engine.
8528 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8529 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8530 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8531 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8532 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8533 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8534 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8536 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8538 PR rtl-optimization/82778
8539 PR rtl-optimization/82597
8540 * compare-elim.c (struct comparison): Add in_a_setter field.
8541 (find_comparison_dom_walker::before_dom_children): Remove killed
8542 bitmap and df_simulate_find_defs call, instead walk the defs.
8543 Compute last_setter and initialize in_a_setter. Merge definitions
8544 with first initialization for a few variables.
8545 (try_validate_parallel): Use insn_invalid_p instead of
8546 recog_memoized. Return insn rather than just the pattern.
8547 (try_merge_compare): Fix up comment. Don't uselessly test if
8548 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8550 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8553 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8554 Alan Hayward <alan.hayward@arm.com>
8555 David Sherwood <david.sherwood@arm.com>
8557 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8558 aarch64_hard_regno_nregs to get the number of registers
8561 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8562 Alan Hayward <alan.hayward@arm.com>
8563 David Sherwood <david.sherwood@arm.com>
8565 * config/aarch64/constraints.md (Upl): Rename to...
8567 * config/aarch64/aarch64.md
8568 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8571 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8572 Alan Hayward <alan.hayward@arm.com>
8573 David Sherwood <david.sherwood@arm.com>
8575 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8576 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8579 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8580 Alan Hayward <alan.hayward@arm.com>
8581 David Sherwood <david.sherwood@arm.com>
8583 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8584 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8585 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8587 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8588 of a single element.
8589 * config/aarch64/iterators.md: Add a comment above the permute
8590 unspecs to say that they are generated directly by
8591 aarch64_expand_vec_perm_const.
8592 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8594 2017-11-01 Nathan Sidwell <nathan@acm.org>
8596 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8598 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8600 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8601 and medany code models, and describe what they do.
8603 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8605 Revert accidental duplicate:
8607 * combine.c (can_change_dest_mode): Reject changes in
8608 REGMODE_NATURAL_SIZE.
8610 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8612 PR rtl-optimization/64682
8613 PR rtl-optimization/69567
8614 PR rtl-optimization/69737
8615 PR rtl-optimization/82683
8616 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8617 register mentioned in the note, drop the note, unless it came from I3,
8618 in which case it should go to I3 again.
8620 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8622 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8623 and return false if not.
8624 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8626 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8628 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8629 range is known to be empty.
8631 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8632 Alan Hayward <alan.hayward@arm.com>
8633 David Sherwood <david.sherwood@arm.com>
8635 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8636 and CONST_VECTOR_NUNITS instead of computing the number of units from
8637 the byte sizes of the vector and element.
8638 (simplify_binary_operation_1): Likewise.
8639 (simplify_const_binary_operation): Likewise.
8640 (simplify_ternary_operation): Likewise.
8642 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8643 Alan Hayward <alan.hayward@arm.com>
8644 David Sherwood <david.sherwood@arm.com>
8646 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8647 (int_mem_offset): ...this new function.
8648 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8649 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8650 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8653 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8654 Alan Hayward <alan.hayward@arm.com>
8655 David Sherwood <david.sherwood@arm.com>
8657 * lower-subreg.c (interesting_mode_p): New function.
8658 (compute_costs, find_decomposable_subregs, decompose_register)
8659 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8660 (resolve_clobber, dump_choices): Use it.
8662 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8663 Alan Hayward <alan.hayward@arm.com>
8664 David Sherwood <david.sherwood@arm.com>
8666 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8668 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8669 Alan Hayward <alan.hayward@arm.com>
8670 David Sherwood <david.sherwood@arm.com>
8672 * alias.c (find_base_value, find_base_term): Only process integer
8673 truncations. Check the precision rather than the size.
8675 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8676 Alan Hayward <alan.hayward@arm.com>
8677 David Sherwood <david.sherwood@arm.com>
8679 * machmode.h (is_narrower_int_mode): New function
8680 * optabs.c (expand_float, expand_fix): Use it.
8681 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8683 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8684 Alan Hayward <alan.hayward@arm.com>
8685 David Sherwood <david.sherwood@arm.com>
8687 * rtl.h (narrower_subreg_mode): New function.
8688 * ira-color.c (update_costs_from_allocno): Use it.
8690 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8691 Alan Hayward <alan.hayward@arm.com>
8692 David Sherwood <david.sherwood@arm.com>
8694 * optabs-query.h (convert_optab_p): New function, split out from...
8695 (convert_optab_handler): ...here.
8696 (widening_optab_handler): Delete.
8697 (find_widening_optab_handler): Remove permit_non_widening parameter.
8698 (find_widening_optab_handler_and_mode): Likewise. Provide an
8699 override that operates on mode class wrappers.
8700 * optabs-query.c (widening_optab_handler): Delete.
8701 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8702 parameter. Assert that the two modes are the same class and that
8703 the "from" mode is narrower than the "to" mode. Use
8704 convert_optab_handler instead of widening_optab_handler.
8705 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8706 instead of widening_optab_handler.
8707 * expr.c (expand_expr_real_2): Update calls to
8708 find_widening_optab_handler.
8709 * optabs.c (expand_widen_pattern_expr): Likewise.
8710 (expand_binop_directly): Take the insn_code as a parameter.
8711 (expand_binop): Only call find_widening_optab_handler for
8712 conversion optabs; use optab_handler otherwise. Update calls
8713 to find_widening_optab_handler and expand_binop_directly.
8714 Use convert_optab_handler instead of widening_optab_handler.
8715 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8716 find_widening_optab_handler and use scalar_mode rather than
8718 (convert_plusminus_to_widen): Likewise.
8720 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8721 Alan Hayward <alan.hayward@arm.com>
8722 David Sherwood <david.sherwood@arm.com>
8724 * machmode.h (fixed_size_mode): New class.
8725 * rtl.h (get_pool_mode): Return fixed_size_mode.
8726 * gengtype.c (main): Add fixed_size_mode.
8727 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8728 (get_raw_arg_mode): Likewise.
8729 * doc/tm.texi: Regenerate.
8730 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8731 * targhooks.c (default_get_reg_raw_mode): Likewise.
8732 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8733 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8734 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8735 (msp430_get_raw_result_mode): Likewise.
8736 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8737 * dbxout.c (dbxout_parms): Require fixed-size modes.
8738 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8739 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8740 * omp-low.c (lower_oacc_reductions): Likewise.
8741 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8742 (simplify_subreg): Update accordingly.
8743 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8744 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8745 that aren't fixed-size.
8746 (get_pool_mode): Return a fixed_size_mode.
8747 (output_constant_pool_2): Take a fixed_size_mode.
8749 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8750 Alan Hayward <alan.hayward@arm.com>
8751 David Sherwood <david.sherwood@arm.com>
8753 * doc/rtl.texi (vec_series): Document.
8754 (const): Say that the operand can be a vec_series.
8755 * rtl.def (VEC_SERIES): New rtx code.
8756 * rtl.h (const_vec_series_p_1): Declare.
8757 (const_vec_series_p): New function.
8758 * emit-rtl.h (gen_const_vec_series): Declare.
8759 (gen_vec_series): Likewise.
8760 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8761 (gen_vec_series): Likewise.
8762 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8763 * simplify-rtx.c (simplify_unary_operation): Handle negations
8765 (simplify_binary_operation_series): New function.
8766 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8767 (test_vector_ops_series): New function.
8768 (test_vector_ops): Call it.
8769 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8770 gen_const_vec_series.
8771 (altivec_lvsr): Likewise.
8772 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8774 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8775 Alan Hayward <alan.hayward@arm.com>
8776 David Sherwood <david.sherwood@arm.com>
8778 * doc/rtl.texi (const): Update description of address constants.
8779 Say that vector constants are allowed too.
8780 * common.md (E, F): Use CONSTANT_P instead of checking for
8782 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8783 checking for CONST_VECTOR.
8784 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8786 * expr.c (expand_expr_real_2): Check for vector modes instead
8787 of checking for CONST_VECTOR.
8788 * rtl.h (const_vec_p): New function.
8789 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8790 (unwrap_const_vec_duplicate): Handle them here too.
8792 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8793 David Malcolm <dmalcolm@redhat.com>
8794 Alan Hayward <alan.hayward@arm.com>
8795 David Sherwood <david.sherwood@arm.com>
8797 * rtl.h (vec_duplicate_p): New function.
8798 * selftest-rtl.c (assert_rtx_eq_at): New function.
8799 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8800 (assert_rtx_eq_at): Declare.
8801 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8802 * selftest-run-tests.c (selftest::run_tests): Call it.
8803 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8804 (simplify_unary_operation_1): Recursively handle vector duplicates.
8805 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8807 (simplify_subreg): Handle subregs of vector duplicates.
8808 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8809 (selftest::simplify_rtx_c_tests): New functions.
8811 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8812 Alan Hayward <alan.hayward@arm.com>
8813 David Sherwood <david.sherwood@arm.com>
8815 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8816 (gen_vec_duplicate): Likewise.
8817 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8819 (gen_const_vector): ...here.
8820 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8821 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8822 whose elements are all equal.
8823 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8824 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8825 (simplify_relational_operation): Likewise.
8826 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8828 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8829 (aarch64_expand_vector_init): Likewise.
8830 * config/arm/arm.c (neon_vdup_constant): Likewise.
8831 (neon_expand_vector_init): Likewise.
8832 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8833 (arm_block_set_unaligned_vect): Likewise.
8834 (arm_block_set_aligned_vect): Likewise.
8835 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8836 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8837 (expand_vec_perm_even_odd_pack): Likewise.
8838 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8839 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8840 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8841 gen_const_vec_duplicate.
8842 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8843 * config/mips/mips.c (mips_gen_const_int_vector): Use
8844 gen_const_vec_duplicate.
8845 (mips_expand_vector_init): Use CONST0_RTX.
8846 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8847 (define_split): Use gen_const_vec_duplicate.
8848 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8849 (define_split): Use gen_const_vec_duplicate.
8850 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8851 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8852 * config/spu/spu.c (spu_const): Likewise.
8854 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8855 Alan Hayward <alan.hayward@arm.com>
8856 David Sherwood <david.sherwood@arm.com>
8858 * combine.c (can_change_dest_mode): Reject changes in
8859 REGMODE_NATURAL_SIZE.
8861 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8863 * configure.ac (--enable-libssp): New.
8864 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8865 trying to determine target-specific default. Adjust indentation.
8866 * configure: Regenerated.
8867 * doc/install.texi (Configuration): Expand --disable-libssp
8870 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8872 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8875 2017-10-31 Martin Jambor <mjambor@suse.cz>
8878 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8880 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8882 * auto-profile.c (autofdo_source_profile::read): Use
8883 UNKNOWN_LOCATION rather than 0.
8884 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8885 (warning_at): ...this overload.
8886 (warning_at_rich_loc_n): Rename to...
8887 (warning_n): ...this overload.
8888 (error_at_rich_loc): Rename to...
8889 (error_at): ...this overload.
8890 (pedwarn_at_rich_loc): Rename to...
8891 (pedwarn): ...this overload.
8892 (permerror_at_rich_loc): Rename to...
8893 (permerror): ...this overload.
8894 (inform_at_rich_loc): Rename to...
8895 (inform): ...this overload.
8896 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8897 (diagnostic_n_impl_richloc): Rename to...
8898 (diagnostic_n_impl): ...this rich_location *-based decl.
8899 (inform_at_rich_loc): Rename to...
8900 (inform): ...this, and add an assertion.
8901 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8902 (warning_at_rich_loc): Rename to...
8903 (warning_at): ...this, and add an assertion.
8904 (warning_at_rich_loc_n): Rename to...
8905 (warning_n): ...this, and add an assertion.
8906 (warning_n): Update location_t-based implementation for removal of
8907 location_t-based diagnostic_n_impl.
8908 (pedwarn_at_rich_loc): Rename to...
8909 (pedwarn): ...this, and add an assertion.
8910 (permerror_at_rich_loc): Rename to...
8911 (permerror): ...this, and add an assertion.
8912 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8913 (error_at_rich_loc): Rename to...
8914 (error_at): ...this, and add an assertion.
8915 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8916 (driver::do_spec_on_infiles): Likewise.
8917 * substring-locations.c (format_warning_va): Update for renaming
8918 of inform_at_rich_loc.
8920 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8922 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8923 _Float<N>X built-in functions so that the variant without the
8924 "__builtin_" prefix is only enabled for the GNU C and Objective C
8925 languages when they are in non-strict ANSI/ISO mode.
8926 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8927 * target.def (floatn_builtin_p): Add a target hook to control
8928 whether _Float<N> and _Float<N>X built-in functions without the
8929 "__builtin_" prefix are enabled, and return true for C and
8930 Objective C in the default hook. Include langhooks.h in
8932 * targhooks.h (default_floatn_builtin_p): Likewise.
8933 * targhooks.c (default_floatn_builtin_p): Likewise.
8934 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8935 floatn_builtin_p target hook.
8936 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8938 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8939 Eric Botcazou <ebotcazou@adacore.com>
8941 PR rtl-optimization/81803
8942 * lra-constraints.c (curr_insn_transform): Also reload the whole
8943 register for a strict subreg no wider than a word if this is for
8944 a WORD_REGISTER_OPERATIONS target.
8946 2017-10-31 Jason Merrill <jason@redhat.com>
8948 * gdbinit.in: Skip over inlines from timevar.h.
8950 2017-10-31 Martin Liska <mliska@suse.cz>
8952 * doc/gcov.texi: Document new option.
8953 * gcov.c (print_usage): Likewise print it.
8954 (process_args): Support the argument.
8955 (format_count): New function.
8956 (format_gcov): Use the function.
8958 2017-10-31 Martin Liska <mliska@suse.cz>
8960 * gcov.c (struct name_map): do not use typedef.
8961 Define operator== and operator<.
8962 (name_search): Remove.
8963 (name_sort): Remove.
8964 (main): Do not allocate names.
8965 (process_file): Add vertical space.
8966 (generate_results): Use std::find.
8967 (release_structures): Do not release memory.
8968 (find_source): Use std::find.
8970 2017-10-31 Martin Liska <mliska@suse.cz>
8972 * gcov.c (struct line_info): Remove it's typedef.
8973 (line_info::line_info): Add proper ctor.
8974 (line_info::has_block): Do not use a typedef.
8975 (struct source_info): Do not use typedef.
8976 (circuit): Likewise.
8977 (get_cycles_count): Likewise.
8978 (output_intermediate_file): Iterate via vector iterator.
8979 (add_line_counts): Use std::vector methods.
8980 (accumulate_line_counts): Likewise.
8981 (output_lines): Likewise.
8983 2017-10-31 Martin Liska <mliska@suse.cz>
8985 * gcov.c (struct source_info): Remove typedef.
8986 (source_info::source_info): Add proper ctor.
8987 (accumulate_line_counts): Use struct, not it's typedef.
8988 (output_gcov_file): Likewise.
8989 (output_lines): Likewise.
8990 (main): Do not allocate an array.
8991 (output_intermediate_file): Use size of vector container.
8992 (process_file): Resize the vector.
8993 (generate_results): Do not preallocate, use newly added vector
8995 (release_structures): Do not release sources.
8996 (find_source): Use vector methods.
8997 (add_line_counts): Do not use typedef.
8999 2017-10-31 Martin Liska <mliska@suse.cz>
9001 * doc/gcov.texi: Document that.
9002 * gcov.c (add_line_counts): Mark lines with a non-executed
9004 (output_line_beginning): Handle such lines.
9005 (output_lines): Pass new argument.
9006 (output_intermediate_file): Print it in intermediate format.
9008 2017-10-31 Martin Liska <mliska@suse.cz>
9010 * color-macros.h: New file.
9011 * diagnostic-color.c: Factor out color related to macros to
9013 * doc/gcov.texi: Document -k option.
9014 * gcov.c (INCLUDE_STRING): Include string.h.
9015 (print_usage): Add -k option.
9016 (process_args): Parse it.
9017 (pad_count_string): New function.
9018 (output_line_beginning): Likewise.
9019 (DEFAULT_LINE_START): New macro.
9020 (output_lines): Support color output.
9022 2017-10-31 Martin Liska <mliska@suse.cz>
9024 PR gcov-profile/82633
9025 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9026 their interaction with GCOV infrastructure.
9027 * configure.ac: Add -fkeep-{inline,static}-functions to
9029 * configure: Regenerate.
9031 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9034 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9036 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9039 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9040 into a register if it does not fit into an immediate offset field.
9042 2017-10-31 Olivier Hainque <hainque@adacore.com>
9044 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9046 2017-10-31 Julia Koval <julia.koval@intel.com>
9048 * config.gcc: Add gfniintrin.h.
9049 * config/i386/gfniintrin.h: New.
9050 * config/i386/i386-builtin-types.def
9051 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9052 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9053 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9054 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9055 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9056 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9057 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9058 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9059 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9060 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9061 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9062 * config/i386/immintrin.h: Include gfniintrin.h.
9063 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9065 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9067 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9069 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9071 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9072 (arm_ashldi3_1bit): Remove pattern.
9073 (ashrdi3): Remove shift by 1 expansion.
9074 (arm_ashrdi3_1bit): Remove pattern.
9075 (lshrdi3): Remove shift by 1 expansion.
9076 (arm_lshrdi3_1bit): Remove pattern.
9077 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9078 cost of ashldi3 by 1.
9079 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9080 (<shift>di3_neon): Likewise.
9082 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9084 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9085 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9086 and (*aarch64_simd_mov<VQ:mode>).
9087 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9088 pattern alternative.
9089 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9090 attributes to match pattern alternative.
9092 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9094 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9095 * config/rs6000/emmintrin.h: New file.
9096 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9098 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9100 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9101 * (movdi_vfp_cortexa8): Remove pattern.
9103 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9105 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9107 (Specific, alpha*-dec-osf5.1): Remove.
9108 (Specific, mips-sgi-irix5): Remove.
9109 (Specific, mips-sgi-irix6): Remove.
9111 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9114 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9115 arguments to clear_bit_region_be.
9117 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9119 * gimplify.c: Include memmodel.h.
9121 2017-10-30 Martin Jambor <mjambor@suse.cz>
9123 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9124 condition whether loop should be executed at all.
9126 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9128 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9129 gimple folding of vec_madd() intrinsics.
9130 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9131 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9132 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9134 2017-10-30 Richard Biener <rguenther@suse.de>
9136 PR tree-optimization/82762
9138 2017-10-23 Richard Biener <rguenther@suse.de>
9140 PR tree-optimization/82129
9142 2017-08-01 Richard Biener <rguenther@suse.de>
9144 PR tree-optimization/81181
9145 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9146 (compute_antic): ... end of iteration here.
9148 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9150 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9151 -std=iso9899:2017 and -std=gnu17.
9152 * doc/standards.texi (C Language): Document C17 support.
9153 * doc/cpp.texi (Overview): Mention -std=c17.
9154 (Standard Predefined Macros): Document C11 and C17 values of
9155 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9156 * doc/extend.texi (Inline): Do not list individual options for
9157 standards newer than C99.
9158 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9160 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9163 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9165 * asan.c (asan_finish_file): Align asan globals array by shadow
9168 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9171 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9172 (struct store_immediate_info): Add bitregion_start and bitregion_end
9174 (store_immediate_info::store_immediate_info): Add brs and bre
9175 arguments and initialize bitregion_{start,end} from those.
9176 (struct merged_store_group): Add bitregion_start, bitregion_end,
9177 align_base and mask fields. Drop unnecessary struct keyword from
9178 struct store_immediate_info. Add do_merge method.
9179 (clear_bit_region_be): Use memset instead of loop storing zeros.
9180 (merged_store_group::do_merge): New method.
9181 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9182 stores as long as the surrounding bitregions have no gaps.
9183 (merged_store_group::merge_overlapping): Use do_merge.
9184 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9185 is byte aligned, rather than requiring that start and width are
9186 byte aligned. Drop unnecessary struct keyword from
9187 struct store_immediate_info. Allocate and populate also mask array.
9188 Make start of the arrays relative to bitregion_start rather than
9189 start and size them according to bitregion_{end,start} difference.
9190 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9191 struct store_immediate_info.
9192 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9193 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9194 struct keyword from struct store_immediate_info.
9195 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9196 between stores as long as the surrounding bitregions have no gaps.
9198 (struct split_store): Add orig non-static data member.
9199 (split_store::split_store): Initialize orig to false.
9200 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9201 if there is exactly a single original stmt. Change stmts argument
9202 to pointer from reference, if NULL, don't push anything to it. Add
9203 first argument, use it to optimize skipping over orig stmts that
9204 are known to be before bitpos already. Simplify.
9205 (split_group): Return unsigned int count how many stores are or
9206 would be needed rather than a bool. Add allow_unaligned argument.
9207 Change split_stores argument from reference to pointer, if NULL,
9208 only do a dry run computing how many stores would be produced.
9209 Rewritten algorithm to use both alignment and misalign if
9210 !allow_unaligned and handle bitfield stores with gaps.
9211 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9212 from bitregion_start instead of start. Compute allow_unaligned
9213 here, if true, do 2 split_group dry runs to compute which one
9214 produces fewer stores and prefer aligned if equal. Punt if
9215 new count is bigger or equal than original before emitting any
9216 statements, rather than during that. Remove no longer needed
9217 new_ssa_names tracking. Replace num_stmts with
9218 split_stores.length (). Use 32-bit stack allocated entries
9219 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9220 if possible. Handle bitfields with gaps.
9221 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9222 Compute bitregion_{start,end} for the stores and construct
9223 store_immediate_info with that. Formatting fixes.
9225 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9228 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9229 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9231 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9233 * gimplify.c: Include tm_p.h.
9235 * common.opt (gcoff): Re-add as ignored option.
9236 (gcoff1, gcoff2, gcoff3): Likewise.
9238 * Makefile.in (OBJS): Delete sdbout.o.
9239 (GTFILES): Delete $(srcdir)/sdbout.c.
9240 * debug.h: Delete sdb_debug_hooks.
9241 * final.c: Delete sdbout.h include.
9242 (final_scan_insn): Delete SDB_DEBUG check.
9243 (rest_of_clean_state): Likewise.
9244 * output.h: Delete sdb_begin_function_line.
9247 * toplev.c: Delete sdbout.h include.
9248 (process_options): Delete SDB_DEBUG check.
9249 * tree-core.h (tree_type_common): Delete pointer field of
9251 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9252 TYPE_SYMTAB_POINTER.
9253 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9254 (TYPE_SYMTAB_IS_POINTER): Delete.
9255 (TYPE_SYMTAB_IS_DIE): Renumber.
9256 * xcoffout.c: Refer to former sdbout.c file.
9257 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9259 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9260 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9261 (Debugging Options): Delete -gcoff.
9262 (-gstabs): Delete SDB reference.
9264 (-gcoff@var{level}): Delete.
9265 * doc/passes.texi (Debugging information output): Delete SDB and
9266 sdbout.c references.
9267 * doc/tm.texi: Regenerate.
9268 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9269 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9271 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9272 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9273 (SDB_DEBUGGING_INFO): Delete.
9274 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9275 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9276 * target.def (output_source_filename): Delete COFF reference.
9278 * common.opt (gcoff): Delete.
9279 (gxcoff+): Update Negative chain.
9280 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9282 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9283 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9284 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9286 (expand_function_start): Change sdb reference to past tense.
9287 (expand_function_end): Change sdb reference to past tense.
9288 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9289 * opts.c (debug_type_names): Delete coff entry.
9290 (common_handle_option): Delete OPT_gcoff case.
9291 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9293 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9294 * config/cris/cris.h: Delete SDB reference in comment.
9295 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9296 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9297 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9298 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9300 (ix86_expand_prologue): Likewise.
9301 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9302 * config/ia64/ia64.h: Likewise.
9303 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9304 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9306 * config/mmix/mmix.h: Likewise.
9307 * config/nds32/nds32.c: Likewise.
9308 * config/stormy/storym16.h: Likewise.
9309 * config/visium/visium.h: Likewise.
9310 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9312 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9314 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9315 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9316 FRAME_POINTER_REGNUM point at high end of local var area.
9318 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9320 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9321 Move comment around. Do not reset best_edge for a copiable
9322 destination if the copy would cause a partition change.
9323 (better_edge_p): Remove redundant check.
9325 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9327 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9329 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9331 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9332 for math functions that have _Float<N> and _Float<N>X variants.
9333 (mathfn_built_in_2): Add support for math functions that have
9334 _Float<N> and _Float<N>X variants.
9335 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9336 (expand_builtin_mathfn_ternary): Add support for fma with
9337 _Float<N> and _Float<N>X variants.
9338 (expand_builtin): Likewise.
9339 (fold_builtin_3): Likewise.
9340 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9341 create math function _Float<N> and _Float<N>X variants as external
9343 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9344 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9345 the __builtin_ prefix and if not strict ansi, without the prefix.
9346 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9347 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9348 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9349 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9350 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9351 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9352 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9353 function signatures for fma _Float<N> and _Float<N>X variants.
9354 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9355 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9356 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9357 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9358 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9359 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9360 * gencfn-macros.c (print_case_cfn): Add support for math functions
9361 that have _Float<N> and _Float<N>X variants.
9362 (print_define_operator_list): Likewise.
9363 (fltfn_suffixes): Likewise.
9365 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9366 for math functions that have _Float<N> and _Float<N>X variants.
9367 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9368 and _Float<N>X variants.
9369 (COPYSIGN): Likewise.
9372 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9373 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9375 (integer_valued_read_call_p): Likewise.
9376 * fold-const-call.c (fold_const_call_ss): Likewise.
9377 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9378 _Float<N> and _Float<N>X variants.
9379 (fold_const_call_ssss): Add support for fma _Float<N> and
9380 _Float<N>X variants.
9381 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9382 support for copysign and fma _Float<N> and _Float<N>X variants.
9383 (backprop::process_builtin_call_use): Likewise.
9384 * tree-call-cdce.c (can_test_argument_range); Add support for
9385 sqrt _Float<N> and _Float<N>X variants.
9386 (edom_only_function): Likewise.
9387 (get_no_error_domain): Likewise.
9388 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9389 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9390 copysign _Float<N> and _Float<N>X variants.
9391 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9392 handled by machine independent code.
9393 (FMAF128): Likewise.
9394 * doc/cpp.texi (Common Predefined Macros): Document defining
9395 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9396 fma _Float<N> and _Float<N>X variants.
9398 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9401 * config/i386/i386-modes.def (CCFPU): Remove definition.
9402 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9403 (ix86_cc_modes_compatible): Ditto.
9404 (ix86_expand_carry_flag_compare): Ditto.
9405 (ix86_expand_int_movcc): Ditto.
9406 (ix86_expand_int_addcc): Ditto.
9407 (ix86_reverse_condition): Ditto.
9408 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9409 Return true/false for unordered/ordered fp comparisons.
9410 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9411 (ix86_prepare_fp_compare_args): Update for rename.
9412 (ix86_expand_fp_compare): Update for rename. Generate unordered
9413 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9414 (ix86_expand_sse_compare_and_jump): Ditto.
9415 * config/i386/predicates.md (fcmov_comparison_operator):
9416 Remove CCFPU mode handling.
9417 (ix86_comparison_operator): Ditto.
9418 (ix86_carry_flag_operator): Ditto.
9419 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9420 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9421 (*cmpu<mode>_cc_i387): Ditto.
9422 (FPCMP): Remove mode iterator.
9423 (unord): Remove mode attribute.
9424 (unord_subst): New define_subst transformation
9425 (unord): New define_subst attribute.
9427 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9428 (*cmpi<unord>xf_i387): Ditto.
9429 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9430 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9431 using unord_subst transformation.
9432 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9433 (round_saeonly): Also handle CCFP mode.
9434 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9435 Remove UNSPEC_SAHF unspec handling.
9437 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9439 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9441 2017-10-27 Jeff Law <law@redhat.com>
9443 * gimple-ssa-sprintf.c: Include domwalk.h.
9444 (class sprintf_dom_walker): New class, derived from dom_walker.
9445 (sprintf_dom_walker::before_dom_children): New function.
9446 (struct call_info): Moved into sprintf_dom_walker class
9447 (compute_formath_length, handle_gimple_call): Likewise.
9448 (sprintf_length::execute): Call the dominator walker rather
9449 than walking the statements.
9451 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9452 gimple statement locations.
9453 (check_array_bounds): Corresponding changes. Get the statement's
9454 location directly from wi->stmt.
9456 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9459 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9461 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9463 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9464 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9466 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9469 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9470 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9471 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9472 maybe_get_pool_constant.
9473 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9476 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9478 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9479 2.26 caveat. Update gas and gld versions.
9480 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9483 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9485 * cgraph.h (set_malloc_flag): Declare.
9486 * cgraph.c (set_malloc_flag_1): New function.
9487 (set_malloc_flag): Likewise.
9488 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9489 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9491 (read_ipa_call_summary): Add support for reading is_return_callee.
9492 (write_ipa_call_summary): Stream is_return_callee.
9493 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9494 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9495 ipa-prop.h, ipa-fnsummary.h.
9496 (pure_const_names): Change to static.
9497 (malloc_state_e): Define.
9498 (malloc_state_names): Define.
9499 (funct_state_d): Add field malloc_state.
9500 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9501 (check_retval_uses): New function.
9502 (malloc_candidate_p): Likewise.
9503 (analyze_function): Add support for malloc attribute.
9504 (pure_const_write_summary): Stream malloc_state.
9505 (pure_const_read_summary): Add support for reading malloc_state.
9506 (dump_malloc_lattice): New function.
9507 (propagate_malloc): New function.
9508 (warn_function_malloc): New function.
9509 (ipa_pure_const::execute): Call propagate_malloc and
9510 ipa_free_fn_summary.
9511 (pass_local_pure_const::execute): Add support for malloc attribute.
9512 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9513 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9515 2017-10-27 Martin Liska <mliska@suse.cz>
9517 PR gcov-profile/82457
9518 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9519 for fork-like functions to be properly instrumented.
9521 2017-10-27 Richard Biener <rguenther@suse.de>
9524 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9525 info when we redirected EH.
9527 2017-10-26 Michael Collison <michael.collison@arm.com>
9529 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9531 (<optab>_trunchf<GPI:mode>2: New pattern.
9532 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9533 * config/aarch64/iterators.md (wv): New mode attribute.
9534 (vf, VF): New mode attributes.
9535 (vgp, VGP): New mode attributes.
9536 (s): Update attribute with SImode and DImode prefixes.
9538 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9540 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9541 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9542 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9543 (nios2_option_overide): Initialize it. Don't allow R0-relative
9544 addressing with PIC.
9545 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9546 (nios2_symbolic_constant_p): Likewise.
9547 (nios2_legitimate_address_p): Likewise.
9548 (nios2_r0rel_section_name_p): New.
9549 (nios2_symbol_ref_in_r0rel_data_p): New.
9550 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9551 (r0rel_constant_p): New.
9552 (nios2_print_operand_address): Handle r0rel_constant_p.
9553 (nios2_cdx_narrow_form_p): Likewise.
9554 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9555 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9556 (Nios II Options): Document -mr0rel-sec.
9558 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9560 * config/nios2/nios2.c: Include xregex.h.
9561 (nios2_gprel_sec_regex): New.
9562 (nios2_option_overide): Initialize it. Don't allow GP-relative
9563 addressing with PIC.
9564 (nios2_small_section_name_p): Check for regex match.
9565 * config/nios2/nios2.opt (mgprel-sec=): New option.
9566 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9567 (Nios II Options): Document -mgprel-sec.
9569 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9571 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9573 2017-10-26 Tom de Vries <tom@codesourcery.com>
9575 PR tree-optimization/82707
9576 * gimple.c (gimple_copy): Fix unsharing of
9577 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9579 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9581 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9582 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9583 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9584 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9585 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9586 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9587 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9588 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9589 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9590 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9591 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9592 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9593 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9594 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9595 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9596 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9597 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9599 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9601 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9603 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9604 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9605 warning to rs6000.c. Remove the Undocumented flag, since it has
9607 (-mabi=ibmlongdouble): Likewise.
9608 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9609 already set, set the default format for long double.
9610 (rs6000_debug_reg_global): Print whether long double is IBM or
9612 (rs6000_option_override_internal): Rework setting long double
9613 format. Only warn if the user is changing the long double default
9614 and they did not use -Wno-psabi.
9615 * doc/invoke.texi (PowerPC options): Update the documentation for
9616 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9618 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9619 Alan Hayward <alan.hayward@arm.com>
9620 David Sherwood <david.sherwood@arm.com>
9622 * rtl.h (wider_subreg_mode): New function.
9623 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9624 rather than an unsigned int *.
9625 * ira-color.c (regno_max_ref_width): Replace with...
9626 (regno_max_ref_mode): ...this new variable.
9627 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9628 Use wider_subreg_mode.
9629 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9630 rather than an unsigned int *.
9631 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9632 (process_alt_operands): Likewise.
9633 (invariant_p): Likewise.
9634 * lra-spills.c (assign_mem_slot): Likewise.
9635 (add_pseudo_to_slot): Likewise.
9636 * lra.c (collect_non_operand_hard_regs): Likewise.
9637 (add_regs_to_insn_regno_info): Likewise.
9638 * reload1.c (regno_max_ref_width): Replace with...
9639 (regno_max_ref_mode): ...this new variable.
9640 (reload): Update accordingly. Update call to
9641 ira_sort_regnos_for_alter_reg.
9642 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9643 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9644 (scan_paradoxical_subregs): Likewise.
9646 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9648 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9649 (aarch64_frame): Add emit_frame_chain boolean.
9650 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9651 Move eh_return case to aarch64_layout_frame.
9652 (aarch64_layout_frame): Initialize emit_frame_chain.
9653 (aarch64_expand_prologue): Use emit_frame_chain.
9655 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9657 * config/aarch64/aarch64.c (aarch64_layout_frame):
9658 Ensure LR is always stored at the bottom of the callee-saves.
9659 Remove rarely used frame layout which saves callee-saves at top of
9660 frame, so the store of LR can be used as a valid probe in all cases.
9662 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9664 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9665 Improve unaligned TImode/TFmode base/offset split.
9667 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9668 Alan Hayward <alan.hayward@arm.com>
9669 David Sherwood <david.sherwood@arm.com>
9671 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9672 * combine.c (find_single_use_1): Likewise.
9673 (expand_field_assignment): Likewise.
9674 (move_deaths): Likewise.
9675 * lra-constraints.c (simplify_operand_subreg): Likewise.
9676 (curr_insn_transform): Likewise.
9677 * lra.c (collect_non_operand_hard_regs): Likewise.
9678 (add_regs_to_insn_regno_info): Likewise.
9679 * rtlanal.c (reg_referenced_p): Likewise.
9680 (covers_regno_no_parallel_p): Likewise.
9682 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9684 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9685 Don't print any bits outside the precision of the value.
9686 * wide-int.cc (test_printing): Add some new tests.
9688 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9690 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9691 supports -xbrace_comment option.
9692 * configure: Regenerate.
9693 * config.in: Regenerate.
9694 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9695 (ASM_CPU_SPEC): Use it.
9697 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9699 * target.def (static_rtx_alignment): New hook.
9700 * targhooks.h (default_static_rtx_alignment): Declare.
9701 * targhooks.c (default_static_rtx_alignment): New function.
9702 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9703 * doc/tm.texi: Regenerate.
9704 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9705 instead of targetm.constant_alignment. Remove call to
9707 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9708 (cris_preferred_mininum_alignment): New function, split out from...
9709 (cris_constant_alignment): ...here.
9710 (cris_static_rtx_alignment): New function.
9711 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9713 (ix86_constant_alignment): ...here.
9714 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9715 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9716 (mmix_static_rtx_alignment): New function.
9717 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9718 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9720 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9723 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9724 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9726 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9728 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9729 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9730 mark as a sign-extending load.
9731 (local_pic_load_u): Define.
9733 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9736 * fold-const.c (operand_equal_for_comparison_p): Also return true
9737 if ARG0 is a simple variant of ARG1 with narrower precision.
9738 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9740 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9742 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9744 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9745 scatter_static, scatter_per_elt.
9746 * x86-tune-costs.h: Add new cost entries.
9748 2017-10-25 Richard Biener <rguenther@suse.de>
9750 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9751 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9752 class pass_fre): Move to ...
9753 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9754 class pass_fre): ... here and adjust for statistics.
9756 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9759 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9760 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9761 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9762 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9765 2017-10-25 Richard Biener <rguenther@suse.de>
9767 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9768 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9769 eliminate_push_avail, eliminate_insert): Move inside...
9770 (class eliminate_dom_walker): ... this class in preparation
9772 (fini_eliminate): Remove by merging with ...
9773 (eliminate): ... this function. Adjust for class changes.
9774 (pass_pre::execute): Remove fini_eliminate call.
9775 (pass_fre::execute): Likewise.
9777 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9780 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9781 (VPERMI2, VPERMI2I): New mode iterators.
9782 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9783 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9785 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9786 mode iterator. Remove 3 old define_insn patterns.
9787 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9788 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9789 VPERMI2 mode iterator, remove the other two expanders.
9790 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9791 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9792 instructions, remove the other two patterns.
9793 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9794 mode iterator, remove the other two patterns.
9795 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9796 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9797 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9798 and adjust argument order accordingly.
9799 (ix86_expand_vec_perm): Adjust caller.
9800 (expand_vec_perm_1): Likewise.
9801 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9802 (expand_vec_perm_vpermt2_vpshub2): ... this.
9803 (ix86_expand_vec_perm_const_1): Adjust caller.
9804 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9807 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9808 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9809 (vec_shl_<mode>): Remove unused expander.
9810 (avx512bw_<shift_insn><mode>3): New define_insn.
9811 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9812 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9814 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9817 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9820 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9822 PR rtl-optimization/82396
9823 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9824 (autopref_multipass_init): Simplify initialization.
9825 (autopref_rank_data): Simplify sort order.
9826 * gcc/sched-int.h (autopref_multipass_data_): Remove
9827 multi_mem_insn_p, min_offset and max_offset.
9829 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9832 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9833 Check special value of flag_omit_frame_pointer.
9834 (aarch64_can_eliminate): Likewise.
9835 (aarch64_override_options_after_change_1): Simplify handling of
9836 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9838 2017-10-24 Richard Biener <rguenther@suse.de>
9840 PR tree-optimization/82697
9841 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9842 zero for conditional load and unconditional store.
9844 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9846 * doc/install.texi: Document bootstrap-cet.
9848 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9851 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9852 ENDBR instruction at function entrance if function is only
9855 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9858 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9859 patterns to better describe from which operation the CF is computed.
9860 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9861 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9862 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9863 is 0, use _0 suffixed expanders instead of emitting a comparison
9866 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9868 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9869 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9871 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9874 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9875 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9876 * loop-iv.c (iv_get_reaching_def): Likewise.
9877 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9878 variable is promoted and the partition contains undefined values.
9880 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9882 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9884 (nios2_address_cost): Define.
9885 (nios2_legitimize_address): Recognize (exp + constant) directly.
9886 (TARGET_ADDRESS_COST): Define.
9888 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9890 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9891 (nios2_symbolic_memory_operand_p): Declare.
9892 (nios2_split_large_constant): Declare.
9893 (nios2_split_symbolic_memory_operand): Declare.
9894 * config/nios2/nios2.c: Adjust includes.
9895 (nios2_symbolic_constant_allowed): New.
9896 (nios2_symbolic_constant_p): New.
9897 (nios2_plus_symbolic_constant_p): New.
9898 (nios2_valid_addr_expr_p): Recognize addresses involving
9900 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9901 (nios2_symbolic_memory_operand_p): New.
9902 (nios2_large_constant_p): New.
9903 (nios2_split_large_constant): New.
9904 (nios2_split_plus_large_constant): New.
9905 (nios2_split_symbolic_memory_operand): New.
9906 (nios2_legitimize_address): Code refactoring. Handle addresses
9907 involving symbolic constants.
9908 (nios2_emit_move_sequence): Likewise.
9909 (nios2_print_operand): Improve error output.
9910 (nios2_print_operand_address): Handle LO_SUM.
9911 (nios2_cdx_narrow_form_p): Likewise.
9912 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9913 operands involving symbolic constants.
9914 (movhi_internal, movsi_internal): Likewise.
9915 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9916 (extendhisi2, extendqi<mode>2): Likewise.
9918 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9920 * tree-pass.h (PROP_rtl_split_insns): Define.
9921 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9923 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9925 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9927 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9930 * target.def (const_not_ok_for_debug_p): Default to
9931 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9932 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9933 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9934 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9935 which targetm.const_not_ok_for_debug_p returned true.
9936 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9938 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9940 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9941 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9942 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9944 (ix86_const_not_ok_for_debug_p): New function.
9945 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9946 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9948 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9951 * system.h: Conditionally include "unique-ptr.h" if
9952 INCLUDE_UNIQUE_PTR is defined.
9953 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9954 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9956 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9958 * config/rl78/rl78.md: New define_expand "subdi3".
9960 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9963 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9964 DF_REF_INSN if DF_REF_INSN_INFO is false.
9966 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9968 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9969 xmm_move instead of sse_move.
9970 (sse_store_index): New function.
9971 (ix86_register_move_cost): Be more sensible about mismatch stall;
9972 model AVX moves correctly; make difference between sse->integer and
9974 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9975 moves; make difference between SSE and AVX.
9976 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9977 and zmm_move. Increase size of sse load and store tables;
9978 add unaligned load and store tables; add ssemmx_to_integer.
9979 * x86-tune-costs.h: Update all entries according to real
9980 move latencies from Agner Fog's manual and chip documentation.
9982 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9985 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9986 * config/i386/constraints.md (Wf): New constraint.
9987 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9988 (cmp<dwi>_doubleword): Removed.
9989 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9990 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9991 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9992 expand with cmp<dwi>_doubleword. For LTU and GEU use
9993 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9995 * common.opt (gcolumn-info): Enable by default.
9996 * doc/invoke.texi (gcolumn-info): Document new default.
9998 2017-10-23 Richard Biener <rguenther@suse.de>
10000 PR tree-optimization/82672
10001 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10002 Fold the stmt if we propagated into it.
10004 2017-10-23 Richard Biener <rguenther@suse.de>
10006 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10007 (bitmap_remove_expr_from_set): ... this. All callers call this
10008 for non-constant values.
10009 (bitmap_set_subtract): Rename to...
10010 (bitmap_set_subtract_expressions): ... this. Adjust and
10012 (bitmap_set_contains_value): Remove superfluous check.
10013 (bitmap_set_replace_value): Inline into single caller ...
10014 (bitmap_value_replace_in_set): ... here and simplify.
10015 (dependent_clean): Merge into ...
10016 (clean): ... this using an overload. Adjust.
10017 (prune_clobbered_mems): Adjust.
10018 (compute_antic_aux): Likewise.
10019 (compute_partial_antic_aux): Likewise.
10021 2017-10-23 Richard Biener <rguenther@suse.de>
10023 PR tree-optimization/82129
10025 2017-08-01 Richard Biener <rguenther@suse.de>
10027 PR tree-optimization/81181
10028 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10029 (compute_antic): ... end of iteration here.
10031 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10033 * target.def (starting_frame_offset): New hook.
10034 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10035 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10036 * doc/tm.texi.in: Regenerate.
10037 * hooks.h (hook_hwi_void_0): Declare.
10038 * hooks.c (hook_hwi_void_0): New function.
10039 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10040 STARTING_FRAME_OFFSET.
10041 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10042 * reload1.c (reload): Likewise.
10043 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10044 instead of STARTING_FRAME_OFFSET.
10045 * function.c (try_fit_stack_local): Likewise.
10046 (assign_stack_local_1): Likewise
10047 (instantiate_virtual_regs): Likewise.
10048 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10049 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10050 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10051 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10052 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10053 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10054 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10055 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10056 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10057 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10058 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10059 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10060 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10061 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10062 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10063 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10064 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10065 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10066 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10067 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10068 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10069 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10070 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10071 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10072 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10073 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10074 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10075 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10076 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10077 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10078 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10079 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10080 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10081 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10082 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10083 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10084 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10085 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10086 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10087 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10088 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10089 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10090 return a HOST_WIDE_INT.
10091 (avr_builtin_setjmp_frame_value): Use it instead of
10092 STARTING_FRAME_OFFSET.
10093 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10094 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10095 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10097 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10098 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10099 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10100 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10101 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10102 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10103 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10104 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10105 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10106 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10107 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10108 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10110 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10111 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10112 * config/mips/mips.c (mips_compute_frame_info): Refer to
10113 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10114 (mips_starting_frame_offset): New function.
10115 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10116 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10117 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10118 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10119 and return a HOST_WIDE_INT.
10120 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10121 (mmix_initial_elimination_offset): Refer to
10122 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10123 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10124 * config/pa/pa.c (pa_starting_frame_offset): New function.
10125 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10126 (pa_expand_prologue): Likewise.
10127 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10128 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10129 !FRAME_GROWS_DOWNWARD handling to...
10130 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10131 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10132 !FRAME_GROWS_DOWNWARD handling to...
10133 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10134 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10135 !FRAME_GROWS_DOWNWARD handling to...
10136 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10137 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10139 (rs6000_starting_frame_offset): New function.
10140 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10141 !FRAME_GROWS_DOWNWARD handling to...
10142 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10143 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10144 !FRAME_GROWS_DOWNWARD handling to...
10145 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10146 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10147 !FRAME_GROWS_DOWNWARD handling to...
10148 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10149 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10150 (rs6000_starting_frame_offset): New function.
10151 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10152 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10153 * config/vax/vax.c (vax_starting_frame_offset): New function.
10154 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10155 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10156 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10157 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10158 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10159 * system.h (STARTING_FRAME_OFFSET): Poison.
10161 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10163 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10164 SCALAR_TYPE_MODE instead of TYPE_MODE.
10166 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10167 Alan Hayward <alan.hayward@arm.com>
10168 David Sherwood <david.sherwood@arm.com>
10170 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10172 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10173 Alan Hayward <alan.hayward@arm.com>
10174 David Sherwood <david.sherwood@arm.com>
10176 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10178 2017-10-23 Richard Biener <rguenther@suse.de>
10180 PR tree-optimization/82129
10181 * tree-ssa-pre.c (bitmap_set_and): Remove.
10182 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10183 canonicalizing expressions in the set to those with lowest
10184 ID rather than taking that from the first edge.
10186 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10188 * combine.c (rtx_equal_for_field_assignment_p): Use
10189 byte_lowpart_offset.
10191 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10192 Alan Hayward <alan.hayward@arm.com>
10193 David Sherwood <david.sherwood@arm.com>
10195 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10196 to a SUBREG_PROMOTED_VAR.
10198 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10199 Alan Hayward <alan.hayward@arm.com>
10200 David Sherwood <david.sherwood@arm.com>
10202 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10203 (expand_debug_source_expr): Likewise.
10204 * combine.c (combine_simplify_rtx): Likewise.
10205 * cse.c (fold_rtx): Likewise.
10206 * optabs.c (expand_float): Likewise.
10207 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10208 (simplify_binary_operation_1): Likewise.
10210 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10211 Alan Hayward <alan.hayward@arm.com>
10212 David Sherwood <david.sherwood@arm.com>
10214 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10215 (record_promoted_value): Likewise.
10216 * expr.c (expand_expr_real_2): Likewise.
10217 * ree.c (update_reg_equal_equiv_notes): Likewise.
10218 (combine_set_extension): Likewise.
10219 * rtlanal.c (low_bitmask_len): Likewise.
10220 * simplify-rtx.c (neg_const_int): Likewise.
10221 (simplify_binary_operation_1): Likewise.
10223 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10224 Alan Hayward <alan.hayward@arm.com>
10225 David Sherwood <david.sherwood@arm.com>
10227 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10228 * regcprop.c (maybe_mode_change): Likewise.
10229 * reload1.c (alter_reg): Likewise.
10231 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10233 * inchash.h (inchash::hash::add_wide_int): New function.
10234 * lto-streamer-out.c (hash_tree): Use it.
10236 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10238 * inchash.h (inchash::hash::add_wide_int): Rename to...
10239 (inchash::hash::add_hwi): ...this.
10240 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10241 (polymorphic_call_target_hasher::hash): Likewise.
10242 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10243 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10244 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10245 * lto-streamer-out.c (hash_tree): Likewise.
10246 * optc-save-gen.awk: Likewise.
10247 * tree.c (add_expr): Likewise.
10249 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10252 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10253 for ordered inequality comparisons even with TARGET_IEEE_FP.
10255 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10258 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10259 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10260 Expand with cmp<dwi>_doubleword.
10262 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10264 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10265 List CET intrinsics.
10266 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10267 specific to -fcf-protection option.
10269 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10271 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10272 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10273 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10274 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10275 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10276 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10277 (extra_objs): Add cet.o for Linux/x86 targets.
10278 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10279 * config/i386/cet.c: New file.
10280 * config/i386/cetintrin.h: Likewise.
10281 * config/i386/t-cet: Likewise.
10282 * config/i386/cpuid.h (bit_SHSTK): New.
10283 (bit_IBT): Likewise.
10284 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10285 pass IBT and SHSTK bits.
10286 * config/i386/i386-builtin-types.def
10287 (VOID_FTYPE_UNSIGNED_PVOID): New.
10288 (VOID_FTYPE_UINT64_PVOID): Likewise.
10289 * config/i386/i386-builtin.def: Add CET intrinsics.
10290 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10291 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10292 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10293 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10295 * config/i386/i386.c (rest_of_insert_endbranch): New.
10296 (pass_data_insert_endbranch): Likewise.
10297 (pass_insert_endbranch): Likewise.
10298 (make_pass_insert_endbranch): Likewise.
10299 (ix86_notrack_prefixed_insn_p): Likewise.
10300 (ix86_target_string): Add -mibt, -mshstk flags.
10301 (ix86_option_override_internal): Add flag_cf_protection
10303 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10304 (ix86_print_operand): Add 'notrack' prefix output.
10305 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10306 (ix86_expand_builtin): Expand CET intrinsics.
10307 (x86_output_mi_thunk): Add 'endbranch' instruction.
10308 * config/i386/i386.h (TARGET_IBT): New.
10309 (TARGET_IBT_P): Likewise.
10310 (TARGET_SHSTK): Likewise.
10311 (TARGET_SHSTK_P): Likewise.
10312 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10313 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10314 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10315 (builtin_setjmp_setup): New pattern.
10316 (builtin_longjmp): Likewise.
10317 (rdssp<mode>): Likewise.
10318 (incssp<mode>): Likewise.
10319 (saveprevssp): Likewise.
10320 (rstorssp): Likewise.
10321 (wrss<mode>): Likewise.
10322 (wruss<mode>): Likewise.
10323 (setssbsy): Likewise.
10324 (clrssbsy): Likewise.
10325 (nop_endbr): Likewise.
10326 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10328 * config/i386/immintrin.h: Include <cetintrin.h>.
10329 * config/i386/linux-common.h
10330 (file_end_indicate_exec_stack_and_cet): New prototype.
10331 (TARGET_ASM_FILE_END): New.
10333 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10335 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10336 latencies instead of having separate table; make difference between
10337 integer and float costs.
10338 * i386.h (processor_costs): Remove scalar_stmt_cost,
10339 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10340 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10342 * x86-tune-costs.h: Remove entries which has been removed in
10343 procesor_costs from all tables; make cond_taken_branch_cost
10344 and cond_not_taken_branch_cost COST_N_INSNS based.
10346 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10348 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10350 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10352 * config/i386/i386.md (isa): Remove fma_avx512f.
10353 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10354 <avx512>_fmadd_<mode>_mask3<round_name>,
10355 <avx512>_fmsub_<mode>_mask<round_name>,
10356 <avx512>_fmsub_<mode>_mask3<round_name>,
10357 <avx512>_fnmadd_<mode>_mask<round_name>,
10358 <avx512>_fnmadd_<mode>_mask3<round_name>,
10359 <avx512>_fnmsub_<mode>_mask<round_name>,
10360 <avx512>_fnmsub_<mode>_mask3<round_name>,
10361 <avx512>_fmaddsub_<mode>_mask<round_name>,
10362 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10363 <avx512>_fmsubadd_<mode>_mask<round_name>,
10364 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10365 (*vec_widen_umult_even_v16si<mask_name>,
10366 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10367 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10369 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10371 * extend.texi: Add 'nocf_check' documentation.
10372 * gimple.texi: Add second parameter to
10373 gimple_build_call_from_tree.
10374 * invoke.texi: Add -fcf-protection documentation.
10375 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10377 2017-10-20 Richard Biener <rguenther@suse.de>
10379 PR tree-optimization/82473
10380 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10381 the largest input type.
10383 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10385 * c-attribs.c (handle_nocf_check_attribute): New function.
10386 (c_common_attribute_table): Add 'nocf_check' handling.
10387 * gimple-parser.c: Add second argument NULL to
10388 gimple_build_call_from_tree.
10389 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10390 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10392 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10393 * common.opt: Add fcf-protection flag.
10394 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10395 * flag-types.h: Add enum cf_protection_level.
10396 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10397 Add 'nocf_check' attribute propagation to gimple call.
10398 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10399 (gimple_build_call_from_tree): Update prototype.
10400 (gimple_call_nocf_check_p): New function.
10401 (gimple_call_set_nocf_check): Likewise.
10402 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10403 * ipa-icf.c: Add nocf_check attribute in statement hash.
10404 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10405 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10406 * toplev.c (process_options): Add flag_cf_protection handling.
10408 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10410 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10412 2017-10-20 Richard Biener <rguenther@suse.de>
10414 PR tree-optimization/82603
10415 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10416 remove false predicated stores.
10418 2017-10-20 Richard Biener <rguenther@suse.de>
10420 * graphite-isl-ast-to-gimple.c
10421 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10422 Remove return value and simplify, dump copied stmt after lhs
10424 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10425 Reduce dump verbosity.
10426 (gsi_insert_earliest): Likewise.
10427 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10428 * graphite.c (print_global_statistics): Adjust dumping.
10429 (print_graphite_scop_statistics): Likewise.
10430 (print_graphite_statistics): Do not dump loops here.
10431 (graphite_transform_loops): But here.
10433 2017-10-20 Nicolas Roche <roche@adacore.com>
10435 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10436 * configure: Regenerate.
10438 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10441 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10442 functions when optimizing replace GIMPLE_RETURN stmts with
10443 calls to __builtin_unreachable ().
10446 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10447 for -fsanitize=thread link of executables.
10448 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10449 link of executables.
10452 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10453 New mode iterators.
10454 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10455 define_insns for logical vector shifts to use VI248_AVX512BW
10456 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10457 condition, useless isa and prefix attributes. Change the first
10458 2 of these define_insns to ...
10459 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10460 define_insn for avx512vl.
10461 (<shift_insn><mode>3): ... and this, new define_insn without
10462 masking for non-avx512vl.
10465 * config/i386/sse.md (*andnot<mode>3,
10466 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10467 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10468 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10469 not applied use empty suffix even for TARGET_AVX512VL.
10470 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10471 is applied, supply evex,evex or evex,evex,evex instead of just
10474 2017-10-20 Julia Koval <julia.koval@intel.com>
10476 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10477 (OPTION_MASK_ISA_GFNI_UNSET): New.
10478 (ix86_handle_option): Handle OPT_mgfni.
10479 * config/i386/cpuid.h (bit_GFNI): New.
10480 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10481 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10482 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10483 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10484 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10485 * config/i386/i386.opt: Add mgfni.
10487 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10489 * config/msp430/msp430.c (msp430_option_override): Disable
10490 -fdelete-null-pointer-checks.
10491 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10493 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10495 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10496 of x87 and SSE instructions.
10498 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10500 * asan.c (create_cond_insert_point): Do not update edge count.
10501 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10502 (afdo_propagate_circuit): Likewise.
10503 (afdo_calculate_branch_prob): Likewise.
10504 (afdo_annotate_cfg): Likewise.
10505 * basic-block.h (struct edge_def): Remove count.
10506 (edge_def::count): New accessor.
10507 * bb-reorder.c (rotate_loop): Update.
10508 (find_traces_1_round): Update.
10509 (connect_traces): Update.
10510 (sanitize_hot_paths): Update.
10511 * cfg.c (unchecked_make_edge): Update.
10512 (make_single_succ_edge): Update.
10513 (check_bb_profile): Update.
10514 (dump_edge_info): Update.
10515 (update_bb_profile_for_threading): Update.
10516 (scale_bbs_frequencies_int): Update.
10517 (scale_bbs_frequencies_gcov_type): Update.
10518 (scale_bbs_frequencies_profile_count): Update.
10519 (scale_bbs_frequencies): Update.
10520 * cfganal.c (connect_infinite_loops_to_exit): Update.
10521 * cfgbuild.c (compute_outgoing_frequencies): Update.
10522 (find_many_sub_basic_blocks): Update.
10523 * cfgcleanup.c (try_forward_edges): Update.
10524 (try_crossjump_to_edge): Update
10525 * cfgexpand.c (expand_gimple_cond): Update
10526 (expand_gimple_tailcall): Update
10527 (construct_exit_block): Update
10528 * cfghooks.c (verify_flow_info): Update
10529 (redirect_edge_succ_nodup): Update
10530 (split_edge): Update
10531 (make_forwarder_block): Update
10532 (duplicate_block): Update
10533 (account_profile_record): Update
10534 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10535 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10536 * cfgloopmanip.c (scale_loop_profile): Update.
10538 (lv_adjust_loop_entry_edge): Update.
10539 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10540 (force_nonfallthru_and_redirect): Update.
10541 (purge_dead_edges): Update.
10542 (rtl_flow_call_edges_add): Update.
10543 * cgraphunit.c (init_lowered_empty_function): Update.
10544 (cgraph_node::expand_thunk): Update.
10545 * gimple-pretty-print.c (dump_probability): Update.
10546 (dump_edge_probability): Update.
10547 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10548 * haifa-sched.c (sched_create_recovery_edges): Update.
10549 * hsa-gen.c (convert_switch_statements): Update.
10550 * ifcvt.c (dead_or_predicable): Update.
10551 * ipa-inline-transform.c (inline_transform): Update.
10552 * ipa-split.c (split_function): Update.
10553 * ipa-utils.c (ipa_merge_profiles): Update.
10554 * loop-doloop.c (add_test): Update.
10555 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10556 * lto-streamer-in.c (input_cfg): Update.
10557 (input_function): Update.
10558 * lto-streamer-out.c (output_cfg): Update.
10559 * modulo-sched.c (sms_schedule): Update.
10560 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10561 * predict.c (maybe_hot_edge_p): Update.
10562 (unlikely_executed_edge_p): Update.
10563 (probably_never_executed_edge_p): Update.
10564 (dump_prediction): Update.
10565 (drop_profile): Update.
10566 (propagate_unlikely_bbs_forward): Update.
10567 (determine_unlikely_bbs): Update.
10568 (force_edge_cold): Update.
10569 * profile.c (compute_branch_probabilities): Update.
10570 * reg-stack.c (better_edge): Update.
10571 * shrink-wrap.c (handle_simple_exit): Update.
10572 * tracer.c (better_p): Update.
10573 * trans-mem.c (expand_transaction): Update.
10574 (split_bb_make_tm_edge): Update.
10575 * tree-call-cdce.c: Update.
10576 * tree-cfg.c (gimple_find_sub_bbs): Update.
10577 (gimple_split_edge): Update.
10578 (gimple_duplicate_sese_region): Update.
10579 (gimple_duplicate_sese_tail): Update.
10580 (gimple_flow_call_edges_add): Update.
10581 (insert_cond_bb): Update.
10582 (execute_fixup_cfg): Update.
10583 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10584 * tree-complex.c (expand_complex_div_wide): Update.
10585 * tree-eh.c (lower_resx): Update.
10586 (unsplit_eh): Update.
10587 (cleanup_empty_eh_move_lp): Update.
10588 * tree-inline.c (copy_edges_for_bb): Update.
10589 (freqs_to_counts): Update.
10590 (copy_cfg_body): Update.
10591 * tree-ssa-dce.c (remove_dead_stmt): Update.
10592 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10593 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10594 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10595 (unloop_loops): Update.
10596 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10597 * tree-ssa-loop-split.c (connect_loops): Update.
10598 (split_loop): Update.
10599 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10600 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10601 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10602 * tree-ssa-reassoc.c (branch_fixup): Update.
10603 * tree-ssa-tail-merge.c (replace_block_by): Update.
10604 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10605 (compute_path_counts): Update.
10606 (update_profile): Update.
10607 (recompute_probabilities): Update.
10608 (update_joiner_offpath_counts): Update.
10609 (estimated_freqs_path): Update.
10610 (freqs_to_counts_path): Update.
10611 (clear_counts_path): Update.
10612 (ssa_fix_duplicate_block_edges): Update.
10613 (duplicate_thread_path): Update.
10614 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10615 (case_bit_test_cmp): Update.
10616 (collect_switch_conv_info): Update.
10617 (gen_inbound_check): Update.
10618 (do_jump_if_equal): Update.
10619 (emit_cmp_and_jump_insns): Update.
10620 * tree-tailcall.c (decrease_profile): Update.
10621 (eliminate_tail_call): Update.
10622 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10623 (vect_do_peeling): Update.
10624 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10625 * ubsan.c (ubsan_expand_null_ifn): Update.
10626 (ubsan_expand_ptr_ifn): Update.
10627 * value-prof.c (gimple_divmod_fixed_value): Update.
10628 (gimple_mod_pow2): Update.
10629 (gimple_mod_subtract): Update.
10630 (gimple_ic): Update.
10631 (gimple_stringop_fixed_value): Update.
10633 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10636 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10638 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10640 PR rtl-optimization/82395
10641 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10642 based on non_spilled_static_chain_regno_p.
10644 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10646 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10647 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10648 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10650 2017-10-19 Martin Sebor <msebor@redhat.com>
10652 PR tree-optimization/82596
10653 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10655 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10657 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10658 (get_mem_refs_of_builtin_call): Likewise.
10659 * builtins.c (expand_builtin_apply): Adjust call to
10660 allocate_dynamic_stack_space.
10661 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10662 the third argument to allocate_dynamic_stack_space, otherwise -1.
10663 (expand_builtin): Deal with all alloca variants.
10664 (is_inexpensive_builtin): Likewise.
10665 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10666 * calls.c (special_function_p): Deal with all alloca variants.
10667 (initialize_argument_information): Adjust call to
10668 allocate_dynamic_stack_space.
10669 (expand_call): Likewise.
10670 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10671 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10672 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10673 use it for the stack usage computation.
10674 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10675 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10676 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10677 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10678 (in_loop_p): Remove first argument and useless check.
10679 (pass_walloca::execute): Remove useless test and adjust call to above.
10680 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10681 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10682 (gimplify_call_expr): Deal with all alloca variants.
10683 * hsa-gen.c (gen_hsa_alloca): Likewise.
10684 (gen_hsa_insns_for_call): Likewise.
10685 * ipa-pure-const.c (special_builtin_state): Likewise.
10686 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10687 * tree-object-size.c (alloc_object_size): Likewise.
10688 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10689 (call_may_clobber_ref_p_1): Likewise.
10690 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10691 (ccp_fold_stmt): Likewise.
10692 (optimize_stack_restore): Likewise.
10693 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10694 (mark_all_reaching_defs_necessary_1): Likewise.
10695 (propagate_necessity): Likewise.
10696 (eliminate_unnecessary_stmts): Likewise.
10697 * tree.c (build_common_builtin_nodes): Build
10698 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10699 (build_alloca_call_expr): New function.
10700 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10701 (CASE_BUILT_IN_ALLOCA): Likewise.
10702 (build_alloca_call_expr): Declare.
10703 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10705 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10708 * dwarf2out.c (new_die_raw): New static inline function.
10709 (new_die): Use it to create the DIE.
10710 (add_AT_external_die_ref): Likewise.
10711 (clone_die): Likewise.
10712 (clone_as_declaration): Likewise.
10713 (dwarf2out_vms_debug_main_pointer): Likewise.
10714 (base_type_die): Likewise. Remove early return for corner cases.
10715 Do not call add_pubtype on the DIE here.
10716 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10717 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10718 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10719 native order exists for base types, attach the DIE manually and call
10720 add_pubtype on it. Do not equate a reverse order DIE to the type.
10722 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10724 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10725 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10727 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10729 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10731 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10732 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10733 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10734 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10735 * opts.c (sanitizer_opts): Add builtin.
10736 * ubsan.c (instrument_builtin): New function.
10737 (pass_ubsan::execute): Call it.
10738 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10739 * doc/invoke.texi: Document -fsanitize=builtin.
10741 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10742 builtins, store max (log2 (align), 0) into uchar field instead of
10743 align into uptr field.
10744 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10745 store uchar 0 field instead of uptr 0 field.
10746 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10747 instead of passing one address of struct with 2 locations pass
10748 two addresses of structs with 1 location each.
10749 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10750 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10751 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10752 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10753 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10754 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10755 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10756 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10758 2017-10-19 Martin Liska <mliska@suse.cz>
10761 * file-find.c (remove_prefix): Remove.
10762 * file-find.h (remove_prefix): Likewise.
10763 * gcc-ar.c: Remove smartness of lookup.
10765 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10767 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10768 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10769 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10771 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10774 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10775 (setcc + and to xor + setcc): New peephole2.
10777 2017-10-19 Tom de Vries <tom@codesourcery.com>
10779 * doc/sourcebuild.texi (Test Directives, Variants of
10780 dg-require-support): Add dg-require-stack-size.
10782 2017-10-19 Martin Liska <mliska@suse.cz>
10785 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10786 that have a large alignment.
10787 (gimplify_target_expr): Likewise.
10789 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10791 PR rtl-optimization/82602
10792 * ira.c (rtx_moveable_p): Return false for volatile asm.
10794 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10797 * config/i386/i386-modes.def (CCGZ): New CC mode.
10798 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10799 * config/i386/predicates.md (ix86_comparison_operator):
10801 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10802 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10803 with double-word subtraction.
10804 (put_condition_code): Handle CCGZmode.
10806 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10808 * wide-int.cc (debug (const wide_int &)): New.
10809 (debug (const wide_int *)): New.
10810 (debug (const widest_int &)): New.
10811 (debug (const widest_int *)): New.
10813 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10815 PR middle-end/82556
10816 * lra-constraints.c (curr_insn_transform): Use non-input operand
10817 instead of output one for matched reload.
10819 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10821 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10822 (tree-ssa-loop-ivopts.h): New header file.
10823 (struct builtin_info): New fields.
10824 (classify_builtin_1): Compute and record base and offset parts for
10825 memset builtin partition by calling strip_offset.
10826 (offset_cmp, fuse_memset_builtins): New functions.
10827 (finalize_partitions): Fuse adjacent memset partitions by calling
10829 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10830 Expose the interface.
10831 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10833 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10835 PR tree-optimization/82574
10836 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10837 that data reference must be executed exactly once per iteration
10838 against the outermost loop in nest.
10839 (classify_partition): Update call to above function.
10841 2017-10-18 Richard Biener <rguenther@suse.de>
10843 PR tree-optimization/82591
10844 * graphite.c (graphite_transform_loops): Move code gen message
10846 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10847 Here. Handle scop_to_isl_ast failing.
10848 (scop_to_isl_ast): Limit the number of ISL operations.
10850 2017-10-18 Richard Biener <rguenther@suse.de>
10852 * graphite-isl-ast-to-gimple.c
10853 (translate_isl_ast_to_gimple::set_rename): Simplify.
10854 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10855 (graphite_copy_stmts_from_block): ... here.
10856 (copy_bb_and_scalar_dependences): Simplify.
10857 (add_parameters_to_ivs_params): Canonicalize.
10858 (generate_entry_out_of_ssa_copies): Simplify.
10859 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10860 by passing in ISL dimension.
10861 (parameter_index_in_region_1): Rename to ...
10862 (parameter_index_in_region): ... this.
10863 (extract_affine): Adjust assert, pass down parameter index.
10864 (add_param_constraints): Use range-info when available.
10865 (build_scop_context): Adjust.
10866 * sese.c (new_sese_info): Adjust.
10867 (free_sese_info): Likewise.
10868 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10869 Remove unused typedefs.
10870 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10872 2017-10-18 Martin Liska <mliska@suse.cz>
10874 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10876 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10878 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10879 (s390_sched_init): Do not reset s390_sched_state if we entered the
10880 current basic block via a fallthru edge and all others are unlikely.
10882 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10884 * config/s390/s390.c (NUM_SIDES): New variable.
10885 (LONGRUNNING_THRESHOLD): New variable.
10886 (LATENCY_FACTOR): New variable.
10887 (s390_sched_score): Decrease score for long-running instructions on
10889 (s390_sched_variable_issue): Perform bookkeeping for long-running
10892 2017-10-18 Richard Biener <rguenther@suse.de>
10894 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10895 Simplify with removal of the parameter rename map.
10896 (set_rename): Likewise.
10897 (should_copy_to_new_region): Likewise.
10898 (graphite_copy_stmts_from_block): Likewise.
10899 (copy_bb_and_scalar_dependences): Remove initialization of
10900 unused copied_bb_map.
10901 (copy_def): Remove.
10902 (copy_internal_parameters): Likewise.
10903 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10904 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10905 Use INTEGRAL_TYPE_P.
10906 (parameter_index_in_region_1): Rename to ...
10907 (assign_parameter_index_in_region): ... this. Assert we have
10908 a parameter we handle.
10909 (scan_tree_for_params): Adjust.
10910 * sese.h (parameter_rename_map_t): Remove.
10911 (struct sese_info_t): Remove unused parameter_rename_map and
10912 copied_bb_map members.
10913 * sese.c (new_sese_info): Adjust.
10914 (free_sese_info): Likewise.
10916 2017-10-18 Martin Liska <mliska@suse.cz>
10919 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10920 on an abnormal edge.
10922 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10924 * doc/invoke.texi (ffunction-sections and fdata-sections):
10927 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10929 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10930 the use statement can throw internally.
10932 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10934 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10935 any RTX present on the RHS of a SET.
10936 * compare-elim.c (try_eliminate_compare): Restore comment.
10938 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10940 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10941 may be also called on tcc_type nodes.
10942 * langhooks.c (lhd_tree_size): Likewise.
10944 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10946 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10947 format_warning_at_substring.
10948 (maybe_warn): Convert source_range * param to a location_t. Pass
10949 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10950 (format_directive): Remove code to extract source_ranges and
10951 source_range * in favor of just a location_t.
10952 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10954 * substring-locations.c (format_warning_va): Convert
10955 source_range * param to a location_t.
10956 (format_warning_at_substring): Likewise.
10957 * substring-locations.h (format_warning_va): Likewise.
10958 (format_warning_at_substring): Likewise.
10960 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10962 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10964 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10965 and scatter/gather ops.
10967 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10968 vec_gather_load and vec_scatter_store.
10969 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10970 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10971 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10972 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10973 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10974 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10976 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10978 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10979 Detect FTST instruction and handle its register pops. Only pop
10980 second operand if can_pop_second_op is true.
10981 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10982 set can_pop_second_op to false in the compare_for_stack_reg call.
10984 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10985 output_fp_compare for stack register operands.
10986 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10987 instructions here. Do not emit stack register pops here. Assert
10988 that FCOMPP pops next to top stack register. Rewrite function.
10990 2017-10-17 Nathan Sidwell <nathan@acm.org>
10992 PR middle-end/82577
10993 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10994 use DECL_ASSEMBLER_NAME_RAW.
10996 PR middle-end/82546
10997 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11000 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11001 Wilco Dijkstra <wilco.dijkstra@arm.com>
11003 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11004 converstion to Pmode from the buf_addr.
11006 2017-10-17 Richard Biener <rguenther@suse.de>
11008 * graphite-dependences.c (scop_get_reads_and_writes): Change
11009 output parameters to references.
11011 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11013 PR 71026/tree-optimization
11014 * fold-const.c (distribute_real_division): Removed.
11015 (fold_binary_loc): Remove calls to distribute_real_divison.
11017 2017-10-17 Richard Biener <rguenther@suse.de>
11019 * graphite-scop-detection.c
11020 (scop_detection::stmt_has_simple_data_refs_p): Always use
11021 the full nest as region.
11022 (try_generate_gimple_bb): Likewise.
11023 * sese.c (scalar_evolution_in_region): Simplify now that
11024 SCEV can handle instantiation in regions.
11025 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11026 in the non-loop part of a function if requested.
11028 2017-10-17 Richard Biener <rguenther@suse.de>
11030 PR tree-optimization/82563
11031 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11033 (graphite_regenerate_ast_isl): Call it.
11034 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11036 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11038 PR tree-optimization/82549
11039 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11040 Formatting fixes. Instead of calling make_bit_field_ref with negative
11043 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11045 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11046 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11047 _mm_maskz_reduce_ss): New.
11048 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11049 __builtin_ia32_reducess_mask): Ditto..
11050 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11051 * config/i386/sse.md (reduces<mode>): Renamed to ...
11052 (reduces<mode><mask_scalar_name>): ... this.
11053 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11055 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11056 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11058 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11060 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11061 * selftest-run-tests.c (selftest::run_tests): Call
11062 selftest::unique_ptr_tests_cc_tests.
11063 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11064 * unique-ptr-tests.cc: New file.
11066 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11069 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11071 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11072 (make_hard_regno_born, make_hard_regno_dead): Update
11073 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11075 2017-10-16 Jeff Law <law@redhat.com>
11077 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11079 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11081 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11083 2017-10-16 Olivier Hainque <hainque@adacore.com>
11085 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11086 with_cpu if we were configured for an e500v2 target cpu name.
11088 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11090 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11091 * doc/invoke.texi: Document +nodsp as a valid extension for
11094 2017-10-16 Martin Liska <mliska@suse.cz>
11096 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11097 (test_set_range): Likewise.
11098 (test_range_functions): Rename to ...
11099 (test_bit_in_range): ... this.
11100 (sbitmap_c_tests): Add new test.
11102 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11104 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11106 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11107 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11109 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11111 * config/aarch64/aarch64-builtins.c
11112 (aarch64_types_quadopu_lane_qualifiers): New.
11113 (TYPES_QUADOPU_LANE): New.
11114 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11115 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11116 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11117 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11118 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11119 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11120 (Vdottype, DOTPROD): New.
11121 (sur): Add SDOT and UDOT.
11123 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11125 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11126 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11127 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11128 Add TARGET_DOTPROD.
11129 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11130 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11131 Enable TARGET_DOTPROD.
11132 (cortex-a75.cortex-a55): Likewise.
11133 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11135 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11137 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11138 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11140 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11142 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11143 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11144 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11145 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11146 * config/arm/types.md (neon_dot, neon_dot_q): New.
11147 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11149 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11151 * config/arm/arm.h (TARGET_DOTPROD): New.
11152 * config/arm/arm.c (arm_arch_dotprod): New.
11153 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11154 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11155 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11156 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11157 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11158 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11159 * doc/invoke.texi (armv8.2-a): Document dotprod
11161 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11163 * i386.c (ix86_vec_cost): New function.
11164 (ix86_rtx_costs): Handle vector operations better.
11165 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11166 * x86-tune-costs.h: Add new costs to all tables.
11168 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11170 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11172 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11173 divsd, sqrtss and sqrtsd
11174 * x86-tune-costs.h: Add new entries to all costs.
11175 (znver1_cost): Fix to match real instruction latencies.
11177 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11178 Michael Collison <michael.collison@arm.com>
11180 * compare-elim.c: Include emit-rtl.h.
11181 (can_merge_compare_into_arith): New function.
11182 (try_validate_parallel): Likewise.
11183 (try_merge_compare): Likewise.
11184 (try_eliminate_compare): Call the above when no previous clobber
11186 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11189 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11191 PR middle-end/62263
11192 PR middle-end/82498
11193 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11194 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11196 PR middle-end/62263
11197 PR middle-end/82498
11198 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11199 to be any operand_equal_p operands. For & (B - 1) require
11200 B to be power of 2. Recognize
11201 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11203 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11206 * optabs.c (expand_memory_blockage): Fix call of
11207 targetm.have_memory_blockage.
11209 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11212 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11213 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11214 objects to extra_objs instead of overwriting it.
11216 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11218 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11219 Use any_fp_register_operand as operand[3] predicate. Simplify
11220 equality test for operands[2] and operands[4] memory location.
11221 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11222 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11223 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11224 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11225 any_fp_register_operand as operand[1] predicate. Simplify
11226 equality test for operands[0] and operands[3] memory location.
11227 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11228 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11229 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11231 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11233 * target-insns.def: Add memory_blockage.
11234 * optabs.c (expand_memory_blockage): New function.
11235 (expand_asm_memory_barrier): Rename ...
11236 (expand_asm_memory_blockage): ... to this.
11237 (expand_mem_thread_fence): Call expand_memory_blockage
11238 instead of expand_asm_memory_barrier.
11239 (expand_mem_singnal_fence): Ditto.
11240 (expand_atomic_load): Ditto.
11241 (expand_atomic_store): Ditto.
11242 * doc/md.texi (Standard Pattern Names For Generation):
11243 Document memory_blockage instruction pattern.
11245 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11247 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11248 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11249 * config/rl78/rl78.md: New define_expand "adddi3".
11251 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11253 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11256 2017-10-13 Jeff Law <law@redhat.com>
11258 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11260 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11263 * internal-fn.c (expand_mul_overflow): If both operands have
11264 the same highpart of -1 or 0 and the topmost bit of lowpart
11265 is different, overflow is if res <= 0 rather than res < 0.
11267 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11269 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11270 TARGET_P9_VECTOR code for unaligned_load case.
11272 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11274 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11276 2017-10-13 Nathan Sidwell <nathan@acm.org>
11278 * tree-core.h (tree_contains_struct): Make bool.
11279 * tree.c (tree_contains_struct): Likewise.
11280 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11281 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11282 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11283 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11284 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11286 2017-10-13 Richard Biener <rguenther@suse.de>
11288 * graphite-isl-ast-to-gimple.c
11289 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11290 parameters and dominance check.
11291 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11292 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11293 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11294 Do not update SSA form here or do intermediate IL verification.
11295 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11296 (graphite_initialize): Remove check on the number of loops in
11297 the function and inline into graphite_transform_loops.
11298 (graphite_finalize): Inline into graphite_transform_loops.
11299 (graphite_transform_loops): Perform SSA update and IL verification
11301 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11303 2017-10-13 Richard Biener <rguenther@suse.de>
11305 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11306 graphite_expression_type_precision): Avoid global constructor
11308 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11309 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11310 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11311 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11312 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11313 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11315 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11318 * config/i386/i386.h (ix86_red_zone_size): New.
11319 * config/i386/i386.md (push peephole2s): Replace
11320 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11322 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11323 Alan Hayward <alan.hayward@arm.com>
11324 David Sherwood <david.sherwood@arm.com>
11326 * combine.c (can_change_dest_mode): Reject changes in
11327 REGMODE_NATURAL_SIZE.
11329 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11330 Alan Hayward <alan.hayward@arm.com>
11331 David Sherwood <david.sherwood@arm.com>
11333 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11334 (expand_debug_source_expr): Likewise.
11335 * combine.c (combine_simplify_rtx): Likewise.
11336 * cse.c (fold_rtx): Likewise.
11337 * fwprop.c (canonicalize_address): Likewise.
11338 * targhooks.c (default_shift_truncation_mask): Likewise.
11340 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11341 Alan Hayward <alan.hayward@arm.com>
11342 David Sherwood <david.sherwood@arm.com>
11344 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11345 (widened_mode): Likewise.
11346 (expand_unop): Likewise.
11347 * ree.c (transform_ifelse): Likewise.
11348 (merge_def_and_ext): Likewise.
11349 (combine_reaching_defs): Likewise.
11350 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11352 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11353 Alan Hayward <alan.hayward@arm.com>
11354 David Sherwood <david.sherwood@arm.com>
11356 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11357 * combine.c (gen_lowpart_for_combine): Likewise.
11358 * dwarf2out.c (rtl_for_decl_location): Likewise.
11359 * final.c (alter_subreg): Likewise.
11360 * rtlhooks.c (gen_lowpart_general): Likewise.
11361 (gen_lowpart_if_possible): Likewise.
11363 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11364 Alan Hayward <alan.hayward@arm.com>
11365 David Sherwood <david.sherwood@arm.com>
11367 * calls.c (expand_call): Use subreg_lowpart_offset.
11368 * cse.c (cse_insn): Likewise.
11369 * regcprop.c (copy_value): Likewise.
11370 (copyprop_hardreg_forward_1): Likewise.
11372 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11375 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11376 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11377 =Q constraints to +Q and into insn condition add check
11378 that operands[0] and operands[1] are equal.
11379 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11380 =Q constraints to +Q and into insn condition add check
11381 that operands[0] is equal to either operands[1] or operands[2].
11384 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11385 instead of handling MINUS_EXPR twice (once for each argument),
11386 canonicalize operand order and handle just once, use rtype where
11387 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11390 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11391 any values of __C while still being pattern recognizable as a simple
11392 rotate instruction.
11394 2017-10-13 Richard Biener <rguenther@suse.de>
11396 PR tree-optimization/82451
11398 2017-10-02 Richard Biener <rguenther@suse.de>
11400 PR tree-optimization/82355
11401 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11402 a mapping for the enclosing loop but avoid generating one for
11403 the loop tree root.
11404 (copy_bb_and_scalar_dependences): Remove premature codegen
11405 error on PHIs in blocks duplicated into multiple places.
11406 * graphite-scop-detection.c
11407 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11408 in the region use it as loop and nest to analyze the DR in.
11409 (try_generate_gimple_bb): Likewise.
11410 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11411 (add_loop_constraints): For blocks in a loop not in the region
11412 create a dimension with a single iteration.
11413 * sese.h (gbb_loop_at_index): Remove assert.
11415 * cfgloop.c (loop_preheader_edge): For the loop tree root
11416 return the single successor of the entry block.
11417 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11418 Reset the SCEV hashtable and niters.
11419 * graphite-scop-detection.c
11420 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11421 assert that we only have POLYNOMIAL_CHREC that vary in loops
11422 contained in the region.
11423 (scop_detection::graphite_can_represent_expr): Adjust.
11424 (scop_detection::stmt_has_simple_data_refs_p): For loops
11425 not in the region set loop to NULL. The nest is now the
11426 entry edge to the region.
11427 (try_generate_gimple_bb): Likewise.
11428 * sese.c (scalar_evolution_in_region): Adjust for
11429 instantiate_scev change.
11430 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11431 Make nest parameter the edge into the region.
11432 (create_data_ref): Likewise.
11433 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11434 entry edge into a region and adjust instantiate_scev calls.
11435 (create_data_ref): Likewise.
11436 (graphite_find_data_references_in_stmt): Likewise.
11437 (find_data_references_in_stmt): Pass the loop preheader edge
11438 from the nest argument.
11439 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11440 parameter the edge into the region.
11441 (instantiate_parameters): Use the loop preheader edge as entry.
11442 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11444 (get_instantiated_value_entry): Make instantiate_below parameter
11445 the edge into the region.
11446 (instantiate_scev_name): Likewise. Adjust dominance checks,
11447 when we cannot use loop-based instantiation instantiate by
11448 walking use-def chains.
11449 (instantiate_scev_poly): Adjust.
11450 (instantiate_scev_binary): Likewise.
11451 (instantiate_scev_convert): Likewise.
11452 (instantiate_scev_not): Likewise.
11453 (instantiate_array_ref): Remove.
11454 (instantiate_scev_3): Likewise.
11455 (instantiate_scev_2): Likewise.
11456 (instantiate_scev_1): Likewise.
11457 (instantiate_scev_r): Do not blindly handle N-operand trees.
11458 Do not instantiate array-refs. Handle all constants and invariants.
11459 (instantiate_scev): Make instantiate_below parameter
11460 the edge into the region.
11461 (resolve_mixers): Use the loop preheader edge for the region
11462 parameter to instantiate_scev_r.
11463 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11465 2017-10-13 Richard Biener <rguenther@suse.de>
11467 PR tree-optimization/82525
11468 * graphite-isl-ast-to-gimple.c
11469 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11471 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11472 Fail code generation when we cannot represent the isl integer.
11473 (binary_op_to_tree): Elide modulo operations that are no-ops
11474 in the type we code generate. Remove now superfluous code
11477 2017-10-13 Richard Biener <rguenther@suse.de>
11479 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11480 (scop_detection::harmful_loop_in_region): Remove premature
11481 IV type restriction.
11482 (scop_detection::graphite_can_represent_scev): We can handle
11483 pointer IVs just fine.
11485 2017-10-13 Alan Modra <amodra@gmail.com>
11487 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11488 "Clobbers and Scratch Registers". Add paragraph on
11489 alternative to clobbers for scratch registers and OpenBLAS
11492 2017-10-13 Alan Modra <amodra@gmail.com>
11494 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11495 example of a memory input for a string of known length. Move
11496 commentary out of table. Add a number of new examples
11497 covering array memory inputs.
11499 2017-10-12 Martin Liska <mliska@suse.cz>
11501 PR tree-optimization/82493
11502 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11503 (test_range_functions): New function.
11504 (sbitmap_c_tests): Likewise.
11505 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11506 * selftest.h (sbitmap_c_tests): New function.
11508 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11510 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11512 * config/rs6000/amo.h: Fix spacing issue.
11514 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11517 * config/i386/i386.md (*ashl<mode>3_mask_1,
11518 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11519 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11522 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11524 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11525 (profile_probability): Set max_probability
11526 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11528 * profile-count.c (profile_probability::differs_from_p): Do not
11529 rely on max_probaiblity == 10000
11531 2017-10-12 Jeff Law <law@redhat.com>
11533 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11536 2017-10-12 Martin Sebor <msebor@redhat.com>
11540 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11541 (handle_alias_pairs): Call it.
11542 * common.opt (-Wattribute-alias): New option.
11543 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11544 * doc/invoke.texi (-Wattribute-alias): Document.
11546 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11549 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11551 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11553 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11554 (make_hard_regno_born, make_hard_regno_dead): Update
11555 bb_killed_pseudos and bb_gen_pseudos.
11557 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11559 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11561 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11563 * config/alpha/alpha.c (alpha_split_conditional_move):
11564 Use std::swap instead of manually swapping.
11565 (alpha_stdarg_optimize_hook): Ditto.
11566 (alpha_canonicalize_comparison): Ditto.
11568 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11570 * tree-loop-distribution.c (struct builtin_info): New struct.
11571 (struct partition): Refactor fields into struct builtin_info.
11572 (partition_free): Free struct builtin_info.
11573 (build_size_arg_loc, build_addr_arg_loc): Delete.
11574 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11575 information from struct builtin_info.
11576 (find_single_drs): New function refactored from classify_partition.
11577 Also moved builtin validity checks to this function.
11578 (compute_access_range, alloc_builtin): New functions.
11579 (classify_builtin_st, classify_builtin_ldst): New functions.
11580 (classify_partition): Refactor code into functions find_single_drs,
11581 classify_builtin_st and classify_builtin_ldst.
11582 (distribute_loop): Don't do runtime alias check when distributing
11584 (find_seed_stmts_for_distribution): New function.
11585 (pass_loop_distribution::execute): Refactor code finding seed
11586 stmts into above function. Support distribution for the innermost
11587 two-level loop nest. Adjust dump information.
11589 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11591 * tree-loop-distribution.c: Adjust the general comment.
11592 (NUM_PARTITION_THRESHOLD): New macro.
11593 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11594 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11595 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11596 in call to build_partition_graph.
11597 (finalize_partitions): New parameter. Make loop distribution more
11598 conservative by fusing more partitions.
11599 (distribute_loop): Don't do runtime alias check in case of loop nest
11601 (find_seed_stmts_for_distribution): New function.
11602 (prepare_perfect_loop_nest): New function.
11603 (pass_loop_distribution::execute): Refactor code finding seed stmts
11604 and loop nest into above functions. Support loop nest distribution.
11605 Adjust dump information accordingly.
11607 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11609 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11610 and set PTYPE_SEQUENTIAL for merged partition.
11612 2017-10-12 Richard Biener <rguenther@suse.de>
11614 PR tree-optimization/69728
11616 2017-09-19 Richard Biener <rguenther@suse.de>
11618 PR tree-optimization/69728
11619 * graphite-sese-to-poly.c (schedule_error): New global.
11620 (add_loop_schedule): Handle empty domain by failing the
11622 (build_original_schedule): Handle schedule_error.
11624 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11625 domain by returning an unchanged schedule.
11627 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11629 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11630 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11632 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11634 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11637 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11640 * expr.c (store_field): Don't optimize away bitsize == 0 store
11641 from CALL_EXPR with addressable return type.
11643 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11645 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11646 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11647 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11648 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11649 TARGET_ISEL instead of TARGET_ISEL<sel>.
11651 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11653 * config/rs6000/rs6000.c
11654 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11656 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11658 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11660 (reg_or_cint_operand): Fix comment.
11661 (reg_or_zero_operand): New predicate.
11662 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11663 * config/rs6000/rs6000.c (output_isel): Delete.
11664 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11665 instead of reg_or_cint_operand. Output instruction directly (not via
11667 (isel_unsigned_<mode>): Ditto.
11668 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11669 gpc_reg_operand. Add an instruction alternative for this. Output
11670 instruction directly.
11671 (*isel_reversed_unsigned_<mode>): Ditto.
11673 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11675 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11676 (TARGET_CANONICALIZE_COMPARISON): Define.
11678 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11681 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11682 Check whether the dest is REG before adding REG_EQUIV note.
11684 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11687 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11689 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11690 (make_hard_regno_born, make_hard_regno_dead): Update
11691 bb_killed_pseudos and bb_gen_pseudos.
11693 2017-10-11 Nathan Sidwell <nathan@acm.org>
11695 * incpath.h (enum incpath_kind): Name enum, prefix values.
11696 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11697 * incpath.c (heads, tails): Use INC_MAX.
11698 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11699 (merge_include_chains, split_quote_chain,
11700 register_include_chains): Update incpath_kind names.
11701 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11702 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11704 (add_framework_path, darwin_register_objc_includes): Likewise.
11705 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11707 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11709 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11710 Do not use float_operator operator predicate.
11711 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11712 * config/i386/predicates.md (float_operator): Remove predicate.
11714 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11716 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11717 (*jccxf_i387): Ditto.
11718 (*jcc<mode>_i387): Ditto.
11719 (*jccu<mode>_i387): Ditto.
11720 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11721 (*jcc_*_i387 splitters): Remove.
11722 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11723 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11724 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11727 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11729 * profile-count.h (slow_safe_scale_64bit): New function.
11730 (safe_scale_64bit): New inline.
11731 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11732 * profile-count.c: Include wide-int.h
11733 (slow_safe_scale_64bit): New.
11735 2017-10-11 Nathan Sidwell <nathan@acm.org>
11737 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11738 HAS_DECL_ASSEMBLER_NAME_P.
11739 * gimple-expr.c (gimple_decl_printable_name: Check
11740 HAS_DECL_ASSEMBLER_NAME_P too.
11741 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11742 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11743 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11744 * passes.c (rest_of_decl_compilation): Check
11745 HAS_DECL_ASSEMBLER_NAME_P too.
11746 * recog.c (verify_changes): Likewise.
11747 * tree-pretty-print.c (dump_decl_name): Likewise.
11748 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11750 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11751 (SET_DECL_ASSEMBLER_NAME): Use it.
11752 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11753 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11754 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11756 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11758 * config.gcc (i386, x86_64): Add extra objects.
11759 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11760 (ix86_min_insn_size): Declare.
11761 (ix86_issue_rate): Declare.
11762 (ix86_adjust_cost): Declare.
11763 (ia32_multipass_dfa_lookahead): Declare.
11764 (ix86_macro_fusion_p): Declare.
11765 (ix86_macro_fusion_pair_p): Declare.
11766 (ix86_bd_has_dispatch): Declare.
11767 (ix86_bd_do_dispatch): Declare.
11768 (ix86_core2i7_init_hooks): Declare.
11769 (ix86_atom_sched_reorder): Declare.
11770 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11771 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11772 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11773 (rip_relative_addr_p): Rename to ...
11774 (ix86_rip_relative_addr_p): ... this one; export.
11775 (memory_address_length): Update.
11776 (ix86_issue_rate): Move to x86-tune-sched.c.
11777 (ix86_flags_dependent): Move to x86-tune-sched.c.
11778 (ix86_agi_dependent): Move to x86-tune-sched.c.
11779 (exact_dependency_1): Move to x86-tune-sched.c.
11780 (exact_store_load_dependency): Move to x86-tune-sched.c.
11781 (ix86_adjust_cost): Move to x86-tune-sched.c.
11782 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11783 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11784 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11785 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11786 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11787 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11788 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11789 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11790 (min_insn_size): Rename to ...
11791 (ix86_min_insn_size): ... this one; export.
11792 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11793 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11794 (core2i7_first_cycle_multipass_backtrack): Move to
11795 x86-tune-sched-core.c.
11796 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11797 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11798 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11799 (ix86_avoid_jump_mispredicts): Update.
11800 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11801 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11802 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11803 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11804 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11805 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11806 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11807 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11808 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11809 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11810 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11811 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11812 (BIG): Move to ix86-tune-sched-bd.c.
11813 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11814 (enum insn_path): Move to ix86-tune-sched-bd.c.
11815 (get_mem_group): Move to ix86-tune-sched-bd.c.
11816 (is_cmp): Move to ix86-tune-sched-bd.c.
11817 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11818 (is_branch): Move to ix86-tune-sched-bd.c.
11819 (is_prefetch): Move to ix86-tune-sched-bd.c.
11820 (init_window): Move to ix86-tune-sched-bd.c.
11821 (allocate_window): Move to ix86-tune-sched-bd.c.
11822 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11823 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11824 (process_end_window): Move to ix86-tune-sched-bd.c.
11825 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11826 (find_constant): Move to ix86-tune-sched-bd.c.
11827 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11828 (has_immediate): Move to ix86-tune-sched-bd.c.
11829 (get_insn_path): Move to ix86-tune-sched-bd.c.
11830 (get_insn_group): Move to ix86-tune-sched-bd.c.
11831 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11832 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11833 (add_insn_window): Move to ix86-tune-sched-bd.c.
11834 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11835 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11836 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11837 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11838 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11839 (do_dispatch): Move to ix86-tune-sched-bd.c.
11840 (has_dispatch): Move to ix86-tune-sched-bd.c.
11841 * config/i386/t-i386: Add new object files.
11842 * config/i386/x86-tune-costs.h: New file.
11843 * config/i386/x86-tune-sched-atom.c: New file.
11844 * config/i386/x86-tune-sched-bd.c: New file.
11845 * config/i386/x86-tune-sched-core.c: New file.
11846 * config/i386/x86-tune-sched.c: New file.
11848 2017-10-11 Liu Hao <lh_mouse@126.com>
11850 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11851 the generic version below instead.
11852 (should_colorize): Recognize Windows consoles as terminals
11854 * pretty-print.c [__MINGW32__] (write_all): New function.
11855 [__MINGW32__] (find_esc_head): Likewise.
11856 [__MINGW32__] (find_esc_terminator): Likewise.
11857 [__MINGW32__] (eat_esc_sequence): Likewise.
11858 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11860 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11863 2017-10-11 Richard Biener <rguenther@suse.de>
11865 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11866 Properly call analyze_scalar_evolution with the loop of the stmt.
11868 2017-10-11 Richard Biener <rguenther@suse.de>
11870 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11871 * tree-core.h (tree_base): Add chrec_var union member.
11872 * tree.h (CHREC_VAR): Remove.
11873 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11874 * tree-chrec.h (build_polynomial_chrec): Adjust.
11875 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11876 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11878 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11880 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11881 * match.pd: ... here.
11882 ((T) X == (T) Y): Relax condition.
11884 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11886 PR tree-optimization/82472
11887 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11889 (break_alias_scc_partitions): Update postorder number.
11891 2017-10-11 Martin Liska <mliska@suse.cz>
11894 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11896 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11898 2017-10-11 Martin Liska <mliska@suse.cz>
11900 * print-rtl.c (print_insn): Move declaration of idbuf
11901 to same scope as name.
11903 2017-10-11 Martin Liska <mliska@suse.cz>
11908 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11909 volatile arguments.
11911 2017-10-11 Martin Liska <mliska@suse.cz>
11914 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11915 volatile arguments.
11917 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11919 * config.gcc (default_gnu_indirect_function): Default to yes for
11920 arm*-*-linux* with glibc.
11922 2017-10-11 Richard Biener <rguenther@suse.de>
11924 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11925 default-defs and types we do not want to analyze.
11926 (interpret_loop_phi): Replace unreachable code with an assert.
11927 (compute_scalar_evolution_in_loop): Remove and inline ...
11928 (analyze_scalar_evolution_1): ... here, replacing condition with
11929 what makes the intent clearer. Remove handling of cases
11930 get_scalar_evolution now handles.
11932 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11934 PR rtl-optimization/81434
11935 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11936 comment for main loop. In sched_group_found if, also add checks for
11937 pass and min_cost_group.
11939 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11941 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11942 (rs6000_insn_cost): New function.
11943 * config/rs6000/rs6000.md (cost): New attribute.
11945 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11946 H.J. Lu <hongjiu.lu@intel.com>
11950 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11951 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11952 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11953 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11954 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11955 OPTION_MASK_ISA_MMX and some other bit set require both
11956 mmx and the ISAs without the mmx bit.
11957 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11958 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11959 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11960 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11961 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11962 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11963 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11964 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11965 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11966 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11967 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11968 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11969 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11970 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11971 Add OPTION_MASK_ISA_MMX.
11973 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11975 * config.gcc (armv7*-*-freebsd*): New target.
11976 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11978 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11980 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11981 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11984 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11986 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11987 a template parameter.
11988 (WIDE_INT_REF_FOR): Update accordingly.
11989 * tree.h (wi::int_traits <const_tree>): Delete.
11990 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11991 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11992 (wi::tree_to_wide_ref): New typedef.
11993 (wi::to_wide): New function.
11994 * calls.c (get_size_range): Use wi::to_wide when operating on
11995 trees as wide_ints.
11996 * cgraph.c (cgraph_node::create_thunk): Likewise.
11997 * config/i386/i386.c (ix86_data_alignment): Likewise.
11998 (ix86_local_alignment): Likewise.
11999 * dbxout.c (stabstr_O): Likewise.
12000 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12001 * expr.c (const_vector_from_tree): Likewise.
12002 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12003 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12004 (fold_negate_expr_1, int_const_binop_1, const_binop)
12005 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12006 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12007 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12008 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12009 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12010 (fold_not_const, round_up_loc): Likewise.
12011 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12012 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12013 (alloca_call_type): Likewise.
12014 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12015 * godump.c (go_output_typedef): Likewise.
12016 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12017 * internal-fn.c (get_min_precision): Likewise.
12018 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12019 * ipa-polymorphic-call.c
12020 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12021 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12022 (ipa_modify_call_arguments): Likewise.
12023 * match.pd: Likewise.
12024 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12025 * print-tree.c (print_node_brief, print_node): Likewise.
12026 * stmt.c (expand_case): Likewise.
12027 * stor-layout.c (layout_type): Likewise.
12028 * tree-affine.c (tree_to_aff_combination): Likewise.
12029 * tree-cfg.c (group_case_labels_stmt): Likewise.
12030 * tree-data-ref.c (dr_analyze_indices): Likewise.
12031 (prune_runtime_alias_test_list): Likewise.
12032 * tree-dump.c (dequeue_and_dump): Likewise.
12033 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12034 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12035 * tree-pretty-print.c (dump_generic_node): Likewise.
12036 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12037 (simple_iv_with_niters): Likewise.
12038 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12039 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12040 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12041 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12042 (refine_value_range_using_guard, number_of_iterations_ne_max)
12043 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12044 (get_cst_init_from_scev, record_nonwrapping_iv)
12045 (scev_var_range_cant_overflow): Likewise.
12046 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12047 * tree-ssa-pre.c (compute_avail): Likewise.
12048 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12049 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12050 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12051 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12052 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12053 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12054 (dump_case_nodes, try_switch_expansion): Likewise.
12055 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12056 (vect_do_peeling): Likewise.
12057 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12058 * tree-vect-stmts.c (vectorizable_load): Likewise.
12059 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12060 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12061 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12062 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12063 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12064 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12065 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12066 (evrp_dom_walker::before_dom_children): Likewise.
12067 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12068 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12069 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12070 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12071 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12072 (get_range_pos_neg): Likewise.
12073 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12074 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12075 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12076 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12077 * config/avr/avr.c (avr_fold_builtin): Likewise.
12078 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12079 * config/msp430/msp430.c (msp430_attr): Likewise.
12080 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12081 * config/powerpcspe/powerpcspe-c.c
12082 (altivec_resolve_overloaded_builtin): Likewise.
12083 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12084 (rs6000_expand_ternop_builtin): Likewise.
12085 * config/rs6000/rs6000-c.c
12086 (altivec_resolve_overloaded_builtin): Likewise.
12087 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12088 (rs6000_expand_ternop_builtin): Likewise.
12089 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12091 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12093 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12094 when copying loop nest with only one inner loop.
12096 2017-10-10 Richard Biener <rguenther@suse.de>
12098 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12099 blocks if SCEV is active.
12100 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12102 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12103 (scev_initialize): Assert we are not yet initialized.
12105 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12107 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12108 inner loop's exit stmt by making it always exit the loop, otherwise
12109 we would generate an infinite empty loop.
12111 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12113 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12114 renaming variables in new preheader if it's deleted.
12116 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12118 * tree-loop-distribution.c (struct partition): Remove unused field
12119 loops of the structure.
12120 (partition_alloc, partition_free): Ditto.
12121 (build_rdg_partition_for_vertex): Ditto.
12123 2017-10-09 Jeff Law <law@redhat.com>
12125 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12126 return type to match prototype and documentation.
12128 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12130 * config/rs6000/rs6000.c (processor_costs): Move to ...
12131 * config/rs6000/rs6000.h: ... here.
12132 (rs6000_cost): Declare.
12134 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12136 * except.c (setjmp_fn): New global variable.
12137 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12138 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12139 if DONT_USE_BUILTIN_SETJMP is defined.
12141 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12143 * target.def (insn_cost): New hook.
12144 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12145 * doc/tm.texi: Regenerate.
12146 * rtlanal.c (insn_cost): Use the new hook.
12148 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12150 * combine.c (combine_validate_cost): Compute the new insn_cost,
12151 not just pattern_cost.
12152 (try_combine): Adjust comment.
12154 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12156 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12158 * combine.c (uid_insn_cost): Adjust comment.
12159 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12161 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12162 * dse.c (find_shift_sequence): Ditto.
12163 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12164 (bb_valid_for_noce_process_p): Use pattern_cost.
12165 * rtl.h (insn_rtx_cost): Delete.
12166 (pattern_cost): New prototype.
12167 (insn_cost): New prototype.
12168 * rtlanal.c (insn_rtx_cost): Rename to...
12169 (pattern_cost): ... this.
12172 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12174 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12175 (*jcc<mode>_0_r_i387): Ditto.
12176 (*jccxf_r_i387): Ditto.
12177 (*jcc<mode>_r_i387): Ditto.
12178 (*jccu<mode>_r_i387): Ditto.
12179 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12180 (*jcc): Rename from *jcc_1.
12182 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12184 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12185 deferred rescans after the lvx/stvx recombination pre-pass.
12187 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12189 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12190 memory operation instruction support.
12191 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12192 (rs6000-ibm-aix[789]*): Likewise.
12193 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12194 Document new functions.
12196 2017-10-09 Richard Biener <rguenther@suse.de>
12198 PR tree-optimization/82397
12199 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12200 equality only for semantically equal trees.
12202 2017-10-09 Richard Biener <rguenther@suse.de>
12204 PR tree-optimization/82449
12205 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12206 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12207 allow constant addresses.
12208 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12211 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12213 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12216 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12219 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12222 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12225 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12227 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12230 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12231 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12233 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12235 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12236 (WI_BINARY_PREDICATE_RESULT): Likewise.
12237 (wi::binary_traits::operator_result): New type.
12238 (wi::binary_traits::predicate_result): Likewise.
12239 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12240 (generic_wide_int::operator==, generic_wide_int::operator!=)
12241 (generic_wide_int::operator&, generic_wide_int::and_not)
12242 (generic_wide_int::operator|, generic_wide_int::or_not)
12243 (generic_wide_int::operator^, generic_wide_int::operator+
12244 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12246 (operator~, unary operator-, operator==, operator!=, operator&)
12247 (operator|, operator^, operator+, binary operator-, operator*): New
12249 * expr.c (get_inner_reference): Use wi::bit_and_not.
12250 * fold-const.c (fold_binary_loc): Likewise.
12251 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12252 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12253 (bit_value_binop): Likewise.
12254 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12255 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12256 (extract_range_from_binary_expr_1): Likewise.
12257 (masked_increment): Likewise.
12258 (simplify_bit_ops_using_ranges): Likewise.
12260 2017-10-09 Martin Jambor <mjambor@suse.cz>
12263 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12264 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12265 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12267 (hsa_fixup_mov_insn_type): New function.
12268 (hsa_op_with_type::get_in_type): Use it.
12269 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12270 immediates in an assert.
12271 (hsa_op_with_type::extend_int_to_32bit): New method.
12272 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12273 types. Convert to dest type if necessary.
12274 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12275 (reg_for_gimple_ssa): Pass false as min32int to
12276 hsa_type_for_scalar_tree_type.
12277 (gen_hsa_addr): Fixup type when creating addresable temporary.
12278 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12279 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12280 necessary. Call hsa_fixup_mov_insn_type.
12281 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12282 extend operands and convert to dest type if necessary.
12283 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12284 to dest type if necessary.
12285 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12286 if conversion nt necessary and size matches.
12287 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12288 to dest type if necessary.
12289 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12290 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12292 (gen_hsa_clrsb): Likewise.
12293 (gen_hsa_ffs): Likewise.
12294 (gen_hsa_divmod): Extend operands and convert to dest type if
12296 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12298 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12300 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12301 Remove empty default arguments. Use a brace block as output
12303 (conditional return): Ditto.
12305 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12306 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12308 (group_ending_nop): Ditto.
12309 (doloop_end): Ditto.
12310 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12311 (splitters for those): Ditto.
12313 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12315 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12316 a conditional jump (and the compare for it) so that pc_rtx is the
12318 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12319 for the deleted and renamed ctr<mode>_internal[234] patterns.
12320 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12321 Delete second conditional return pattern.
12322 (ctr<mode>_internal2): Delete this second bdnz pattern.
12323 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12324 (ctr<mode>_internal4): Delete this second bdz pattern.
12326 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12328 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12329 (always_initialized_rtx_for_ssa_name_p): New predicate.
12330 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12331 (finish_out_of_ssa): Free new field of SA.
12332 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12333 * tree-ssa-coalesce.c: Include tree-ssa.h.
12334 (get_parm_default_def_partitions): Remove extern keyword.
12335 (get_undefined_value_partitions): New function.
12336 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12337 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12338 uninitialized bits.
12339 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12341 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12343 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12345 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12347 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12348 for targets that preffer 128bit.
12350 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12352 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12354 2017-10-08 Olivier Hainque <hainque@adacore.com>
12356 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12357 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12359 (thumb_set_return_address): Likewise.
12361 2017-10-08 Olivier Hainque <hainque@adacore.com>
12363 * common/config/arm/arm-common.c (arm_except_unwind_info):
12364 Handle DWARF2_UNWIND_INFO.
12366 2017-10-07 Michael Collison <michael.collison@arm.com>
12368 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12371 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12373 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12374 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12375 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12376 defined, force the creation of a new block for a dispatch label.
12378 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12380 * invoke.texi (Wsuggest-attribute=cold): Document.
12381 * common.opt (Wsuggest-attribute=cold): New
12382 * ipa-pure-const.c (warn_function_cold): New function.
12383 * predict.c (compute_function_frequency): Use it.
12384 * predict.h (warn_function_cold): Declare.
12386 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12388 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12391 2017-10-06 Martin Liska <mliska@suse.cz>
12393 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12394 keyword for member functions.
12395 (struct sanopt_tree_couple): New struct.
12396 (struct sanopt_tree_couple_hash): New function.
12397 (struct sanopt_ctx): Add new hash_map.
12398 (has_dominating_ubsan_ptr_check): New function.
12399 (record_ubsan_ptr_check_stmt): Likewise.
12400 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12401 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12402 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12404 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12407 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12408 aarch64_simd_valid_immediate on CONST_VECTORs.
12409 (aarch64_reg_or_bic_imm): Likewise.
12411 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12413 PR rtl-optimization/82396
12414 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12416 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12418 * graphite-dependences.c (scop_get_reads): Move code to...
12419 (scop_get_must_writes): Move code to...
12420 (scop_get_may_writes): Move code to...
12421 (scop_get_reads_and_writes): ... here.
12422 (scop_get_dependences): Call scop_get_reads_and_writes.
12424 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12426 PR tree-optimization/82434
12427 * fold-const.h (can_native_encode_type_p,
12428 can_native_encode_string_p): Remove.
12429 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12430 don't encode anything, just return what would be otherwise returned.
12431 (native_encode_fixed, native_encode_complex, native_encode_vector):
12433 (native_encode_string): Likewise. Inline by hand
12434 can_native_encode_string_p.
12435 (can_native_encode_type_p): Remove.
12436 (can_native_encode_string_p): Remove.
12437 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12438 STRING_CSTs using can_native_encode_string_p, test all
12439 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12440 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12441 argument from native_encode_expr.
12442 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12443 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12446 2017-10-06 Richard Biener <rguenther@suse.de>
12448 PR tree-optimization/82397
12449 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12450 operand_equal_p but rely on data_ref_compare_tree for detecting
12452 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12453 to match up with dr_group_sort_cmp.
12455 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12458 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12460 * config/s390/s390-builtin-types.def: Regenerate.
12462 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12465 * config/s390/s390-builtin-types.def: Regenerate.
12466 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12467 Change flag from B_VXE to B_VX.
12468 (s390_vec_min_dbl): Remove B_VXE flag.
12470 2017-10-06 Richard Biener <rguenther@suse.de>
12472 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12473 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12474 translate_isl_ast_to_gimple::is_valid_rename,
12475 translate_isl_ast_to_gimple::get_rename,
12476 translate_isl_ast_to_gimple::get_def_bb_for_const,
12477 translate_isl_ast_to_gimple::get_new_name,
12478 translate_isl_ast_to_gimple::collect_all_ssa_names,
12479 translate_isl_ast_to_gimple::copy_loop_phi_args,
12480 translate_isl_ast_to_gimple::collect_all_ssa_names,
12481 translate_isl_ast_to_gimple::copy_loop_phi_args,
12482 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12483 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12484 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12485 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12486 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12487 translate_isl_ast_to_gimple::copy_cond_phi_args,
12488 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12489 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12490 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12491 translate_isl_ast_to_gimple::rename_uses,
12492 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12493 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12494 (set_rename_for_each_def): Likewise.
12495 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12496 here. Handle rewriting SCEV analyzable uses here.
12497 (copy_bb_and_scalar_dependences): Generate code for PHI
12499 (graphite_regenerate_ast_isl): Adjust.
12500 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12501 (add_write, add_read): New functions.
12502 (build_cross_bb_scalars_def): Use it and simplify.
12503 (build_cross_bb_scalars_use): Likewise.
12504 (graphite_find_cross_bb_scalar_vars): Inline into...
12505 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12506 simulating out-of-SSA. Compute liveout and add dependencies.
12507 (build_scops): Force an empty entry block.
12508 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12510 (sese_build_liveouts): Declare.
12511 (sese_trivially_empty_bb_p): Likewise.
12512 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12513 compute liveout and debug_liveout.
12514 (sese_bad_liveouts_use): Remove.
12515 (sese_reset_debug_liveouts_bb): Likewise.
12516 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12517 (sese_build_liveouts): Build liveout and debug_liveout and store
12519 (new_sese_info): Adjust.
12520 (free_sese_info): Likewise.
12521 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12522 do not build liveout here.
12523 (move_sese_in_condition): Adjust region entry.
12524 (scev_analyzable_p): Match up with chrec_apply requirements.
12525 (sese_trivially_empty_bb_p): New.
12526 * tree-into-ssa.c (get_reaching_def): Properly support generating
12527 default-defs for incremental rewrite of anonymous names.
12529 2017-10-06 Richard Biener <rguenther@suse.de>
12531 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12532 precision do not perform modulo reduction.
12534 2017-10-06 Richard Biener <rguenther@suse.de>
12536 PR tree-optimization/82436
12537 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12538 conservatively choose the vectorization factor when checking
12539 whether we can perform the required load permutation.
12540 (vect_transform_slp_perm_load): Assert when we may not fail.
12542 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12544 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12545 message for incompatible -msdata=* and -mcall-* options.
12547 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12549 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12550 rate for post-reload scheduling.
12552 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12554 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12556 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12558 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12559 to improve monte carlo in scimark.
12561 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12563 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12564 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12565 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12566 pentium4_cost, nocona_cost): Set reassociation width to 1.
12567 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12568 width to 2 for fp operations and 1 otherwise.
12569 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12571 (atom_cost): Set reassociation width to 2.
12572 (slm_cost, generic_cost): Set fp reassociation width
12573 to 2 and 1 otherwise.
12574 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12575 (core_cost): Set fp reassociation width to 4 and vector to 2.
12576 (ix86_reassociation_width): Rewrite using cost table; special case
12577 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12578 and TARGET_AVX128_OPTIMAL.
12579 * config/i386/i386.h (processor_costs): Add
12580 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12581 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12582 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12583 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12584 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12585 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12587 2017-10-05 Nathan Sidwell <nathan@acm.org>
12589 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12591 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12593 * config/arm/arm.c (arm_test_fpu_data): New.
12594 (arm_run_selftests): Call arm_test_fpu_data.
12596 2017-10-04 Nathan Sidwell <nathan@acm.org>
12598 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12600 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12602 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12604 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12605 check type for aarch64_simd_valid_immediate.
12606 (aarch64_output_simd_mov_immediate): Update prototype.
12607 (aarch64_simd_valid_immediate): Update prototype.
12608 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12609 support for ORR-immediate.
12610 (and<mode>3): modified pattern to add support for BIC-immediate.
12611 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12612 now checks for valid immediate for BIC and ORR based on new enum
12614 (aarch64_output_simd_mov_immediate): Function now used to output
12615 BIC/ORR imm as well based on new enum argument.
12616 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12618 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12619 (aarch64_reg_or_bic_imm): Likewise.
12621 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12623 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12624 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12626 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12630 PR rtl-optimization/82396
12631 * haifa-sched.c (autopref_multipass_init): Simplify
12633 (autopref_rank_data): Simplify sort order.
12634 * sched-int.h (autopref_multipass_data_): Remove
12635 multi_mem_insn_p, min_offset and max_offset.
12637 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12639 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12641 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12643 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12644 vect_doubleint_cvt.
12646 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12648 * doc/sourcebuild.texi: Document vect_long_mult.
12650 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12652 PR tree-optimization/82413
12653 * fold-const.c (build_range_check): Use widest_int when comparing
12654 the maximum ETYPE value with HIGH.
12656 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12658 PR rtl-optimization/82396
12659 * haifa-sched.c (autopref_multipass_init): Simplify
12661 (autopref_rank_data): Simplify sort order.
12662 * sched-int.h (autopref_multipass_data_): Remove
12663 multi_mem_insn_p, min_offset and max_offset.
12665 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12667 PR tree-optimization/82381
12668 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12669 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12672 PR tree-optimization/82374
12673 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12674 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12675 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12676 current_function_decl to the new decl.
12678 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12680 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12681 helper macro for IEEE float128 hardware built-in functions.
12682 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12684 (TRUNCF128_ODD): Likewise.
12685 (ADDF128_ODD): Likewise.
12686 (SUBF128_ODD): Likewise.
12687 (MULF128_ODD): Likewise.
12688 (DIVF128_ODD): Likewise.
12689 (FMAF128_ODD): Likewise.
12690 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12691 UNSPEC_TRUNC_ROUND_TO_ODD.
12692 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12693 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12694 floating point round to odd instructions.
12695 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12696 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12697 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12698 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12699 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12700 (trunc<mode>sf2_hw): Change the truncate with round to odd
12701 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12702 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12703 to odd hardware instructions.
12704 (sub<mode>3_odd): Likewise.
12705 (mul<mode>3_odd): Likewise.
12706 (div<mode>3_odd): Likewise.
12707 (sqrt<mode>2_odd): Likewise.
12708 (fma<mode>4_odd): Likewise.
12709 (fms<mode>4_odd): Likewise.
12710 (nfma<mode>4_odd): Likewise.
12711 (nfms<mode>4_odd): Likewise.
12712 (trunc<mode>df2_odd): Change the truncate with round to odd
12713 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12715 * doc/extend.texi (PowerPC built-in functions): Update documentation
12716 for existing IEEE float128-bit built-in functions. Add built-in
12717 functions that generate the IEEE 128-bit floating point round to
12720 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12722 PR rtl-optimization/77729
12723 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12724 to (X&(C1&~C2))|C2 transformations.
12726 2017-10-03 Martin Jambor <mjambor@suse.cz>
12728 PR tree-optimization/82363
12729 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12730 mismatch, mark lacc written regardless of racc.
12732 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12734 PR tree-optimization/82381
12735 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12736 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12737 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12738 fallthrough into reassoc_stmt_dominates_stmt_p.
12741 * combine.c (combine_instructions): Don't combine in unreachable
12744 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12747 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12748 function to not use the have_cpu variable. Do not set cpu_index,
12749 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12750 or the default cpu.
12751 (rs6000_valid_attribute_p): Remove duplicate initializations of
12752 old_optimize and func_optimize.
12753 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12754 (rs6000_activate_target_options): Make global.
12755 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12758 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12760 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12761 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12762 Set *poffset to 0 and *psize and *pmax_size to -1 if
12763 *poffset + *psize overflows in HOST_WIDE_INT.
12765 PR tree-optimization/82387
12766 PR tree-optimization/82388
12767 PR tree-optimization/82389
12768 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12769 instead of live_bytes non-NULL.
12771 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12774 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12777 2017-10-02 Richard Biener <rguenther@suse.de>
12779 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12780 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12781 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12783 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12785 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12786 requested precision matches the type's.
12787 * calls.c (alloc_max_size): Calculate the new candidate size as
12788 a widest_int and use wi::to_widest when comparing it with the
12789 current candidate size.
12790 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12791 zero rather than integer_zero_node.
12792 * match.pd: Check for a no-op conversion before using wi::add
12793 rather than after. Use tree_to_uhwi when summing small shift
12794 counts into an unsigned int.
12796 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12797 Alan Hayward <alan.hayward@arm.com>
12798 David Sherwood <david.sherwood@arm.com>
12801 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12802 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12803 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12804 POINTER_AND_FP_REGS.
12806 2017-10-02 Richard Biener <rguenther@suse.de>
12808 PR tree-optimization/82355
12809 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12810 a mapping for the enclosing loop but avoid generating one for
12811 the loop tree root.
12812 (copy_bb_and_scalar_dependences): Remove premature codegen
12813 error on PHIs in blocks duplicated into multiple places.
12814 * graphite-scop-detection.c
12815 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12816 in the region use it as loop and nest to analyze the DR in.
12817 (try_generate_gimple_bb): Likewise.
12818 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12819 (add_loop_constraints): For blocks in a loop not in the region
12820 create a dimension with a single iteration.
12821 * sese.h (gbb_loop_at_index): Remove assert.
12823 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12825 * omp-expand.c (adjust_context_scope): New function.
12826 (expand_parallel_call): Call adjust_context_scope.
12828 2017-10-01 Jeff Law <law@redhat.com>
12830 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12831 dom_opt_dom_walker class with direct access to private members.
12832 Add comments. Call test_for_singularity.
12833 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12834 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12835 m_dummy_cond anymore.
12836 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12838 (pass_dominator:execute): Build the dummy_cond here and pass it
12839 to the dom_opt_dom_walker ctor.
12840 (test_for_singularity): New function.
12842 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12843 Maya Rashish <coypu@sdf.org>
12845 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12846 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12847 (alpha*-*-netbsd*) Use nbsd_tm_file.
12848 (arm*-*-netbsdelf*) Likewise.
12849 (i[34567]86-*-netbsdelf*) Likewise.
12850 (x86_64-*-netbsd*) Likewise.
12851 (mips*-*-netbsd*) Likewise.
12852 (powerpc-*-netbsd*) Likewise.
12853 (sh*-*-netbsd*) Likewise.
12854 (sparc-*-netbsdelf*) Likewise.
12855 (sparc64-*-netbsd*) Likewise.
12856 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12858 (vax-*-netbsdelf*) Likewise.
12859 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12860 (UINT_FAST8_TYPE) Likewise.
12861 (INT_FAST16_TYPE) Check CHAR_FAST16.
12862 (UINT_FAST16_TYPE) Likewise.
12864 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12867 * config/i386/i386.md
12868 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12869 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12870 *divmodsi4_zext_2): New define_insn_and_split.
12871 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12872 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12873 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12874 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12875 New define_insn_and_split.
12876 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12877 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12878 operands[1] having DImode when mode is SImode.
12880 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12881 always SImode for DIV and MOD in REG_EQUAL notes.
12883 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12885 PR middle-end/82319
12886 * match.pd: Fix handling of NaNs in pattern.
12888 2017-09-29 Jeff Law <law@redhat.com>
12890 * sbitmap.c (bitmap_bit_in_range_p): New function.
12891 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12892 * tree-ssa-dse.c (live_bytes_read): New function.
12893 (dse_classify_store): Ignore reads of dead bytes.
12895 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12896 typos and whitespace errors.
12897 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12898 * config/s390/s390.c (s390_emit_prologue): Likewise.
12900 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12903 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12904 with a symbol for LRA.
12906 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12908 PR rtl-optimization/82338
12909 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12911 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12913 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12914 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12915 (qsort_chk): Declare.
12916 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12917 (qsort_chk): New function.
12919 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12921 PR tree-optimization/82337
12922 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12923 phi definition if the PHI result appears in an abnormal PHI.
12924 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12925 basis appears in an abnormal PHI.
12927 2017-09-29 Richard Biener <rguenther@suse.de>
12929 * graphite-isl-ast-to-gimple.c
12930 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12931 (binary_op_to_tree): Use it.
12932 (get_rename_from_scev): Likewise.
12933 (copy_loop_phi_nodes): Likewise.
12934 (copy_bb_and_scalar_dependences): Likewise.
12935 (translate_pending_phi_nodes): Likewise.
12937 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12940 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12941 for movabsq $(i32 << shift), r64.
12943 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12945 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12946 index when encoding %esp as %rsp to avoid 0x67 prefix.
12948 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12950 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12951 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12953 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12955 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12956 Extensions with more than 16 double VFP registers.
12957 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12958 to_clear_mask and all code related to it. Replace the remaining
12959 entry by a sbitmap and adapt code accordingly.
12961 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12963 * brig-builtins.def: Change pure attributes to const.
12965 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12967 * config.gcc (default_gnu_indirect_function): Default to yes for
12968 sparc*-*-linux* with glibc.
12970 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12972 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12973 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12974 when creating .init_array and .fini_array sections with priority
12977 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12980 * config/aarch64/aarch64.c
12981 (aarch64_builtin_support_vector_misalignment): Always return false
12982 when misalignment is unknown.
12984 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12986 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12987 this function to return false if the definition used by the swap
12988 instruction is artificial, or if the memory address from which the
12989 constant value is loaded is not represented by a base address held
12990 in a register or if the base address register is a frame or stack
12991 pointer. Additionally, return false if the base address of the
12992 loaded constant is a SYMBOL_REF but is not considered to be a
12994 (replace_swapped_load_constant): New function.
12995 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12996 loaded constant vector with a load of a swapped constant vector.
12998 2017-09-27 Carl Love <cel@us.ibm.com>
13000 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13001 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13002 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13005 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13007 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13008 first, always call autopref_rank_data otherwise.
13010 2017-09-27 Richard Biener <rguenther@suse.de>
13012 * graphite-scop-detection.c (find_scop_parameters): Move
13013 loop bound handling ...
13014 (gather_bbs::before_dom_children): ... here, avoiding the need
13015 to build scop_info->loop_nest.
13016 (record_loop_in_sese): Remove.
13017 * sese.h (sese_info_t::loop_nest): Remove.
13018 * sese.c (new_sese_info): Do not allocate loop_nest.
13019 (free_sese_info): Do not free loop_nest.
13021 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13024 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13025 lhs from calls if the lhs has addressable type.
13027 2017-09-27 Richard Biener <rguenther@suse.de>
13029 * graphite.h (scop::max_alias_set): New member.
13030 * graphite-scop-detection.c: Remove references to non-existing
13031 --param in comments.
13032 (build_alias_sets): Record the maximum alias set used for drs.
13033 (build_scops): Support zero as unlimited for
13034 --param graphite-max-arrays-per-scop.
13035 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13036 and inline into ...
13037 (build_poly_sr_1): ... here. Compute alias set based on the
13038 maximum alias set used for drs rather than
13039 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13041 2017-09-27 Richard Biener <rguenther@suse.de>
13043 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13044 --param loop-block-tile-size=0 to disable tiling.
13046 2017-09-27 Richard Biener <rguenther@suse.de>
13048 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13049 (graphite-max-nb-scop-params): Document special value zero.
13050 * domwalk.h (dom_walker::STOP): New symbolical constant.
13051 (dom_walker::dom_walker): Add optional parameter for bb to
13053 (dom_walker::~dom_walker): Declare.
13054 (dom_walker::before_dom_children): Document STOP return value.
13055 (dom_walker::m_user_bb_to_rpo): New member.
13056 (dom_walker::m_bb_to_rpo): Likewise.
13057 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13058 mapping here if not provided by the user.
13059 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13060 provided by the user.
13061 (dom_walker::STOP): Define.
13062 (dom_walker::walk): Do not compute bb to RPO mapping here.
13063 Support STOP return value from before_dom_children to stop
13065 * graphite-optimize-isl.c (optimize_isl): If the schedule
13066 is the same still generate code if -fgraphite-identity
13067 or -floop-parallelize-all are given.
13068 * graphite-scop-detection.c: Include cfganal.h.
13069 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13071 (gather_bbs::before_dom_children): Return STOP for BBs
13073 (build_scops): Compute bb to RPO mapping and pass it to
13074 the domwalk. Treat --param graphite-max-nb-scop-params=0
13075 as not limiting the number of params.
13076 * graphite.c (graphite_initialize): Remove limit on the
13077 number of basic-blocks in a function.
13078 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13079 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13080 default value of 10.
13082 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13084 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13085 Adjust code to eliminate needing to do the shift right 32-bits
13086 operation after XSCVDPSPN.
13088 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13090 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13091 ((X / Y) != 0 -> X >= Y): Likewise.
13093 2017-09-26 Carl Love <cel@us.ibm.com>
13095 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13096 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13097 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13098 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13099 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13100 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13101 definitions and overloading.
13102 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13103 statement for P9V_BUILTIN_XST_LEN_R.
13104 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13105 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13106 define_expand and define_insn for the instructions and builtins.
13107 * doc/extend.texi: Update the built-in documentation file for the new
13108 built-in functions.
13109 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13110 define_insn for the instructions
13112 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13115 * gcc/config/netbsd-protos.h: New file.
13116 * gcc/config/netbsd.c: New file.
13117 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13118 * gcc/config/t-netbsd: New file.
13119 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13120 (tmake_file) Add t-netbsd.
13121 (extra_objs) Add netbsd.o.
13123 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13127 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13129 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13131 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13132 sign extension from a vector register to a GPR by doing a 32-bit
13133 direct move and then an EXTSW.
13134 (extendsi<mode>2 splitter): Likewise.
13135 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13136 right or vector extract after doing XSCVDPSPN. Use
13137 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13139 (movdi_from_sf_zero_ext): Likewise.
13140 (reload_gpr_from_vsxsf): Likewise.
13141 (p8_mfvsrd_4_disf): Delete, no longer used.
13142 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13143 then needing to move the SFmode to a GPR to use the XSCVDPSP
13144 instruction instead of FRSP and XSCVDPSPN.
13145 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13146 it is adjacent to the other XSCVSPDP insns.
13147 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13148 SFmode to be in traditional Altivec registers.
13149 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13150 (vsx_xscvspdpn): Likewise.
13151 (vsx_xscvspdpn_scalar): Likewise.
13153 2017-09-26 Martin Jambor <mjambor@suse.cz>
13155 * tree-sra.c (compare_access_positions): Put integral types first,
13156 stabilize sorting of integral types, remove conditions putting
13157 non-full-precision integers last.
13158 (sort_and_splice_var_accesses): Disable scalarization if a
13159 non-integert would be represented by a non-full-precision integer.
13161 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13163 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13164 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13165 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13166 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13167 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13168 conditionals inside the function instead of around it. Call
13169 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13170 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13172 2017-09-26 Richard Biener <rguenther@suse.de>
13174 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13176 (scop_detection::build_scop_breadth): ... this. Removed.
13177 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13178 (scop_detection::harmful_stmt_in_bb): Likewise.
13179 (scop_detection::graphite_can_represent_stmt): Likewise.
13180 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13181 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13182 (scop_detection::can_represent_loop_1): ... this. Removed.
13183 (scop_detection::harmful_loop_in_region): Simplify after inlining
13184 the above and remove more quadraticness.
13185 (build_scops): Adjust.
13186 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13189 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13192 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13193 REGNO (base) == SP_REG if base is a REG.
13195 PR middle-end/35691
13196 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13197 if it is different SSA_NAME.
13198 (optimize_range_tests_cmp_bitwise): New function.
13199 (optimize_range_tests): Call it.
13201 2017-09-26 Richard Biener <rguenther@suse.de>
13203 PR tree-optimization/82321
13204 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13205 for the def being inside the loop.
13207 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13209 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13211 * config/s390/s390-builtins.def: Fix constraint on op4.
13213 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13215 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13216 independent expanders.
13217 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13218 ("vec_ordered", "vec_unordered"): New expanders.
13220 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13222 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13225 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13227 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13228 vec_unpacks_lo_v16qi.
13229 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13231 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13233 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13234 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13235 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13237 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13239 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13241 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13243 ("*vec_slb<mode>"): New insn pattern.
13244 ("vec_shr_<mode>"): New expander.
13245 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13246 and force the shift count operand to V16QImode.
13247 ("vec_srb<mode>"): Set shift count mode to V16QI.
13249 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13251 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13252 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13253 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13255 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13258 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13260 2017-09-26 Richard Biener <rguenther@suse.de>
13262 PR tree-optimization/82320
13263 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13266 2017-09-25 Jeff Law <law@redhat.com>
13268 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13269 prototype for new argument.
13270 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13271 mostly extracted from rs6000_emit_allocate_stack.
13272 (rs6000_emit_probe_stack_range_stack_clash): New function.
13273 (rs6000_emit_allocate_stack): Call
13274 rs6000_emit_probe_stack_range_stack_clash as needed.
13275 (rs6000_emit_probe_stack_range): Add additional argument
13276 to call to gen_probe_stack_range{si,di}.
13277 (output_probe_stack_range): New.
13278 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13279 (output_probe_stack_range_stack_clash): New.
13280 (rs6000_emit_prologue): Emit notes into dump file as requested.
13281 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13282 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13283 Add additional operand and pass it to output_probe_stack_range.
13285 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13287 PR tree-optimization/82163
13288 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13289 (checking_verify_loop_closed_ssa): New parameter.
13290 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13291 (check_loop_closed_ssa_stmt): Delete.
13292 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13293 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13294 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13297 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13299 * brig-builtins.def: Treat HSAIL barrier builtins as
13300 setjmp/longjump style functions.
13302 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13304 * target.def (constant_alignment): New hook.
13305 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13306 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13307 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13308 * doc/tm.texi: Regenerate.
13309 * targhooks.h (default_constant_alignment): Declare.
13310 (constant_alignment_word_strings): Likewise.
13311 * targhooks.c (default_constant_alignment): New function.
13312 (constant_alignment_word_strings): Likewise.
13313 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13314 instead of CONSTANT_ALIGNMENT.
13315 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13316 (force_const_mem): Likewise.
13317 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13318 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13319 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13320 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13321 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13323 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13324 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13325 constant_alignment_word_strings.
13326 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13327 (CONSTANT_ALIGNMENT): Likewise.
13328 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13329 (arm_constant_alignment): New function.
13330 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13331 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13332 constant_alignment_word_strings.
13333 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13334 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13335 constant_alignment_word_strings.
13336 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13337 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13338 (cris_constant_alignment): New function.
13339 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13340 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13341 (epiphany_constant_alignment): New function.
13342 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13343 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13344 constant_alignment_word_strings.
13345 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13346 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13347 constant_alignment_word_strings.
13348 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13349 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13350 constant_alignment_word_strings.
13351 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13352 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13353 * config/i386/i386.c (ix86_constant_alignment): Make static.
13354 Use the same interface as the target hook.
13355 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13356 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13357 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13358 constant_alignment_word_strings.
13359 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13360 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13361 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13362 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13363 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13364 constant_alignment_word_strings.
13365 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13366 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13367 constant_alignment_word_strings.
13368 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13369 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13370 constant_alignment_word_strings.
13371 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13372 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13374 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13375 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13376 * config/mips/mips.c (mips_constant_alignment): New function.
13377 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13378 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13379 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13380 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13381 (mmix_constant_alignment): Make static. Use the same interface
13382 as the target hook.
13383 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13384 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13385 constant_alignment_word_strings.
13386 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13387 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13388 constant_alignment_word_strings.
13389 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13390 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13391 constant_alignment_word_strings.
13392 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13393 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13394 (rs6000_constant_alignment): New function.
13395 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13396 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13397 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13398 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13399 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13400 (rs6000_constant_alignment): New function.
13401 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13402 * config/s390/s390.c (s390_constant_alignment): New function.
13403 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13404 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13405 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13406 constant_alignment_word_strings.
13407 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13408 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13409 (sparc_constant_alignment): New function.
13410 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13411 * config/spu/spu.c (spu_constant_alignment): New function.
13412 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13413 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13414 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13415 constant_alignment_word_strings.
13416 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13417 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13418 constant_alignment_word_strings.
13419 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13420 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13421 constant_alignment_word_strings.
13422 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13423 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13424 (visium_constant_alignment): New function.
13425 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13426 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13427 (xtensa_constant_alignment): New function.
13428 * system.h (CONSTANT_ALIGNMENT): Poison.
13430 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13432 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13433 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13434 (rs6000_builtin_valid_without_lhs): New helper function.
13435 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13436 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13438 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13440 * target.h (vec_perm_indices): Use unsigned short rather than
13442 (auto_vec_perm_indices): Likewise.
13443 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13444 Use unsigned int rather than unsigned char.
13445 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13447 2017-09-25 Richard Biener <rguenther@suse.de>
13449 * cfgloop.h (sort_sibling_loops): Declare.
13450 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13451 (sort_sibling_loops): New function sorting the sibling loop list
13453 * graphite.c (graphite_transform_loops): Sort sibling loops.
13455 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13457 * target.def (vec_perm_const_ok): Change sel parameter to
13459 * optabs-query.c (can_vec_perm_p): Update accordingly.
13460 * doc/tm.texi: Regenerate.
13461 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13462 auto_vec_perm_indices and remove separate nelt field.
13463 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13464 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13465 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13466 (aarch64_expand_vec_perm_const): Update accordingly.
13467 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13468 to vec_perm_indices.
13469 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13470 auto_vec_perm_indices and remove separate nelt field.
13471 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13472 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13473 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13475 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13476 to vec_perm_indices.
13477 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13478 sel to vec_perm_indices.
13479 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13480 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13481 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13483 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13486 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13489 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13490 on the FUNCTION_DECL function context if it has a DIE that is a
13493 2017-09-25 Richard Biener <rguenther@suse.de>
13495 PR tree-optimization/82285
13496 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13499 2017-09-25 Tom de Vries <tom@codesourcery.com>
13503 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13506 2017-09-25 Richard Biener <rguenther@suse.de>
13508 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13509 ISL errors other than isl_error_quota happen. Dump if the
13510 schedule is the same.
13511 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13512 errors instead of aborting inside ISL.
13514 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13517 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13518 of libgcc_eh for m64.
13519 * config/i386/darwin64.h: Likewise.
13521 2017-09-25 Richard Biener <rguenther@suse.de>
13523 PR middle-end/82144
13524 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13525 attribute for incomplete types nor twice for complete ones.
13527 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13530 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13531 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13534 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13537 * config/i386/i386.opt (mprefer-avx256): Use
13538 ix86_target_flags variable.
13539 * config/i386/i386.c (ix86_target_string): Move
13540 -mprefer-avx256 to flag2_opts.
13542 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13544 PR middle-end/35691
13545 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13546 and x != -1 | y != -1 into (x & y) != -1.
13548 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13550 * config.gcc: Add new case statement to set
13551 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13552 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13553 s390x-*-linux* case statements. Added aarch64 to the list of
13554 supported architectures.
13556 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13558 PR tree-optimization/82289
13559 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13560 STMT_VINFO_RELEVANT_P.
13562 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13563 Alan Hayward <alan.hayward@arm.com>
13564 David Sherwood <david.sherwood@arm.com>
13566 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13567 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13568 (extract_range_from_binary_expr_1): Don't call
13569 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13571 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13572 Alan Hayward <alan.hayward@arm.com>
13573 David Sherwood <david.sherwood@arm.com>
13575 * target.def (preferred_vector_alignment): New hook.
13576 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13578 * doc/tm.texi: Regenerate.
13579 * targhooks.h (default_preferred_vector_alignment): Declare.
13580 * targhooks.c (default_preferred_vector_alignment): New function.
13581 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13583 (DR_TARGET_ALIGNMENT): New macro.
13584 (aligned_access_p): Update commentary.
13585 (vect_known_alignment_in_bytes): New function.
13586 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13588 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13589 Calculate the misalignment based on the target alignment rather than
13591 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13592 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13593 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13594 the target alignment, rather than masking the element misalignment
13595 with the number of elements in a vector. Also use the target
13596 alignment when calculating the maximum number of peels.
13597 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13598 instead of TYPE_ALIGN_UNIT.
13599 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13600 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13601 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13602 (vect_create_data_ref_ptr): Likewise.
13603 (vect_setup_realignment): Realign by ANDing with
13604 -DR_TARGET_MISALIGNMENT.
13605 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13606 the number of peels based on DR_TARGET_ALIGNMENT.
13607 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13608 with the guaranteed alignment boundary when deciding whether
13610 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13611 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13612 (ensure_base_align): Remove stmt_info parameter. Get the
13613 target base alignment from DR_TARGET_ALIGNMENT.
13614 (vectorizable_store): Update call accordingly. Interpret
13615 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13617 (vectorizable_load): Likewise.
13619 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13620 Alan Hayward <alan.hayward@arm.com>
13621 David Sherwood <david.sherwood@arm.com>
13623 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13624 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13625 (vect_enhance_data_refs_alignment): Likewise.
13627 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13629 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13630 error. Only quit immediately if parsing is complete.
13631 (BEGIN): Initialize fatal_err and parse_done.
13632 (begin fpu, end fpu): Check number of arguments.
13633 (begin arch, end arch): Likewise.
13634 (begin cpu, end cpu): Likewise.
13635 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13636 (optalias): Likewise.
13638 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13640 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13641 * config/arm/arm-isa.h: Delete. Move definitions to ...
13642 * arm-cpus.in: ... here. Use new feature and fgroup values.
13643 * config/arm/arm.c (arm_option_override): Use lower case for feature
13645 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13646 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13647 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13649 (print_isa_bits_for): New function.
13650 (gen_isa): New function.
13651 (gen_comm_data): Use print_isa_bits_for.
13652 (define feature): New keyword.
13653 (define fgroup): New keyword.
13654 * config/arm/t-arm (TM_H): Remove.
13655 (GTM_H): Add arm-isa.h.
13656 (arm-isa.h): Add rule to generate file.
13657 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13658 case for feature bit names.
13660 2017-09-22 Richard Biener <rguenther@suse.de>
13662 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13664 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13665 print of no dependency loops ...
13666 * graphite.c (graphite_transform_loops): ... here.
13667 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13669 (same_close_phi_node, remove_duplicate_close_phi,
13670 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13671 (canonicalize_loop_closed_ssa): ... here and simplify.
13672 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13673 (optimize_isl): Use dump_printf_loc to tell when we stopped
13674 optimizing because of an ISL timeout.
13676 2017-09-22 Richard Biener <rguenther@suse.de>
13678 PR tree-optimization/82291
13679 * tree-if-conv.c (predicate_mem_writes): Make sure to
13680 remove writes in blocks predicated with false.
13682 2017-09-22 Richard Biener <rguenther@suse.de>
13684 * sese.c: Include cfganal.h.
13685 (if_region_set_false_region): Remove.
13686 (create_if_region_on_edge): Likewise.
13687 (move_sese_in_condition): Re-implement without destroying
13690 2017-09-22 Richard Biener <rguenther@suse.de>
13692 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13693 Verify both BBs contain loop PHI nodes before dispatching to
13694 copy_loop_phi_args.
13695 (graphite_regenerate_ast_isl): Do not recompute dominators,
13696 do not verify three times. Restructure for clarity.
13697 * graphite-scop-detection.c (same_close_phi_node,
13698 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13699 defined_in_loop_p, canonicalize_loop_closed_ssa,
13700 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13701 checking and SSA rewrite, move to ...
13702 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13703 (graphite_initialize): Do not pass in ctx, do not reset the
13704 SCEV cache, compute only dominators.
13705 (graphite_transform_loops): Allocate ISL ctx after
13706 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13707 Maintain post-dominators only around build_scops.
13708 * sese.c (if_region_set_false_region): Make static. Free
13709 and recompute dominators.
13710 (move_sese_in_condition): Assert we don't get called with
13711 post-dominators computed.
13712 * sese.h (if_region_set_false_region): Remove.
13714 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13716 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13717 mode attribute for TARGET_AVX512VL.
13719 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13721 * config/i386/i386.opt (mprefer-avx256): New option.
13722 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13724 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13725 for TARGET_PREFER_AVX256.
13726 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13728 2017-09-21 Jeff Law <law@redhat.com>
13730 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13731 Fix dump output if the only stack space is for pushed registers.
13733 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13735 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13738 2017-09-21 Martin Sebor <msebor@redhat.com>
13741 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13742 code (in C++) or code that triggers warnings.
13744 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13746 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13748 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13750 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13751 * sched-rgn.c: Ditto.
13752 * sel-sched-ir.c: Ditto.
13754 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13756 * toplev.h (set_random_seed): Adjust return type.
13757 * toplev.c (init_local_tick): Move eager initialization of random_seed
13758 to get_random_seed. Adjust comment.
13759 (init_random_seed): Inline to get_random_seed, delete.
13760 (get_random_seed): Initialize random_seed lazily.
13761 (set_random_seed): Do not return previous value.
13762 (print_switch_value): Do not call get_random_seed.
13764 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13766 * cgraph.c (delete_function_version): New, broken out from...
13767 (cgraph_node::delete_function_version): ...here. Rename to
13768 cgraph_node::delete_function_version_by_decl. Update all uses.
13769 (cgraph_node::remove): Call delete_function_version.
13771 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13774 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13775 VAR_DECLs to which addressable non-volatile parameters are mapped
13776 and for id->retvar after the return value assignment. Clear
13777 id->retval and id->retbnd after inlining.
13779 2017-09-21 Richard Biener <rguenther@suse.de>
13781 PR tree-optimization/82276
13782 PR tree-optimization/82244
13783 * tree-vrp.c (build_assert_expr_for): Set
13784 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13786 (remove_range_assertions): Revert earlier change.
13788 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13791 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13793 2017-09-21 Richard Biener <rguenther@suse.de>
13795 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13796 Restore valid IL after code generation errors.
13797 * graphite.c (graphite_transform_loops): Diagnose code
13798 generation issues as MSG_MISSED_OPTIMIZATION and continue
13799 with processing SCOPs.
13801 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13802 Alan Hayward <alan.hayward@arm.com>
13803 David Sherwood <david.sherwood@arm.com>
13805 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13806 rather than choosing between plus_constant and gen_rtx_<CODE>.
13807 * expr.c (emit_push_insn): Likewise.
13808 (expand_expr_real_2): Likewise.
13810 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13811 Alan Hayward <alan.hayward@arm.com>
13812 David Sherwood <david.sherwood@arm.com>
13814 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13816 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13817 Alan Hayward <alan.hayward@arm.com>
13818 David Sherwood <david.sherwood@arm.com>
13820 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13821 calling tree_to_uhwi.
13823 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13824 Alan Hayward <alan.hayward@arm.com>
13825 David Sherwood <david.sherwood@arm.com>
13827 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13828 INTEGER_CST rather than a negative test for ADDR_EXPR.
13830 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13831 Alan Hayward <alan.hayward@arm.com>
13832 David Sherwood <david.sherwood@arm.com>
13834 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13835 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13837 2017-09-21 Richard Biener <rguenther@suse.de>
13839 PR tree-optimization/71351
13840 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13841 graphite_create_new_loop_guard): Remove, fold remaining parts
13843 (translate_isl_ast_node_for): ... here and simplify.
13845 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13848 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13849 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13850 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13851 alternative always use QI mode, for -Os imov (=R,R) alternative
13852 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13855 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13856 Jeff Law <law@redhat.com>
13858 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13859 (allocate_stack_space): New function, partially extracted from
13860 s390_emit_prologue.
13861 (s390_emit_prologue): Track offset to most recent stack probe.
13862 Code to allocate space moved into allocate_stack_space.
13863 Dump actions when no stack is allocated.
13864 (s390_prologue_plus_offset): New function.
13865 (s390_emit_stack_probe): Likewise.
13867 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13869 * common.opt (Wa, Wl, Wp, g, gz=): Add
13871 (gno-column-info): Remove.
13872 (gcolumn-info): Drop RejectNegative.
13873 (gno-): New prefix.
13874 (gno-record-gcc-switches): Remove.
13875 (grecord-gcc-switches): Drop RejectNegative.
13876 (gno-split-dwarf): Remove.
13877 (gsplit-dwarf): Drop RejectNegative.
13878 (gno-strict-dwarf): Remove.
13879 (gstrict-dwarf): Drop RejectNegative.
13880 * config/darwin.opt (gfull, gused): Add RejectNegative.
13881 * dwarf2out.c (gen_producer_string): Drop
13882 gno-record-gcc-switches handler.
13883 * optc-gen.awk: Add g to prefixes with negative forms.
13884 * opts-common.c (remapping_prefix_p): New.
13885 (find_opt): Check it.
13886 (generate_canonical_option): Test g prefix.
13887 (option_map): Add -gno- mapping.
13888 (add_misspelling_candidates): Check remapping_prefix_p.
13890 2017-09-20 Jeff Law <law@redhat.com>
13892 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13893 thinko in stack clash protection support.
13895 * explow.c (compute_stack_clash_protection_loop_data): Use
13896 CONST_INT_P instead of explicit test. Verify object is a
13897 CONST_INT_P before looking at INTVAL.
13898 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13899 instead of explicit test.
13901 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13904 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13905 address instead of to r1 and r11.
13907 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13909 * config.gcc: Support "knm".
13910 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13911 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13913 * config/i386/i386.c (m_KNM): Define.
13914 (processor_target_table): Add "knm".
13916 (ix86_option_override_internal): Add "knm".
13917 (ix86_issue_rate): Add PROCESSOR_KNM.
13918 (ix86_adjust_cost): Ditto.
13919 (ia32_multipass_dfa_lookahead): Ditto.
13920 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13921 (fold_builtin_cpu): Add M_INTEL_KNM.
13922 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13923 (processor_type): Add PROCESSOR_KNM.
13924 * config/i386/x86-tune.def: Add m_KNM.
13925 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13927 2017-09-20 Richard Biener <rguenther@suse.de>
13929 PR tree-optimization/80213
13930 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13931 are allowed in empty BBs as well.
13932 (canonicalize_loop_closed_ssa): Also look for other complex
13934 (scop_detection::get_sese): Include the loop-closed PHI block
13936 (scop_detection::merge_sese): Remove code adding extra blocks.
13937 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13938 (build_scops): Assert the final returned scop is invalid.
13940 2017-09-20 Richard Biener <rguenther@suse.de>
13942 PR tree-optimization/82264
13943 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13945 (vn_phi_lookup): Likewise.
13946 (vn_phi_insert): Likewise.
13948 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13950 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13951 that fits into uhwi or shwi, add DW_AT_const_value regardless
13952 of early_dwarf without going through RTL, using add_AT_unsigned
13955 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13956 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13957 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13958 (DEBUG_MACRO_SECTION): Likewise.
13959 (DEBUG_LTO_MACRO_SECTION): Likewise.
13960 (DEBUG_STR_DWO_SECTION): Likewise.
13961 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13962 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13963 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13964 (DEBUG_LTO_LINE_STR_SECTION): Define.
13965 (init_sections_and_labels): Initialize debug_line_str_section
13966 variable. Initialize debug_loc_section for -gdwarf-5 to
13967 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13969 2017-09-20 Richard Biener <rguenther@suse.de>
13971 * graphite-sese-to-poly.c (extract_affine): Properly handle
13972 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13974 2017-09-20 Richard Biener <rguenther@suse.de>
13976 PR tree-optimization/81373
13977 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13978 Force SESE live-out defs to be handled even if they are
13981 2017-09-19 Jeff Law <law@redhat.com>
13983 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13984 nothing for stack adjustments with REG_STACK_CHECK.
13985 * sched-deps.c (parse_add_or_inc): Reject insns with
13986 REG_STACK_CHECK from dependency breaking.
13987 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13988 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13989 * reg-notes.def (STACK_CHECK): New note.
13991 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13992 (ix86_expand_prologue): Dump stack clash info as needed.
13993 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13995 * function.c (dump_stack_clash_frame_info): New function.
13996 * function.h (dump_stack_clash_frame_info): Prototype.
13997 (enum stack_clash_probes): New enum.
13999 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14000 flag_stack_clash_protection.
14001 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14002 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14003 (arm_frame_pointer_required): Likewise.
14004 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14005 (ia64_expand_prologue): Likewise.
14006 * config/mips/mips.c (mips_expand_prologue): Likewise.
14007 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14008 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14009 (sparc_flat_expand_prologue): Likewise.
14010 * config/spu/spu.c (spu_expand_prologue): Likewise.
14012 * explow.c: Include "params.h".
14013 (anti_adjust_stack_and_probe_stack_clash): New function.
14014 (get_stack_check_protect): Likewise.
14015 (compute_stack_clash_protection_loop_data): Likewise.
14016 (emit_stack_clash_protection_loop_start): Likewise.
14017 (emit_stack_clash_protection_loop_end): Likewise.
14018 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14019 Use anti_adjust_stack_and_probe_stack_clash.
14020 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14021 (emit_stack_clash_protection_loop_start): Likewise.
14022 (emit_stack_clash_protection_loop_end): Likewise.
14023 * rtl.h (get_stack_check_protect): Prototype.
14024 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14025 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14026 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14028 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14030 * doc/tm.texi: Rebuilt.
14031 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14032 get_stack_check_protect.
14033 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14034 * config/arm/arm.c (arm_expand_prologue): Likewise.
14035 (arm_frame_pointer_required): Likewise.
14036 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14037 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14038 * config/mips/mips.c (mips_expand_prologue): Likewise.
14039 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14040 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14041 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14042 (sparc_flat_expand_prologue): Likewise.
14044 * common.opt (-fstack-clash-protection): New option.
14045 * flag-types.h (enum stack_check_type): Note difference between
14046 -fstack-check= and -fstack-clash-protection.
14047 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14048 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14049 * toplev.c (process_options): Issue warnings/errors for cases
14050 not handled with -fstack-clash-protection.
14051 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14052 (-fstack-check): Note additional problem with -fstack-check=generic.
14053 Note that -fstack-check is primarily for Ada and refer users
14054 to -fstack-clash-protection for stack-clash-protection.
14055 Document new params for stack clash protection.
14057 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14059 * config/i386/i386.c (ix86_split_long_move): Do not handle
14060 address used for LEA in a special way.
14062 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14064 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14066 2017-09-19 Martin Sebor <msebor@redhat.com>
14069 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14070 of incompatible types.
14072 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14074 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14075 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14076 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14077 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14079 2017-09-19 Richard Biener <rguenther@suse.de>
14081 PR tree-optimization/82244
14082 * tree-vrp.c (remove_range_assertions): Do not propagate
14083 a constant to abnormals but replace the assert with a copy.
14085 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14087 PR rtl-optimization/57878
14088 PR rtl-optimization/68988
14089 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14090 avoidance test involving non_reload_pseudos. Move frequency test
14091 below the general fragmentation avoidance test.
14093 2017-09-19 Richard Biener <rguenther@suse.de>
14095 PR tree-optimization/69728
14096 * graphite-sese-to-poly.c (schedule_error): New global.
14097 (add_loop_schedule): Handle empty domain by failing the
14099 (build_original_schedule): Handle schedule_error.
14101 2017-09-19 Richard Biener <rguenther@suse.de>
14103 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14104 Do not iterate to sibling loops but only to siblings of inner
14107 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14110 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14111 that the registers are different.
14113 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14115 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14116 to processor_model and "amdfam17h" to arch_names_table.
14117 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14119 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14122 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14124 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14125 Alan Hayward <alan.hayward@arm.com>
14126 David Sherwood <david.sherwood@arm.com>
14128 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14130 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14132 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14133 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14134 vect_schedule_slp_instance.
14135 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14136 Update call to vect_slp_analyze_node_operations. Simplify return
14138 (vect_slp_analyze_bb_1): Update call accordingly.
14139 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14140 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14141 (vect_schedule_slp): Update call accordingly.
14143 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14144 Alan Hayward <alan.hayward@arm.com>
14145 David Sherwood <david.sherwood@arm.com>
14147 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14148 with types that aren't in fact scalar.
14150 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14152 * tree-vect-slp.c (vect_record_max_nunits): New function,
14154 (vect_build_slp_tree_1): ...here.
14155 (vect_build_slp_tree_2): Call it for phis too.
14157 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14159 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14160 to vect_get_vec_def_for_operand when getting the mask operand.
14162 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14163 Alan Hayward <alan.hayward@arm.com>
14164 David Sherwood <david.sherwood@arm.com>
14166 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14169 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14170 Alan Hayward <alan.hayward@arm.com>
14171 David Sherwood <david.sherwood@arm.com>
14173 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14174 calculation for vector booleans.
14176 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14177 Alan Hayward <alan.hayward@arm.com>
14178 David Sherwood <david.sherwood@arm.com>
14180 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14182 (vect_transform_stmt): ...here.
14183 (vect_analyze_stmt): Use it instead of calling
14184 vectorizable_live_operation directly.
14186 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14188 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14189 non-SIMT targets in acc vector loops.
14191 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14193 * configure.ac: Add arc and check if assembler supports gdwarf2.
14194 * configure: Regenerate.
14196 2017-09-18 Richard Biener <rguenther@suse.de>
14198 PR tree-optimization/82220
14199 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14200 epilogue niters from the min_profitable_iters compute.
14202 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14205 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14206 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14207 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14209 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14212 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14213 switching to a new text section.
14215 2017-09-18 Richard Biener <rguenther@suse.de>
14217 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14219 (build_alias_set): Reject aliases with no access function.
14221 2017-09-18 Richard Biener <rguenther@suse.de>
14223 PR tree-optimization/79622
14224 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14226 (build_cross_bb_scalars_use): Likewise.
14228 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14230 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14232 2017-09-18 Alan Modra <amodra@gmail.com>
14235 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14236 stack_pointer_rtx for count 0. Update comments. Break up
14237 large rtl expression.
14239 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14241 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14242 Increase to 20 bytes.
14243 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14244 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14245 or avx version of the stub.
14247 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14250 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14251 compute the minimum stack alignment. Also update preferred stack
14252 boundary for leaf functions.
14254 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14256 PR tree-optimization/82228
14257 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14260 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14262 * common/config/nds32/nds32-common.c
14263 (nds32_option_optimization_table): Refine formatting.
14264 (nds32_option_optimization_table): Use -fsched-pressure and
14265 -fomit-frame-pointer for specific optimization level.
14267 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14269 * config/nds32/nds32.c: Refine formatting and comments.
14270 * config/nds32/nds32.h: Likewise.
14271 * config/nds32/nds32.md: Likewise.
14272 * config/nds32/nds32-cost.c: Likewise.
14273 * config/nds32/nds32-isr.c: Likewise.
14274 * config/nds32/nds32-md-auxiliary.c: Likewise.
14275 * config/nds32/nds32-multiple.md: Likewise.
14276 * config/nds32/nds32-predicates.c: Likewise.
14278 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14279 Jakub Jelinek <jakub@redhat.com>
14281 Add support for -std=c++2a.
14282 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14284 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14286 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14289 * doc/extend.texi (Common Function Attributes): Add
14290 references to ARM, AArch64, and S/390 specific attributes.
14291 (Function Specific Option Pragmas): Add AArch64 and S/390
14292 to list of back ends that support the target pragma.
14294 2017-09-15 Nathan Sidwell <nathan@acm.org>
14296 * doc/standards.texi: Fix C++17 description. Update URLs for
14299 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14301 * common.opt (Wcast-align=strict): New warning option.
14302 * doc/invoke.texi: Document -Wcast-align=strict.
14304 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14306 * cgraph.h (cgraph_thunk_info): Add comments.
14307 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14308 assert for VIRTUAL_* arguments stricter.
14310 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14312 PR tree-optimization/71026
14313 * match.pd: Move RDIV patterns from fold-const.c
14314 * fold-const.c (distribute_real_division): Removed.
14315 (fold_binary_loc): Remove calls to distribute_real_divison.
14317 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14319 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14320 c++1z and gnu++1z as deprecated. Change other references to
14321 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14322 Change -Wc++1z-compat to -Wc++17-compat.
14323 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14324 * dwarf2out.c (highest_c_language): Handle C++17.
14325 (gen_compile_unit_die): Likewise.
14327 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14329 PR rtl-optimization/82192
14330 * combine.c (make_extraction): Don't look through non-paradoxical
14331 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14334 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14335 Alan Hayward <alan.hayward@arm.com>
14336 David Sherwood <david.sherwood@arm.com>
14338 * target.def (function_arg_offset): New hook.
14339 * targhooks.h (default_function_arg_offset): Declare.
14340 * targhooks.c (default_function_arg_offset): New function.
14341 * function.c (locate_and_pad_parm): Use
14342 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14343 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14344 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14345 * doc/tm.texi: Regenerate.
14346 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14347 * config/spu/spu.c (spu_function_arg_offset): New function.
14348 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14349 * system.h (FUNCTION_ARG_OFFSET): Poison.
14351 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14352 Alan Hayard <alan.hayward@arm.com>
14353 David Sherwood <david.sherwood@arm.com>
14355 * target.def (truly_noop_truncation): New hook.
14356 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14357 than TRULY_NOOP_TRUNCATION.
14358 * hooks.h (hook_bool_uint_uint_true): Declare.
14359 * hooks.c (hook_bool_uint_uint_true): New function.
14360 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14361 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14362 * doc/tm.texi: Regenerate.
14363 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14364 rather than TRULY_NOOP_TRUNCATION in comments.
14365 (simplify_comparison): Likewise.
14366 (record_truncated_value): Likewise.
14367 * expmed.c (extract_bit_field_1): Likewise.
14368 (extract_split_bit_field): Likewise.
14369 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14370 instead of TRULY_NOOP_TRUNCATION.
14371 * function.c (assign_parm_setup_block): Likewise.
14372 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14373 * rtlhooks.c: Include target.h.
14374 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14375 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14376 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14377 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14378 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14379 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14380 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14381 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14382 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14383 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14384 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14385 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14386 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14387 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14388 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14389 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14390 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14391 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14392 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14393 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14394 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14395 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14396 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14397 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14398 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14399 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14400 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14401 rather than TRULY_NOOP_TRUNCATION in comments.
14402 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14403 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14404 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14405 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14406 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14407 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14408 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14409 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14410 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14411 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14412 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14413 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14414 rather than TRULY_NOOP_TRUNCATION in comments.
14415 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14416 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14417 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14418 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14419 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14420 TRULY_NOOP_TRUNCATION condition.
14421 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14422 (TRULY_NOOP_TRUNCATION): Delete.
14423 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14424 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14425 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14426 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14427 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14428 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14429 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14430 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14431 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14432 rather than TRULY_NOOP_TRUNCATION in comments.
14433 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14434 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14435 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14436 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14437 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14438 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14440 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14443 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14444 (*cmp_ior): Likewise.
14445 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14446 (*ior_scc_scc_cmp): Likewise.
14447 (*and_scc_scc): Likewise.
14448 (*and_scc_scc_cmp): Likewise.
14450 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14451 Alan Hayard <alan.hayward@arm.com>
14452 David Sherwood <david.sherwood@arm.com>
14454 * target.def (can_change_mode_class): New hook.
14455 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14456 (hard_regno_nregs): Likewise.
14457 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14458 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14459 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14460 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14461 (LOAD_EXTEND_OP): Update accordingly.
14462 * doc/tm.texi: Regenerate.
14463 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14464 CANNOT_CHANGE_MODE_CLASS.
14465 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14466 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14467 * combine.c (simplify_set): Update accordingly.
14468 * emit-rtl.c (validate_subreg): Likewise.
14469 * recog.c (general_operand): Likewise.
14470 * regcprop.c (mode_change_ok): Likewise.
14471 * reload1.c (choose_reload_regs): Likewise.
14472 (inherit_piecemeal_p): Likewise.
14473 * rtlanal.c (simplify_subreg_regno): Likewise.
14474 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14475 instead of CANNOT_CHANGE_MODE_CLASS.
14476 (reload_cse_simplify_operands): Likewise.
14477 * reload.c (push_reload): Use targetm.can_change_mode_class
14478 instead of CANNOT_CHANGE_MODE_CLASS.
14479 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14480 REG_CANNOT_CHANGE_MODE_P.
14481 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14482 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14483 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14484 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14485 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14486 (arm_can_change_mode_class): New function.
14487 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14488 than CANNOT_CHANGE_MODE_CLASS in comments.
14489 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14490 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14491 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14492 (ix86_can_change_mode_class): ...this new function, inverting the
14493 sense of the return value.
14494 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14495 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14496 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14497 (ia64_can_change_mode_class): New function.
14498 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14499 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14500 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14501 (m32c_can_change_mode_class): ...this new function, inverting the
14502 sense of the return value.
14503 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14504 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14505 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14506 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14507 (mips_can_change_mode_class): ...this new function, inverting the
14508 sense of the return value.
14509 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14510 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14511 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14512 (msp430_can_change_mode_class): New function.
14513 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14514 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14515 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14516 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14517 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14518 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14519 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14520 (pa_cannot_change_mode_class): Replace with...
14521 (pa_can_change_mode_class): ...this new function, inverting the
14522 sense of the return value.
14523 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14524 than CANNOT_CHANGE_MODE_CLASS in comments.
14525 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14526 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14527 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14528 (pdp11_cannot_change_mode_class): Replace with...
14529 (pdp11_can_change_mode_class): ...this new function, inverting the
14530 sense of the return value.
14531 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14532 * config/powerpcspe/powerpcspe-protos.h
14533 (rs6000_cannot_change_mode_class_ptr): Delete.
14534 * config/powerpcspe/powerpcspe.c
14535 (rs6000_cannot_change_mode_class_ptr): Delete.
14536 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14537 (rs6000_option_override_internal): Assign to
14538 targetm.can_change_mode_class instead of
14539 rs6000_cannot_change_mode_class_ptr.
14540 (rs6000_cannot_change_mode_class): Replace with...
14541 (rs6000_can_change_mode_class): ...this new function, inverting the
14542 sense of the return value.
14543 (rs6000_debug_cannot_change_mode_class): Replace with...
14544 (rs6000_debug_can_change_mode_class): ...this new function.
14545 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14546 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14547 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14548 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14549 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14551 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14552 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14553 (rs6000_option_override_internal): Assign to
14554 targetm.can_change_mode_class instead of
14555 rs6000_cannot_change_mode_class_ptr.
14556 (rs6000_cannot_change_mode_class): Replace with...
14557 (rs6000_can_change_mode_class): ...this new function, inverting the
14558 sense of the return value.
14559 (rs6000_debug_cannot_change_mode_class): Replace with...
14560 (rs6000_debug_can_change_mode_class): ...this new function.
14561 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14562 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14563 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14564 (s390_can_change_mode_class): ...this new function, inverting the
14565 sense of the return value.
14566 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14567 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14568 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14569 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14570 (sh_cannot_change_mode_class): Replace with...
14571 (sh_can_change_mode_class): ...this new function, inverting the
14572 sense of the return value.
14573 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14574 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14575 (sparc_can_change_mode_class): New function.
14576 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14577 * config/spu/spu.c (spu_can_change_mode_class): New function.
14578 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14579 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14580 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14581 (visium_can_change_mode_class): New function.
14582 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14584 2017-09-15 Richard Biener <rguenther@suse.de>
14586 PR tree-optimization/82217
14587 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14588 but not undefined case.
14590 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14593 * postreload.c (reload_cse_simplify_operands): Skip
14594 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14596 2017-09-15 Richard Biener <rguenther@suse.de>
14598 PR tree-optimization/68823
14599 * graphite-scop-detection.c (build_alias_set): If we have a
14600 possible dependence check whether we can handle them by just
14601 looking at the DRs DR_ACCESS_FNs.
14602 (build_scops): If build_alias_set fails, fail the SCOP.
14604 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14606 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14607 to support float128 built-in functions that require the ISA 3.0
14609 (BU_FLOAT128_3_HW): Likewise.
14610 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14611 built-in functions.
14612 (FMAF128): Likewise.
14614 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14615 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14616 floating point instructions.
14617 (rs6000_invalid_builtin): Likewise.
14618 (rs6000_builtin_mask_names): Likewise.
14619 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14620 (RS6000_BTM_FLOAT128_HW): Likewise.
14621 (RS6000_BTM_COMMON): Likewise.
14622 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14624 * doc/extend.texi (RS/6000 built-in functions): Document the
14625 IEEE 128-bit floating point square root and fused multiply-add
14626 built-in functions.
14628 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14630 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14631 reg (r2) isn't in the set of registers defined in the prologue.
14633 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14634 Alan Hayward <alan.hayward@arm.com>
14635 David Sherwood <david.sherwood@arm.com>
14637 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14638 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14639 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14640 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14641 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14643 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14644 max_vectorization_factor.
14645 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14647 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14648 Alan Hayward <alan.hayward@arm.com>
14649 David Sherwood <david.sherwood@arm.com>
14651 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14652 (vect_worthwhile_without_simd_p): Declare.
14653 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14654 (vectorizable_reduction): Use it.
14655 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14656 (vectorizable_operation): Likewise.
14658 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14659 Alan Hayward <alan.hayward@arm.com>
14660 David Sherwood <david.sherwood@arm.com>
14662 * tree-vectorizer.h (vect_get_num_copies): New function.
14663 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14664 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14665 (vectorizable_induction): Likewise.
14666 (vectorizable_live_operation): Likewise.
14667 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14668 (vectorizable_bswap): Likewise.
14669 (vectorizable_call): Likewise.
14670 (vectorizable_conversion): Likewise.
14671 (vectorizable_assignment): Likewise.
14672 (vectorizable_shift): Likewise.
14673 (vectorizable_operation): Likewise.
14674 (vectorizable_store): Likewise.
14675 (vectorizable_load): Likewise.
14676 (vectorizable_condition): Likewise.
14677 (vectorizable_comparison): Likewise.
14678 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14680 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14681 Alan Hayward <alan.hayward@arm.com>
14682 David Sherwood <david.sherwood@arm.com>
14684 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14685 of vect_init_vector.
14687 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14688 Alan Hayward <alan.hayward@arm.com>
14689 David Sherwood <david.sherwood@arm.com>
14691 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14692 an inline wrapper that provides a location.
14693 (gimple_build_vector): Likewise.
14694 * gimple-fold.c (gimple_build_vector_from_val): New function.
14695 (gimple_build_vector): Likewise.
14696 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14697 functions to build the initial value. Always return a gimple value.
14698 (get_initial_defs_for_reduction): Likewise. Only compute
14700 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14701 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14702 (vectorizable_induction): Use gimple_build_vector rather than
14705 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14706 Alan Hayward <alan.hayward@arm.com>
14707 David Sherwood <david.sherwood@arm.com>
14709 * target.h (vec_perm_indices): New typedef.
14710 (auto_vec_perm_indices): Likewise.
14711 * optabs-query.h: Include target.h
14712 (can_vec_perm_p): Take a vec_perm_indices *.
14713 * optabs-query.c (can_vec_perm_p): Likewise.
14714 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14715 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14716 * tree-vect-generic.c (lower_vec_perm): Likewise.
14717 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14718 (vect_grouped_load_supported): Likewise.
14719 (vect_shift_permute_load_chain): Likewise.
14720 (vect_permute_store_chain): Use auto_vec_perm_indices.
14721 (vect_permute_load_chain): Likewise.
14722 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14723 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14724 Update uses of can_vec_perm_p.
14725 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14726 mode with a number of elements. Take a vec_perm_indices *.
14727 (vect_create_epilog_for_reduction): Update accordingly.
14728 Use auto_vec_perm_indices.
14729 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14730 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14731 (vect_transform_slp_perm_load): Likewise.
14732 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14733 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14734 (vect_gen_perm_mask_checked): Likewise.
14735 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14736 (vect_gen_perm_mask_checked): Likewise.
14737 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14738 (vectorizable_store): Likewise.
14739 (vectorizable_load): Likewise.
14740 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14741 (vectorizable_bswap): Likewise.
14743 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14744 Alan Hayward <alan.hayward@arm.com>
14745 David Sherwood <david.sherwood@arm.com>
14747 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14748 * tree.c (build_vector): Likewise.
14749 (build_vector_from_ctor): Update accordingly.
14750 (build_vector_from_val): Likewise.
14751 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14752 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14753 * tree-vect-generic.c (add_rshift): Likewise.
14754 (expand_vector_divmod): Likewise.
14755 (optimize_vector_constructor): Likewise.
14756 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14757 (vect_transform_slp_perm_load): Likewise.
14758 (vect_schedule_slp_instance): Likewise.
14759 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14760 (vectorizable_call): Likewise.
14761 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14762 * expmed.c (make_tree): Likewise.
14763 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14764 a vector passed to build_vector.
14765 (fold_convert_const): Likewise.
14766 (exact_inverse): Likewise.
14767 (fold_ternary_loc): Likewise.
14768 (fold_relational_const): Likewise.
14769 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14770 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14771 (const_unop): Likewise. Store the reduction accumulator in a
14772 variable rather than an array.
14773 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14774 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14775 the new vector, rather than constructing it after the input arrays.
14776 (native_interpret_vector): Use auto_vec<tree> when building
14777 a vector passed to build_vector. Add elements in order.
14778 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14779 auto_vec<tree> when building a vector passed to build_vector.
14780 (vect_create_epilog_for_reduction): Likewise.
14781 (vectorizable_induction): Likewise.
14782 (get_initial_def_for_reduction): Likewise. Fix indentation of
14784 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14786 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14787 passed to build_vector.
14789 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14790 Alan Hayward <alan.hayward@arm.com>
14791 David Sherwood <david.sherwood@arm.com>
14793 * tree-core.h (tree_base::u): Add an "nelts" field.
14794 (tree_vector): Use VECTOR_CST_NELTS as the length.
14795 * tree.c (tree_size): Likewise.
14796 (make_vector): Initialize VECTOR_CST_NELTS.
14797 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14798 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14799 TYPE_VECTOR_SUBPARTS.
14800 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14801 as the number of units, setting it from VECTOR_CST_NELTS.
14802 (const_vector_from_tree): Likewise.
14803 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14804 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14805 (fold_negate_expr_1): Likewise.
14806 (fold_convert_const): Likewise.
14807 (const_binop): Likewise. Differentiate the number of output and
14809 (const_unop): Likewise.
14810 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14811 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14812 in cases that did the opposite.
14814 2017-09-14 Richard Biener <rguenther@suse.de>
14816 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14819 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14821 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14823 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14826 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14827 if and where to split a bb, except for splitting before debug insn
14828 sequences followed by non-label real insn. Delete debug insns
14829 in between basic blocks.
14831 * combine.c (make_compound_operation_int): Formatting fixes.
14833 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14834 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14835 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14836 * config/sol2.h (LINK_EH_SPEC): Likewise.
14837 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14838 * config/s390/linux.h (LINK_SPEC): Likewise.
14839 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14840 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14841 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14842 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14843 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14844 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14845 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14846 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14848 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14850 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14852 (ENDFILE_LINUX_SPEC): Likewise.
14853 (LINK_EH_SPEC): Likewise.
14854 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14855 (LINK_OS_LINUX_SPEC32): Likewise.
14856 (LINK_OS_LINUX_SPEC64): Likewise.
14857 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14858 (LINK_OS_LINUX_SPEC): Likewise.
14860 2017-09-13 Martin Liska <mliska@suse.cz>
14862 PR middle-end/82154
14863 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14864 CASE_HIGH is NULL_TREE.
14866 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14867 Alan Hayward <alan.hayward@arm.com>
14868 David Sherwood <david.sherwood@arm.com>
14870 * target.def (secondary_memory_needed): New hook.
14871 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14872 instead of SECONDARY_MEMORY_NEEDED.
14873 (secondary_memory_needed_mode): Likewise.
14874 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14875 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14876 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14877 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14878 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14879 * doc/tm.texi: Regenerate.
14880 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14881 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14882 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14883 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14884 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14885 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14886 mode argument first and change the reg_class arguments to reg_class_t.
14887 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14888 Make static. Update the call to inline_secondary_memory_needed.
14889 (ix86_register_move_cost): Update the call to
14890 inline_secondary_memory_needed.
14891 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14892 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14894 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14895 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14897 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14898 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14899 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14900 and match hook interface. Add comment from mips.h.
14901 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14902 * config/mmix/mmix.md (truncdfsf2): Refer to
14903 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14905 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14906 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14907 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14908 (pa_secondary_memory_needed): New function.
14909 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14910 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14911 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14912 (pdp11_secondary_memory_needed): Make static and match hook interface.
14913 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14914 * config/powerpcspe/powerpcspe-protos.h
14915 (rs6000_secondary_memory_needed_ptr): Delete.
14916 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14918 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14919 (rs6000_option_override_internal): Assign to
14920 targetm.secondary_memory_needed rather than
14921 rs6000_secondary_memory_needed_ptr.
14922 (rs6000_secondary_memory_needed): Match hook interface.
14923 (rs6000_debug_secondary_memory_needed): Likewise.
14924 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14925 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14926 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14927 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14928 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14929 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14931 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14932 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14933 (rs6000_option_override_internal): Assign to
14934 targetm.secondary_memory_needed rather than
14935 rs6000_secondary_memory_needed_ptr.
14936 (rs6000_secondary_memory_needed): Match hook interface.
14937 (rs6000_debug_secondary_memory_needed): Likewise.
14938 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14939 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14940 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14941 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14942 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14943 (sparc_secondary_memory_needed): New function.
14944 * lra-constraints.c (check_and_process_move): Refer to
14945 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14947 (curr_insn_transform): Likewise.
14948 (process_alt_operands): Use targetm.secondary_memory_needed
14949 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14950 (check_secondary_memory_needed_p): Likewise.
14951 (choose_split_class): Likewise.
14952 * reload.c: Unconditionally include code that was previously
14953 conditional on SECONDARY_MEMORY_NEEDED.
14954 (push_secondary_reload): Use targetm.secondary_memory_needed
14955 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14956 (push_reload): Likewise.
14957 * reload1.c: Unconditionally include code that was previously
14958 conditional on SECONDARY_MEMORY_NEEDED.
14959 (choose_reload_regs): Use targetm.secondary_memory_needed
14960 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14961 (gen_reload): Likewise.
14962 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14964 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14965 Alan Hayward <alan.hayward@arm.com>
14966 David Sherwood <david.sherwood@arm.com>
14968 * target.def (secondary_memory_needed_mode): New hook:
14969 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14970 * targhooks.h (default_secondary_memory_needed_mode): New function.
14971 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14972 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14973 * doc/tm.texi: Regenerate.
14974 * lra-constraints.c (check_and_process_move): Use
14975 targetm.secondary_memory_needed_mode instead of
14976 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14977 (curr_insn_transform): Likewise.
14978 * reload.c (get_secondary_mem): Likewise.
14979 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14980 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14982 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14983 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14984 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14985 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14986 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14988 * config/powerpcspe/powerpcspe-protos.h
14989 (rs6000_secondary_memory_needed_mode): Delete.
14990 * config/powerpcspe/powerpcspe.c
14991 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14992 (rs6000_secondary_memory_needed_mode): Make static.
14993 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14994 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14996 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14998 (rs6000_secondary_memory_needed_mode): Make static.
14999 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15000 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15001 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15002 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15003 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15005 (sparc_secondary_memory_needed_mode): New function.
15006 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15008 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15010 * config/aarch64/constraints.md (Umq): New constraint.
15011 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15013 (mov<mode>): Update condition.
15015 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15017 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15018 when bitposition is the same.
15020 2017-09-13 Richard Biener <rguenther@suse.de>
15022 * dwarf2out.c (output_die_symbol): Remove.
15023 (output_die): Do not output a DIEs symbol.
15025 2017-09-13 Richard Biener <rguenther@suse.de>
15027 PR middle-end/82128
15028 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15029 default-def to avoid breaking iterator update with the weird
15030 interaction with cgraph_update_edges_for_call_stmt_node.
15032 2017-09-13 Richard Biener <rguenther@suse.de>
15034 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15035 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15036 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15037 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15039 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15041 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15042 Disable pc relative literal load irrespective of
15043 TARGET_FIX_ERR_A53_84341 for default.
15045 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15047 * config/sparc/sparc.c (output_return): Output the source location of
15048 the insn in the delay slot, if any.
15049 (output_sibcall): Likewise.
15051 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15054 * common.opt (-static-pie): New alias.
15055 (shared): Negate static-pie.
15056 (-no-pie): Update help text.
15058 (static-pie): New option.
15059 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15060 -static-pie support.
15061 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15062 (LINK_EH_SPEC): Likewise.
15063 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15064 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15065 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15066 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15067 (init_gcc_specs): Likewise.
15068 (init_spec): Likewise.
15069 (display_help): Update help message for -pie.
15070 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15073 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15075 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15076 (movdi_aarch64): Likewise.
15077 (movti_aarch64): Likewise.
15079 2017-09-12 Simon Wright <simon@pushface.org>
15082 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15083 calculation of the minor version, always output as 0.
15085 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15088 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15089 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15090 on it early, rather than manual conversion late. For
15091 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15092 instead of performing manual conversion.
15094 2017-09-12 Carl Love <cel@us.ibm.com>
15096 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15097 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15099 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15100 VMULOSW): Add definitions.
15101 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15102 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15103 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15104 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15105 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15107 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15109 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15111 (movti_aarch64): Likewise.
15112 (movdf_aarch64): Likewise.
15113 (movtf_aarch64): Likewise.
15114 (load_pairdi): Likewise.
15115 (store_pairdi): Likewise.
15116 (load_pairdf): Likewise.
15117 (store_pairdf): Likewise.
15118 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15119 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15120 (ldr_got_small_<mode>): Likewise.
15121 (ldr_got_small_28k_<mode>): Likewise.
15122 (ldr_got_tiny): Likewise.
15123 * config/aarch64/iterators.md (ldst_sz): New.
15124 (ldpstp_sz): Likewise.
15125 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15127 (thunderx_load): Split load_8 to load_16.
15128 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15130 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15131 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15132 (xgene1_store_pair): Split store_8 to store_16.
15133 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15134 (falkor_st_0_st_sd): Split store_8 to store_16.
15136 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15138 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15139 and store1/2/3/4 to store_4/8/12/16.
15140 * config/aarch64/aarch64.md: Update for rename.
15141 * config/arm/arm.md: Likewise.: Likewise.
15142 * config/arm/arm.c: Likewise.
15143 * config/arm/thumb1.md: Likewise.
15144 * config/arm/thumb2.md: Likewise.
15145 * config/arm/vfp.md: Likewise.
15146 * config/arm/arm-generic.md: Likewise.
15147 * config/arm/arm1020e.md: Likewise.
15148 * config/arm/arm1026ejs.md: Likewise.
15149 * config/arm/arm1136jfs.md: Likewise.
15150 * config/arm/arm926ejs.md: Likewise.
15151 * config/arm/cortex-a15.md: Likewise.
15152 * config/arm/cortex-a17.md: Likewise.
15153 * config/arm/cortex-a5.md: Likewise.
15154 * config/arm/cortex-a53.md: Likewise.
15155 * config/arm/cortex-a57.md: Likewise.
15156 * config/arm/cortex-a7.md: Likewise.
15157 * config/arm/cortex-a8.md: Likewise.
15158 * config/arm/cortex-a9.md: Likewise.
15159 * config/arm/cortex-m4.md: Likewise.
15160 * config/arm/cortex-m7.md: Likewise.
15161 * config/arm/cortex-r4.md: Likewise.
15162 * config/arm/exynos-m1.md: Likewise.
15163 * config/arm/fa526.md: Likewise.
15164 * config/arm/fa606te.md: Likewise.
15165 * config/arm/fa626te.md: Likewise.
15166 * config/arm/fa726te.md: Likewise.
15167 * config/arm/fmp626.md: Likewise.
15168 * config/arm/iwmmxt.md: Likewise.
15169 * config/arm/ldmstm.md: Likewise.
15170 * config/arm/marvell-pj4.md: Likewise.
15171 * config/arm/xgene1.md: Likewise.
15172 * config/aarch64/thunderx.md: Likewise.
15173 * config/aarch64/thunderx2t99.md: Likewise.
15174 * config/aarch64/falkor.md: Likewise.
15176 2017-09-12 Martin Liska <mliska@suse.cz>
15178 * attribs.c (private_lookup_attribute): New function.
15179 * attribs.h (private_lookup_attribute): Declared here.
15180 (lookup_attribute): Called from this place.
15182 2017-09-12 Richard Biener <rguenther@suse.de>
15184 PR tree-optimization/82157
15185 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15186 stmts with side-effects.
15188 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15189 Alan Hayward <alan.hayward@arm.com>
15190 David Sherwood <david.sherwood@arm.com>
15192 * target.def (hard_regno_nregs): New hook.
15193 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15194 * targhooks.h (default_hard_regno_nregs): Declare.
15195 * targhooks.c (default_hard_regno_nregs): New function.
15196 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15197 (TARGET_HARD_REGNO_NREGS): ...this hook.
15198 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15199 (CLASS_MAX_NREGS): Likewise.
15200 * doc/tm.texi: Regenerate.
15201 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15202 instead of HARD_REGNO_NREGS.
15203 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15204 HARD_REGNO_NREGS in the comment.
15205 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15206 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15207 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15208 Return an unsigned int.
15209 (TARGET_HARD_REGNO_NREGS): Redefine.
15210 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15211 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15212 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15213 (arc_hard_regno_nregs): New function.
15214 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15215 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15216 (arm_hard_regno_nregs): New function.
15217 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15218 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15219 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15220 (TARGET_HARD_REGNO_NREGS): Redefine.
15221 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15222 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15223 (HARD_REGNO_NREGS): Delete.
15224 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15225 (cr16_hard_regno_nregs): New function.
15226 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15227 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15228 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15229 (cris_hard_regno_nregs): New function.
15230 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15231 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15232 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15233 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15234 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15235 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15236 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15237 (frv_hard_regno_nregs): Make static. Take and return an
15239 (frv_class_max_nregs): Remove outdated copy of documentation.
15240 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15241 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15242 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15243 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15244 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15245 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15246 (TARGET_HARD_REGNO_NREGS): Redefine.
15247 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15248 (CLASS_MAX_NREGS): Update comment.
15249 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15250 (ia64_hard_regno_nregs): New function.
15251 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15252 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15253 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15254 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15255 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15257 (m32c_hard_regno_nregs): Likewise. Make static.
15258 (TARGET_HARD_REGNO_NREGS): Redefine.
15259 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15260 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15261 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15262 (m68k_hard_regno_nregs): New function.
15263 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15264 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15265 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15266 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15267 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15268 Take and return an unsigned int.
15269 (TARGET_HARD_REGNO_NREGS): Redefine.
15270 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15271 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15272 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15273 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15274 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15275 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15276 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15277 (msp430_hard_regno_nregs): Make static. Take and return an
15279 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15280 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15281 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15282 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15283 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15284 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15285 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15286 (TARGET_HARD_REGNO_NREGS): Redefine.
15287 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15288 (PA_HARD_REGNO_NREGS): ...this.
15289 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15290 (PA_HARD_REGNO_NREGS): ...this.
15291 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15292 (pa_hard_regno_nregs): New function.
15293 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15294 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15295 (pdp11_hard_regno_nregs): New function.
15296 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15297 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15298 (rs6000_hard_regno_nregs_hook): New function.
15299 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15300 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15301 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15302 Take and return an unsigned int. Move earlier in file.
15303 (TARGET_HARD_REGNO_NREGS): Redefine.
15304 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15305 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15306 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15307 (rl78_hard_regno_nregs): Make static. Take and return an
15309 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15310 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15311 (rs6000_hard_regno_nregs_hook): New function.
15312 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15313 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15314 (TARGET_HARD_REGNO_NREGS): Redefine.
15315 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15316 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15317 instead of HARD_REGNO_NREGS.
15318 (s390_hard_regno_nregs): New function.
15319 (s390_hard_regno_mode_ok): Add comment from s390.h.
15320 (TARGET_HARD_REGNO_NREGS): Redefine.
15321 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15322 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15323 (sh_hard_regno_nregs): New function.
15324 (sh_pass_in_reg_p): Use it.
15325 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15326 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15327 (sparc_hard_regno_nregs): New function.
15328 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15329 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15330 (spu_function_arg_advance): Use it, supplying a valid register number.
15331 (TARGET_HARD_REGNO_NREGS): Redefine.
15332 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15333 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15334 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15335 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15336 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15337 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15338 (CLASS_MAX_NREGS): Remove copy of old documentation.
15339 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15340 (visium_hard_regno_nregs): New function.
15341 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15342 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15343 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15344 xtensa_hard_regno_nregs): New function.
15345 * system.h (HARD_REGNO_NREGS): Poison.
15347 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15349 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15350 hard_regno_nregs instead of HARD_REGNO_NREGS.
15351 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15352 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15353 (c6x_expand_epilogue): Likewise.
15354 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15355 (frv_read_iacc_argument): Likewise.
15356 * config/sh/sh.c: Include regs.h.
15357 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15358 (regs_used): Likewise.
15359 (output_stack_adjust): Likewise.
15360 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15361 * expmed.c: Include regs.h.
15362 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15363 * ree.c: Include regs.h.
15364 (combine_reaching_defs): Use hard_regno_nregs instead of
15366 (add_removable_extension): Likewise.
15368 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15370 * regs.h (hard_regno_nregs): Turn into a function.
15371 (end_hard_regno): Update accordingly.
15372 * caller-save.c (setup_save_areas): Likewise.
15373 (save_call_clobbered_regs): Likewise.
15374 (replace_reg_with_saved_mem): Likewise.
15375 (insert_restore): Likewise.
15376 (insert_save): Likewise.
15377 * combine.c (can_change_dest_mode): Likewise.
15378 (move_deaths): Likewise.
15379 (distribute_notes): Likewise.
15380 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15381 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15382 (rs6000_split_multireg_move): Likewise.
15383 (rs6000_register_move_cost): Likewise.
15384 (rs6000_memory_move_cost): Likewise.
15385 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15386 (rs6000_split_multireg_move): Likewise.
15387 (rs6000_register_move_cost): Likewise.
15388 (rs6000_memory_move_cost): Likewise.
15389 * cselib.c (cselib_reset_table): Likewise.
15390 (cselib_lookup_1): Likewise.
15391 * emit-rtl.c (set_mode_and_regno): Likewise.
15392 * function.c (aggregate_value_p): Likewise.
15393 * ira-color.c (setup_profitable_hard_regs): Likewise.
15394 (check_hard_reg_p): Likewise.
15395 (calculate_saved_nregs): Likewise.
15396 (assign_hard_reg): Likewise.
15397 (improve_allocation): Likewise.
15398 (calculate_spill_cost): Likewise.
15399 * ira-emit.c (modify_move_list): Likewise.
15400 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15401 (ira_hard_reg_in_set_p): Likewise.
15402 * ira.c (setup_reg_mode_hard_regset): Likewise.
15403 (clarify_prohibited_class_mode_regs): Likewise.
15404 (check_allocation): Likewise.
15405 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15406 (lra_setup_reg_renumber): Likewise.
15407 (setup_try_hard_regno_pseudos): Likewise.
15408 (spill_for): Likewise.
15409 (assign_hard_regno): Likewise.
15410 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15411 * lra-constraints.c (in_class_p): Likewise.
15412 (lra_constraint_offset): Likewise.
15413 (simplify_operand_subreg): Likewise.
15414 (lra_constraints): Likewise.
15415 (split_reg): Likewise.
15416 (split_if_necessary): Likewise.
15417 (invariant_p): Likewise.
15418 (inherit_in_ebb): Likewise.
15419 * lra-lives.c (process_bb_lives): Likewise.
15420 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15421 (get_hard_regs): Likewise.
15422 (do_remat): Likewise.
15423 * lra-spills.c (assign_spill_hard_regs): Likewise.
15424 * mode-switching.c (create_pre_exit): Likewise.
15425 * postreload.c (reload_combine_recognize_pattern): Likewise.
15426 * recog.c (peep2_find_free_register): Likewise.
15427 * regcprop.c (kill_value_regno): Likewise.
15428 (set_value_regno): Likewise.
15429 (copy_value): Likewise.
15430 (maybe_mode_change): Likewise.
15431 (find_oldest_value_reg): Likewise.
15432 (copyprop_hardreg_forward_1): Likewise.
15433 * regrename.c (check_new_reg_p): Likewise.
15434 (regrename_do_replace): Likewise.
15435 * reload.c (push_reload): Likewise.
15436 (combine_reloads): Likewise.
15437 (find_dummy_reload): Likewise.
15438 (operands_match_p): Likewise.
15439 (find_reloads): Likewise.
15440 (find_equiv_reg): Likewise.
15441 (reload_adjust_reg_for_mode): Likewise.
15442 * reload1.c (count_pseudo): Likewise.
15443 (count_spilled_pseudo): Likewise.
15444 (find_reg): Likewise.
15445 (clear_reload_reg_in_use): Likewise.
15446 (free_for_value_p): Likewise.
15447 (allocate_reload_reg): Likewise.
15448 (choose_reload_regs): Likewise.
15449 (reload_adjust_reg_for_temp): Likewise.
15450 (emit_reload_insns): Likewise.
15451 (delete_output_reload): Likewise.
15452 * rtlanal.c (subreg_get_info): Likewise.
15453 * sched-deps.c (sched_analyze_reg): Likewise.
15454 * sel-sched.c (init_regs_for_mode): Likewise.
15455 (mark_unavailable_hard_regs): Likewise.
15456 (choose_best_reg_1): Likewise.
15457 (verify_target_availability): Likewise.
15458 * valtrack.c (dead_debug_insert_temp): Likewise.
15459 * var-tracking.c (track_loc_p): Likewise.
15460 (emit_note_insn_var_location): Likewise.
15461 * varasm.c (make_decl_rtl): Likewise.
15462 * reginfo.c (choose_hard_reg_mode): Likewise.
15463 (init_reg_modes_target): Refer directly to
15464 this_target_regs->x_hard_regno_nregs.
15466 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15468 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15469 instead of hard_regno_nregs.
15471 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15473 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15474 end_hard_regno instead of hard_regno_nregs.
15475 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15476 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15477 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15478 * ira-color.c (improve_allocation): Likewise.
15479 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15480 * lra-lives.c (mark_regno_live): Likewise.
15481 (mark_regno_dead): Likewise.
15482 * lra-remat.c (operand_to_remat): Likewise.
15483 * lra.c (collect_non_operand_hard_regs): Likewise.
15484 * postreload.c (reload_combine_note_store): Likewise.
15485 (move2add_valid_value_p): Likewise.
15486 * reload.c (regno_clobbered_p): Likewise.
15488 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15490 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15492 * config/v850/v850.c (v850_reorg): Likewise.
15493 * reload.c (refers_to_regno_for_reload_p): Likewise.
15494 (find_equiv_reg): Likewise.
15495 * reload1.c (reload_reg_reaches_end_p): Likewise.
15497 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15499 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15501 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15502 * config/arm/arm.c (output_move_neon): Likewise.
15503 (arm_attr_length_move_neon): Likewise.
15504 (neon_split_vcombine): Likewise.
15505 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15506 (c6x_mark_reg_written): Likewise.
15507 (c6x_dwarf_register_span): Likewise.
15508 * config/i386/i386.c (ix86_save_reg): Likewise.
15509 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15510 (rws_access_reg): Likewise.
15511 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15512 * mode-switching.c (create_pre_exit): Likewise.
15513 * ree.c (combine_reaching_defs): Likewise.
15514 (add_removable_extension): Likewise.
15515 * regcprop.c (find_oldest_value_reg): Likewise.
15516 (copyprop_hardreg_forward_1): Likewise.
15517 * reload.c (reload_inner_reg_of_subreg): Likewise.
15518 (push_reload): Likewise.
15519 (combine_reloads): Likewise.
15520 (find_dummy_reload): Likewise.
15521 (reload_adjust_reg_for_mode): Likewise.
15522 * reload1.c (find_reload_regs): Likewise.
15523 (forget_old_reloads_1): Likewise.
15524 (reload_reg_free_for_value_p): Likewise.
15525 (reload_adjust_reg_for_temp): Likewise.
15526 (emit_reload_insns): Likewise.
15527 (delete_output_reload): Likewise.
15528 * sel-sched.c (choose_best_reg_1): Likewise.
15529 (choose_best_pseudo_reg): Likewise.
15531 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15532 Alan Hayward <alan.hayward@arm.com>
15533 David Sherwood <david.sherwood@arm.com>
15535 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15536 * target.def (slow_unaligned_access): New hook.
15537 * targhooks.h (default_slow_unaligned_access): Declare.
15538 * targhooks.c (default_slow_unaligned_access): New function.
15539 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15540 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15541 * doc/tm.texi: Regenerate.
15542 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15543 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15544 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15546 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15547 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15549 (rs6000_slow_unaligned_access): New function.
15550 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15551 (expand_block_compare): Likewise.
15552 (expand_strn_compare): Likewise.
15553 (rs6000_rtx_costs): Likewise.
15554 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15555 (riscv_slow_unaligned_access): Likewise.
15556 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15557 (riscv_slow_unaligned_access_p): ...this and make static.
15558 (riscv_option_override): Update accordingly.
15559 (riscv_slow_unaligned_access): New function.
15560 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15561 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15562 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15563 (rs6000_slow_unaligned_access): New function.
15564 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15565 (rs6000_rtx_costs): Likewise.
15566 * config/rs6000/rs6000-string.c (expand_block_compare)
15567 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15568 of SLOW_UNALIGNED_ACCESS.
15569 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15570 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15571 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15572 of SLOW_UNALIGNED_ACCESS.
15573 * expmed.c (simple_mem_bitfield_p): Likewise.
15574 * expr.c (alignment_for_piecewise_move): Likewise.
15575 (emit_group_load_1): Likewise.
15576 (emit_group_store): Likewise.
15577 (copy_blkmode_from_reg): Likewise.
15578 (emit_push_insn): Likewise.
15579 (expand_assignment): Likewise.
15580 (store_field): Likewise.
15581 (expand_expr_real_1): Likewise.
15582 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15583 * lra-constraints.c (simplify_operand_subreg): Likewise.
15584 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15585 * gimple-ssa-store-merging.c: Likewise in block comment at start
15587 * tree-ssa-strlen.c: Include target.h.
15588 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15589 of SLOW_UNALIGNED_ACCESS.
15590 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15592 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15594 PR rtl-optimization/82185
15595 * expmed.c (emit_store_flag_int): Only test tem if it has been
15598 2017-09-12 Richard Biener <rguenther@suse.de>
15600 PR middle-end/82149
15601 * match.pd ((FTYPE) N CMP CST): Fix typo.
15603 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15605 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15607 (mips_near_type_p): Add 'short_call' attribute as a synonym
15609 * doc/extend.texi (short_call): Document new function attribute.
15611 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15614 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15615 assertion check that in the condition.
15616 (get_atomic_generic_size): Likewise. Before testing if parameter
15617 has pointer type, if it has array type, call for C++
15618 default_conversion to perform array-to-pointer conversion.
15620 2017-09-12 Richard Biener <rguenther@suse.de>
15622 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15623 for operations we cannot scalarize.
15625 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15627 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15628 vectors heap vectors. Clean up comments.
15629 Make visited_bbs a reference.
15630 (profitable_jump_thread_path): Make GC
15631 vectors heap vectors. Clean up comments.
15633 (convert_and_register_jump_thread_path): Make GC vectors heap
15635 (check_subpath_and_update_thread_path): Same. Clean up comments.
15636 Make visited_bbs a reference.
15637 (handle_phi): Abstract common code to to
15638 register_jump_thread_path_if_profitable.
15639 Rename VAR_BB to DEF_BB.
15641 Make GC vectors heap vectors.
15642 Make visited_bbs a reference.
15643 (handle_assignment): Same.
15644 (register_jump_thread_path_if_profitable): New.
15645 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15647 Make GC vectors heap vectors. Clean up comments.
15648 Make visited_bbs a reference.
15649 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15650 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15653 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15656 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15657 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15659 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15661 Revert r251800 and r251799.
15663 2017-09-11 Martin Jambor <mjambor@suse.cz>
15666 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15667 arguments in advance.
15668 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15669 use it to find predecessor edges.
15670 (naive_outof_ssa): Collect vector of predecessors.
15672 2017-09-08 Jason Merrill <jason@redhat.com>
15674 PR c++/70029 - ICE with ref-qualifier and -flto
15675 * langhooks.h (struct lang_hooks_for_types): Add
15676 copy_lang_qualifiers.
15677 * attribs.c (build_type_attribute_qual_variant): Use it.
15678 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15680 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15681 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15683 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15686 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15687 (*mulsi3_sp64): New instruction.
15688 (mulsi3): New expander.
15690 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15692 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15694 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15696 * sancov.c: Include memmodel.h.
15698 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15701 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15704 2017-09-07 Carl Love <cel@us.ibm.com>
15706 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15707 the sldi instruction.
15709 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15711 * sancov.c: Include tm_p.h.
15713 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15716 * output.h (switch_to_other_text_partition): New declaration.
15717 * varasm.c (switch_to_other_text_partition): New function.
15718 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15719 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15720 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15721 to the other text partition before emitting LCL label and switch back
15722 after emitting the word after it.
15724 2017-09-07 Richard Biener <rguenther@suse.de>
15726 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15727 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15728 critical edges here, after loop init.
15729 (pass_data_pre): Remove PROP_no_crit_edges flags.
15730 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15731 for valueization of call args to avoid leaking VN_TOP.
15732 (visit_use): Assert we do not visit default defs.
15733 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15734 Use error_mark_node to more easily detect leaking VN_TOP.
15735 All default-defs are varying, not VN_TOP. Mark them visited.
15736 (run_scc_vn): Make code match comment.
15738 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15740 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15741 OPTION_MASK_FLOAT128_KEYWORD.
15742 (POWERPC_MASKS): Likewise.
15743 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15744 support for the -mfloat128-type option, and make -mfloat128
15745 default on PowerPC Linux systems. Define or undefine
15746 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15747 Define __float128 to be __ieee128 if IEEE 128-bit support is
15748 enabled, or undefine it.
15749 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15750 Delete defining __FLOAT128_TYPE__.
15751 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15752 -mfloat128-type option and make -mfloat128 default on PowerPC
15754 (TARGET_FLOAT128_TYPE): Likewise.
15755 (-mfloat128-type): Likewise.
15756 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15757 Delete the -mfloat128-type option and make -mfloat128 default on
15758 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15759 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15760 128-bit floating point is enabled. Change tests from using
15761 -mfloat128-type to -mfloat128.
15762 (rs6000_mangle_type): Use the correct mangling for the __float128
15763 type even if normal long double is restricted to 64-bits.
15764 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15766 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15767 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15768 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15770 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15771 documentation for -mfloat128.
15773 2017-09-06 Olivier Hainque <hainque@adacore.com>
15775 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15777 2017-09-06 Wish Wu <wishwu007@gmail.com>
15778 Jakub Jelinek <jakub@redhat.com>
15780 * asan.c (initialize_sanitizer_builtins): Add
15781 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15782 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15783 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15784 BT_FN_VOID_UINT64_PTR variables.
15785 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15786 (BT_FN_VOID_UINT16_UINT16): Likewise.
15787 (BT_FN_VOID_UINT32_UINT32): Likewise.
15788 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15789 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15790 (BT_FN_VOID_UINT64_PTR): Likewise.
15791 * common.opt (flag_sanitize_coverage): New variable.
15792 (fsanitize-coverage=trace-pc): Remove.
15793 (fsanitize-coverage=): Add.
15794 * flag-types.h (enum sanitize_coverage_code): New enum.
15795 * fold-const.c (fold_range_test): Disable non-short-circuit
15796 optimization if flag_sanitize_coverage.
15797 (fold_truth_andor): Likewise.
15798 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15799 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15800 (coverage_sanitizer_opts): New array.
15801 (get_closest_sanitizer_option): Add OPTS argument, handle also
15802 OPT_fsanitize_coverage_.
15803 (parse_sanitizer_options): Adjusted to also handle
15804 OPT_fsanitize_coverage_.
15805 (common_handle_option): Add OPT_fsanitize_coverage_.
15806 * sancov.c (instrument_comparison, instrument_switch): New function.
15807 (sancov_pass): Add trace-cmp support.
15808 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15809 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15810 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15811 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15812 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15813 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15814 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15815 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15816 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15817 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15819 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15821 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15822 error. Only quit immediately if parsing is complete.
15823 (BEGIN): Initialize fatal_err and parse_done.
15824 (begin fpu, end fpu): Check number of arguments.
15825 (begin arch, end arch): Likewise.
15826 (begin cpu, end cpu): Likewise.
15827 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15828 (optalias): Likewise.
15830 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15832 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15833 * config/arm/arm-isa.h: Delete. Move definitions to ...
15834 * arm-cpus.in: ... here. Use new feature and fgroup values.
15835 * config/arm/arm.c (arm_option_override): Use lower case for feature
15837 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15838 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15839 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15841 (print_isa_bits_for): New function.
15842 (gen_isa): New function.
15843 (gen_comm_data): Use print_isa_bits_for.
15844 (define feature): New keyword.
15845 (define fgroup): New keyword.
15846 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15847 (arm-isa.h): Add rule to generate file.
15848 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15849 case for feature bit names.
15851 2017-09-06 Richard Biener <rguenther@suse.de>
15853 * tree-ssa-pre.c (NECESSARY): Remove.
15854 (create_expression_by_pieces): Do not touch pass-local flags.
15855 (insert_into_preds_of_block): Likewise.
15856 (do_pre_regular_insertion): Likewise.
15857 (eliminate_insert): Likewise.
15858 (eliminate_dom_walker::before_dom_children): Likewise.
15859 (fini_eliminate): Do not look at inserted_exprs.
15860 (mark_operand_necessary): Remove.
15861 (remove_dead_inserted_code): Replace with simple work-list
15862 algorithm based on inserted_exprs and SSA uses.
15863 (pass_pre::execute): Re-order fini_eliminate and
15864 remove_dead_inserted_code.
15866 2017-09-06 Olivier Hainque <hainque@adacore.com>
15868 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15869 for VxWorks 7. Adjust surrounding comments.
15871 2017-09-06 Richard Biener <rguenther@suse.de>
15873 * gimple-ssa-strength-reduction.c
15874 (find_candidates_dom_walker::before_dom_children): Also allow
15877 2017-09-06 Richard Biener <rguenther@suse.de>
15879 PR tree-optimization/82108
15880 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15881 for gap in the non-permutation SLP case.
15883 2017-09-06 Martin Jambor <mjambor@suse.cz>
15885 PR tree-optimization/82078
15886 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15887 add_access_to_work_queue...
15888 (build_accesses_from_assign): ...here.
15889 (propagate_all_subaccesses): Make sure racc is the group
15890 representative, if there is one.
15892 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15894 PR middle-end/82095
15895 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15898 2017-09-06 Richard Biener <rguenther@suse.de>
15900 * gimple-ssa-strength-reduction.c
15901 (find_candidates_dom_walker::before_doom_children): Use a
15902 type and not a mode check.
15904 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15907 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15909 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15911 2017-09-05 Jeff Law <law@redhat.com>
15913 PR tree-optimization/64910
15914 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15915 cases where we have 3 or more operands.
15917 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15919 PR middle-end/81768
15920 * omp-low.c (lower_omp_for): Recompute tree invariant if
15921 gimple_omp_for_initial/final is ADDR_EXPR.
15923 PR middle-end/81768
15924 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15925 into gimple val before gimplification fo the COND_EXPR.
15927 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15929 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15930 REGION_COPY argument.
15931 (thread_through_all_blocks): Remove unused argument to
15932 duplicate_thread_path.
15934 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15935 Alan Hayward <alan.hayward@arm.com>
15936 David Sherwood <david.sherwood@arm.com>
15938 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15939 Take a scalar_mode rather than a machine_mode.
15940 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15941 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15942 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15943 (aarch64_gen_adjusted_ldpstp): Likewise.
15944 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15946 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15947 Alan Hayward <alan.hayward@arm.com>
15948 David Sherwood <david.sherwood@arm.com>
15950 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15951 Take a scalar_int_mode instead of a machine_mode.
15952 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15953 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15954 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15955 (aarch64_simd_attr_length_rglist): Delete.
15956 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15957 a scalar_int_mode instead of a machine_mode.
15958 (aarch64_add_offset): Likewise.
15959 (aarch64_internal_mov_immediate): Likewise
15960 (aarch64_add_constant_internal): Likewise.
15961 (aarch64_add_constant): Likewise.
15962 (aarch64_movw_imm): Likewise.
15963 (aarch64_rtx_arith_op_extract_p): Likewise.
15964 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15965 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15966 Remove assert that the mode isn't a vector.
15967 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15968 (aarch64_expand_mov_immediate): Update calls after above changes.
15969 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15970 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15971 (aarch64_move_imm): Likewise.
15972 (aarch64_can_const_movi_rtx_p): Likewise.
15973 (aarch64_strip_extend): Likewise.
15974 (aarch64_extr_rtx_p): Likewise.
15975 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15976 a CONST_INT when the mode parameter is VOIDmode.
15977 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15979 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15981 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15982 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15983 (bitwise_type_for_mode): Update accordingly.
15985 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15987 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15988 * stor-layout.c (mode_for_size_tree): Likewise.
15989 (mode_for_array): Update accordingly.
15990 (layout_decl): Likewise.
15991 (compute_record_mode): Likewise. Only set the mode once.
15993 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15995 * target.def (get_mask_mode): Change return type to opt_mode.
15997 * doc/tm.texi: Regenerate.
15998 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15999 * targhooks.c (default_get_mask_mode): Likewise.
16000 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16001 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16002 targetm.get_mask_mode.
16003 * tree.c (build_truth_vector_type): Likewise.
16005 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16007 * machmode.h (mode_for_vector): Return an opt_mode.
16008 * stor-layout.c (mode_for_vector): Likewise.
16009 (mode_for_int_vector): Update accordingly.
16010 (layout_type): Likewise.
16011 * config/i386/i386.c (emit_memmov): Likewise.
16012 (ix86_expand_set_or_movmem): Likewise.
16013 (ix86_expand_vector_init): Likewise.
16014 (ix86_get_mask_mode): Likewise.
16015 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16017 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16018 * expmed.c (extract_bit_field_1): Likewise.
16019 * expr.c (expand_expr_real_2): Likewise.
16020 * optabs-query.c (can_vec_perm_p): Likewise.
16021 (can_vec_mask_load_store_p): Likewise.
16022 * optabs.c (expand_vec_perm): Likewise.
16023 * targhooks.c (default_get_mask_mode): Likewise.
16024 * tree-vect-stmts.c (vectorizable_store): Likewise.
16025 (vectorizable_load): Likewise.
16026 (get_vectype_for_scalar_type_and_size): Likewise.
16028 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16030 * machmode.h (mode_for_int_vector): New function.
16031 * stor-layout.c (mode_for_int_vector): Likewise.
16032 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16033 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16034 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16035 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16036 (s390_expand_vcond): Likewise.
16038 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16040 * machmode.h (opt_machine_mode): New type.
16041 (opt_mode<T>): Allow construction from anything that can be
16043 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16044 (mode_for_size): Return an opt_machine_mode.
16045 * stor-layout.c (mode_for_size): Likewise.
16046 (mode_for_size_tree): Update call accordingly.
16047 (bitwise_mode_for_mode): Likewise.
16048 (make_fract_type): Likewise.
16049 (make_accum_type): Likewise.
16050 * caller-save.c (replace_reg_with_saved_mem): Update call
16052 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16053 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16054 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16055 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16056 * expmed.c (extract_bit_field_1): Likewise.
16057 * reload.c (get_secondary_mem): Likewise.
16058 * varasm.c (assemble_integer): Likewise.
16059 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16062 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16064 * machmode.h (decimal_float_mode_for_size): New function.
16065 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16066 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16067 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16068 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16069 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16070 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16072 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16074 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16075 (get_builtin_sync_mode): Likewise.
16076 (expand_ifn_atomic_compare_exchange): Likewise.
16077 (expand_builtin_atomic_clear): Likewise.
16078 (expand_builtin_atomic_test_and_set): Likewise.
16079 (fold_builtin_atomic_always_lock_free): Likewise.
16080 * calls.c (compute_argument_addresses): Likewise.
16081 (emit_library_call_value_1): Likewise.
16082 (store_one_arg): Likewise.
16083 * combine.c (combine_instructions): Likewise.
16084 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16085 * config/arm/arm.c (arm_function_value): Likewise.
16086 (aapcs_allocate_return_reg): Likewise.
16087 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16088 * config/i386/i386.c (construct_container): Likewise.
16089 (ix86_gimplify_va_arg): Likewise.
16090 (ix86_expand_sse_cmp): Likewise.
16091 (emit_memmov): Likewise.
16092 (emit_memset): Likewise.
16093 (expand_small_movmem_or_setmem): Likewise.
16094 (ix86_expand_pextr): Likewise.
16095 (ix86_expand_pinsr): Likewise.
16096 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16097 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16099 * config/mips/mips.c (mips_function_value_1) Likewise.
16100 (mips_block_move_straight): Likewise.
16101 (mips_expand_ins_as_unaligned_store): Likewise.
16102 * config/powerpcspe/powerpcspe.c
16103 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16104 (rs6000_darwin64_record_arg_flush): Likewise.
16105 * config/rs6000/rs6000.c
16106 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16107 (rs6000_darwin64_record_arg_flush): Likewise.
16108 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16109 (sparc_function_value_1): Likewise.
16110 * config/spu/spu.c (adjust_operand): Likewise.
16111 (spu_emit_branch_or_set): Likewise.
16112 (arith_immediate_p): Likewise.
16113 * emit-rtl.c (gen_lowpart_common): Likewise.
16114 * expr.c (expand_expr_real_1): Likewise.
16115 * function.c (assign_parm_setup_block): Likewise.
16116 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16117 * reload1.c (alter_reg): Likewise.
16118 * stor-layout.c (mode_for_vector): Likewise.
16119 (layout_type): Likewise.
16121 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16123 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16124 (spu_convert_move): Likewise.
16125 * lower-subreg.c (resolve_simple_move): Likewise.
16127 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16130 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16131 define_insn to a define_expand.
16132 (altivec_vsum2sws_direct): New define_insn.
16133 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16135 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16137 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16140 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16145 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16146 realign stack if stack alignment needed is less than incoming
16149 2017-09-05 Marek Polacek <polacek@redhat.com>
16152 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16155 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16157 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16159 2017-09-05 Richard Biener <rguenther@suse.de>
16161 PR tree-optimization/82084
16162 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16164 2017-09-05 Richard Biener <rguenther@suse.de>
16166 PR tree-optimization/82102
16167 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16169 2017-09-05 Martin Liska <mliska@suse.cz>
16171 PR tree-optimization/82032
16172 * tree-cfg.c (generate_range_test): New function.
16173 * tree-cfg.h (generate_range_test): Declared here.
16174 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16175 (cleanup_control_expr_graph): Use it.
16176 * tree-switch-conversion.c (try_switch_expansion): Remove
16178 (emit_case_nodes): Use generate_range_test.
16180 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16183 * config/i386/i386.md (*<btsc><mode>_mask): Add
16184 TARGET_USE_BT to insn constraint.
16185 (*btr<mode>_mask): Ditto.
16187 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16189 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16190 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16192 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16195 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16196 TARGET_NEON and TARGET_IWMMXT.
16197 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16198 TARGET_NEON and TARGET_IWMMXT.
16199 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16201 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16203 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16204 (ix86_rewrite_tls_address): Ditto.
16205 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16206 (ix86_rewrite_tls_address_1): Ditto.
16207 (ix86_rewrite_tls_address): Ditto.
16208 * config/i386/predicates.md (tls_address_pattern): New predicate.
16209 * config/i386/i386.md (TLS address splitter): New splitter.
16211 2017-09-04 Richard Biener <rguenther@suse.de>
16213 PR tree-optimization/82084
16214 * fold-const.h (can_native_encode_string_p): Declare.
16215 * fold-const.c (can_native_encode_string_p): Factor out from ...
16216 (native_encode_string): ... here.
16217 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16218 vectorizing stores from constants we later cannot handle.
16220 2017-09-04 Marek Polacek <polacek@redhat.com>
16223 * doc/invoke.texi: Update -Wtautological-compare documentation.
16225 2017-09-04 Jeff Law <law@redhat.com>
16227 PR tree-optimization/64910
16228 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16229 swap the first and last operand if the last is a constant.
16231 2017-09-04 Marek Polacek <polacek@redhat.com>
16234 * convert.c (do_narrow): When sanitizing signed integer overflows,
16235 bail out for signed types.
16236 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16238 2017-09-04 Richard Biener <rguenther@suse.de>
16240 PR tree-optimization/82060
16241 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16242 Move devirtualization after stmt folding and before EH/AB/noreturn
16243 cleanup to get the stmt refs canonicalized. Use a bool instead
16244 of gimple_modified_p since that doesn't work for NOPs. Schedule
16245 NOPs generated by folding for removal.
16247 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16248 Alan Hayward <alan.hayward@arm.com>
16249 David Sherwood <david.sherwood@arm.com>
16251 * coretypes.h (pad_direction): New enum.
16252 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16253 (FUNCTION_ARG_PADDING): Likewise.
16254 * target.def (function_arg_padding): New hook.
16255 * targhooks.h (default_function_arg_padding): Declare.
16256 * targhooks.c (default_function_arg_padding): New function.
16257 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16258 (TARGET_FUNCTION_ARG_PADDING): ...this.
16259 * doc/tm.texi: Regenerate.
16260 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16261 instead of direction.
16262 (compute_argument_addresses): Likewise.
16263 (load_register_parameters): Likewise.
16264 (emit_library_call_value_1): Likewise.
16265 (store_one_arg): Use targetm.calls.function_arg_padding instead
16266 of FUNCTION_ARG_PADDING.
16267 (must_pass_in_stack_var_size_or_pad): Likewise.
16268 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16269 (emit_group_store): Likewise.
16270 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16271 instead of FUNCTION_ARG_PADDING.
16272 (emit_push_insn): Likewise, and propagate enum change throughout
16274 * function.h (direction): Delete.
16275 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16277 * function.c (assign_parm_find_stack_rtl): Likewise.
16278 (assign_parm_setup_block_p): Likewise.
16279 (assign_parm_setup_block): Likewise.
16280 (gimplify_parameters): Likewise.
16281 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16282 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16284 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16285 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16286 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16287 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16288 (aarch64_function_arg_padding): ...this new function.
16289 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16290 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16291 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16292 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16293 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16294 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16295 (arm_pad_arg_upward): Replace with...
16296 (arm_function_arg_padding): ...this new function.
16297 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16299 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16300 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16301 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16302 (ia64_hpux_function_arg_padding): Replace with...
16303 (ia64_function_arg_padding): ...this new function. Use pad_direction
16304 instead of direction. Check for TARGET_HPUX.
16305 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16306 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16307 (iq2000_function_arg_padding): New function.
16308 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16309 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16310 (mips_function_arg_padding): ...this new function.
16311 (mips_pad_reg_upward): Update accordingly.
16312 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16313 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16314 targetm.calls.function_arg_padding.
16315 (FUNCTION_ARG_PADDING): Delete.
16316 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16317 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16318 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16319 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16320 (nios2_block_reg_padding): Return pad_direction instead of direction.
16321 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16322 instead of direction.
16323 (nios2_function_arg_padding): Likewise. Make static.
16324 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16325 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16326 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16327 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16328 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16329 (pa_function_arg_padding): Make static. Return pad_direction instead
16331 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16332 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16333 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16334 instead of direction. Use targetm.calls.function_arg_padding.
16335 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16336 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16337 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16338 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16339 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16341 (function_arg_padding): Rename to...
16342 (rs6000_function_arg_padding): ...this. Make static. Return
16343 pad_direction instead of direction.
16344 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16345 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16346 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16347 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16348 instead of direction. Use targetm.calls.function_arg_padding.
16349 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16350 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16351 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16352 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16353 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16354 (function_arg_padding): Rename to...
16355 (rs6000_function_arg_padding): ...this. Make static. Return
16356 pad_direction instead of direction.
16357 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16358 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16359 * config/s390/s390.c (s390_function_arg_padding): New function.
16360 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16361 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16362 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16363 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16364 (function_arg_padding): Rename to...
16365 (sparc_function_arg_padding): ...this. Make static. Return
16366 pad_direction instead of direction.
16367 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16368 * config/spu/spu.c (spu_function_arg_padding): New function.
16369 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16370 * system.h (FUNCTION_ARG_PADDING): Poison.
16372 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16373 Alan Hayward <alan.hayward@arm.com>
16374 David Sherwood <david.sherwood@arm.com>
16376 * target.def (modes_tieable_p): New hook.
16377 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16378 (TARGET_MODES_TIEABLE_P): ...this.
16379 * doc/tm.texi.in: Regenerate.
16380 * hooks.h (hook_bool_mode_mode_true): Declare.
16381 * hooks.c (hook_bool_mode_mode_true): New function.
16382 * combine.c (subst): Use targetm.modes_tieable_p instead of
16384 * dse.c (find_shift_sequence): Likewise.
16385 * expmed.c (extract_low_bits): Likewise.
16386 * lower-subreg.c: Include target.h.
16387 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16389 * rtlanal.c (rtx_cost): Likewise.
16390 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16391 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16392 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16393 (TARGET_MODES_TIEABLE_P): Redefine.
16394 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16395 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16396 (TARGET_MODES_TIEABLE_P): Redefine.
16397 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16398 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16399 (arc_modes_tieable_p): New function.
16400 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16401 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16402 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16403 (arm_modes_tieable_p): Make static.
16404 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16405 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16406 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16407 (TARGET_MODES_TIEABLE_P): Redefine.
16408 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16409 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16410 (TARGET_MODES_TIEABLE_P): Redefine.
16411 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16412 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16413 (cr16_modes_tieable_p): New function.
16414 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16415 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16416 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16417 (TRULY_NOOP_TRUNCATION): Update comment.
16418 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16419 (TRULY_NOOP_TRUNCATION): Update comment.
16420 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16421 (frv_modes_tieable_p): New function.
16422 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16423 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16424 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16425 (TARGET_MODES_TIEABLE_P): Redefine.
16426 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16427 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16428 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16429 (TARGET_MODES_TIEABLE_P): Redefine.
16430 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16431 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16432 (ia64_modes_tieable_p): New function.
16433 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16434 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16435 (iq2000_modes_tieable_p): New function.
16436 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16437 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16438 (lm32_modes_tieable_p): New function.
16439 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16440 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16441 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16442 (TARGET_MODES_TIEABLE_P): Redefine.
16443 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16444 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16445 (m32r_modes_tieable_p): New function.
16446 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16447 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16448 (m68k_modes_tieable_p): New function.
16449 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16450 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16451 (mcore_modes_tieable_p): New function.
16452 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16453 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16455 (TARGET_MODES_TIEABLE_P): Redefine.
16456 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16457 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16458 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16459 (TARGET_MODES_TIEABLE_P): Redefine.
16460 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16461 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16462 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16463 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16464 (mn10300_modes_tieable_p): ...this and make static.
16465 (TARGET_MODES_TIEABLE_P): Redefine.
16466 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16467 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16468 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16469 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16470 (msp430_modes_tieable_p): Make static.
16471 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16472 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16473 (TARGET_MODES_TIEABLE_P): Redefine.
16474 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16475 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16476 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16477 (TARGET_MODES_TIEABLE_P): Redefine.
16478 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16479 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16480 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16481 (TARGET_MODES_TIEABLE_P): Redefine.
16482 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16483 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16484 (pdp11_modes_tieable_p): New function.
16485 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16486 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16487 (rs6000_modes_tieable_p): New function.
16488 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16489 * config/powerpcspe/powerpcspe.md: Update comment.
16490 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16491 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16492 (TARGET_MODES_TIEABLE_P): Redefine.
16493 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16494 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16495 (rl78_modes_tieable_p): New function.
16496 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16497 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16498 (rs6000_modes_tieable_p): New function.
16499 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16500 * config/rs6000/rs6000.md: Update comment.
16501 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16502 * config/rx/rx.c (rx_modes_tieable_p): New function.
16503 (TARGET_MODES_TIEABLE_P): Redefine.
16504 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16505 * config/s390/s390.c (s390_modes_tieable_p): New function.
16506 (TARGET_MODES_TIEABLE_P): Redefine.
16507 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16508 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16509 (sh_modes_tieable_p): New function.
16510 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16511 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16512 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16513 (sparc_modes_tieable_p): Make static.
16514 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16515 * config/spu/spu.c (spu_modes_tieable_p): New function.
16516 (TARGET_MODES_TIEABLE_P): Redefine.
16517 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16518 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16519 (TARGET_MODES_TIEABLE_P): Redefine.
16520 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16521 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16522 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16523 * config/v850/v850.c (v850_modes_tieable_p): New function.
16524 (TARGET_MODES_TIEABLE_P): Redefine.
16525 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16526 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16527 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16528 (visium_modes_tieable_p): New function.
16529 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16530 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16531 (xtensa_modes_tieable_p): New function.
16532 * system.h (MODES_TIEABLE_P): Poison.
16534 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16535 Alan Hayward <alan.hayward@arm.com>
16536 David Sherwood <david.sherwood@arm.com>
16538 * target.def (hard_regno_mode_ok): New hook.
16539 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16540 (TARGET_HARD_REGNO_MODE_OK): ...this.
16541 * doc/tm.texi.in: Regenerate.
16542 * hooks.h (hook_bool_uint_mode_true): Declare.
16543 * hooks.c (hook_bool_uint_mode_true): New function.
16544 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16545 HARD_REGNO_MODE_OK.
16546 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16547 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16548 instead of HARD_REGNO_MODE_OK.
16549 * caller-save.c: Include target.h.
16550 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16551 HARD_REGNO_MODE_OK.
16552 * combine.c (can_combine_p): Likewise.
16553 (combinable_i3pat): Likewise.
16554 (can_change_dest_mode): Likewise.
16555 * expr.c (init_expr_target): Likewise.
16556 (convert_move): Likewise.
16557 (convert_modes): Likewise.
16558 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16559 (setup_prohibited_mode_move_regs): Likewise.
16560 * ira.h (target_ira): Likewise.
16561 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16562 * lra-constraints.c (process_alt_operands): Likewise.
16563 (split_reg): Likewise.
16564 * recog.c (peep2_find_free_register): Likewise.
16565 * ree.c (combine_reaching_defs): Likewise.
16566 * regcprop.c (maybe_mode_change): Likewise.
16567 * reginfo.c (init_reg_sets_1): Likewise.
16568 (choose_hard_reg_mode): Likewise.
16569 (simplifiable_subregs): Likewise.
16570 * regrename.c (check_new_reg_p): Likewise.
16571 * reload.c (find_valid_class): Likewise.
16572 (find_valid_class_1): Likewise.
16573 (reload_inner_reg_of_subreg): Likewise.
16574 (push_reload): Likewise.
16575 (combine_reloads): Likewise.
16576 (find_dummy_reload): Likewise.
16577 (find_reloads): Likewise.
16578 * reload1.c (find_reg): Likewise.
16579 (set_reload_reg): Likewise.
16580 (allocate_reload_reg): Likewise.
16581 (choose_reload_regs): Likewise.
16582 (reload_adjust_reg_for_temp): Likewise.
16583 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16584 (simplify_subreg_regno): Likewise.
16585 * sel-sched.c (init_regs_for_mode): Likewise.
16586 * varasm.c (make_decl_rtl): Likewise.
16587 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16588 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16589 HARD_REGNO_MODE_OK.
16590 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16591 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16592 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16593 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16594 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16595 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16596 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16597 (arc_mode_class): Delete.
16598 (HARD_REGNO_MODE_OK): Delete.
16599 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16600 (arc_hard_regno_mode_ok): Rename old array to...
16601 (arc_hard_regno_mode_ok_modes): ...this.
16602 (arc_conditional_register_usage): Update accordingly.
16603 (arc_mode_class): Make static.
16604 (arc_hard_regno_mode_ok): New function.
16605 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16606 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16607 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16608 (arm_hard_regno_mode_ok): Make static.
16609 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16610 HARD_REGNO_MODE_OK.
16611 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16612 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16613 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16615 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16616 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16617 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16618 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16619 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16620 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16621 * config/bfin/predicates.md (valid_reg_operand): Use
16622 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16623 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16624 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16625 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16626 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16627 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16628 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16629 (cr16_hard_regno_mode_ok): Make static and return a bool.
16630 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16631 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16632 (cris_hard_regno_mode_ok): New function.
16633 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16634 (epiphany_mode_class): Delete.
16635 (HARD_REGNO_MODE_OK): Delete.
16636 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16637 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16638 (hard_regno_mode_ok): Rename to...
16639 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16640 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16641 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16642 HARD_REGNO_MODE_OK.
16643 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16644 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16645 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16646 (frv_hard_regno_mode_ok): Make static and return a bool.
16647 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16648 HARD_REGNO_MODE_OK.
16649 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16650 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16651 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16652 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16654 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16655 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16656 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16657 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16659 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16660 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16661 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16662 (ia64_hard_regno_mode_ok): New function.
16663 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16664 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16665 (iq2000_hard_regno_mode_ok): New function.
16666 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16667 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16668 (lm32_hard_regno_mode_ok): New function.
16669 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16670 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16671 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16672 instead of HARD_REGNO_MODE_OK.
16673 (m32c_hard_regno_ok): Rename to...
16674 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16675 (m32c_cannot_change_mode_class): Update accordingly.
16676 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16677 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16678 (m32r_mode_class): Delete.
16679 (HARD_REGNO_MODE_OK): Delete.
16680 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16681 (m32r_hard_regno_mode_ok): Rename to...
16682 (m32r_hard_regno_modes): ...this.
16683 (m32r_mode_class): Make static.
16684 (m32r_hard_regno_mode_ok): New function.
16685 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16686 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16687 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16688 (m68k_hard_regno_mode_ok): Make static.
16689 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16690 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16691 (mcore_hard_regno_mode_ok): New function.
16692 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16693 (HARD_REGNO_MODE_OK): Delete.
16694 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16696 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16697 (microblaze_hard_regno_mode_ok): New function.
16698 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16699 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16700 (mips_hard_regno_mode_ok): Delete.
16701 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16702 (mips_hard_regno_mode_ok_p): ...this and make static.
16703 (mips_hard_regno_mode_ok_p): Rename to...
16704 (mips_hard_regno_mode_ok_uncached): ...this.
16705 (mips_hard_regno_mode_ok): New function.
16706 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16707 of HARD_REGNO_MODE_OK.
16708 (mips_option_override): Update after above name changes.
16709 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16710 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16711 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16712 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16713 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16714 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16715 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16716 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16717 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16718 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16719 (msp430_hard_regno_mode_ok): Make static and return a bool.
16720 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16721 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16722 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16724 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16725 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16726 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16727 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16728 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16729 (PA_HARD_REGNO_MODE_OK): ...this
16730 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16731 (PA_HARD_REGNO_MODE_OK): ...this.
16732 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16733 (pa_hard_regno_mode_ok): New function.
16734 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16735 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16736 (pdp11_hard_regno_mode_ok): New function.
16737 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16738 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16740 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16742 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16743 (rs6000_hard_regno_mode_ok): Rename to...
16744 (rs6000_hard_regno_mode_ok_uncached): ...this.
16745 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16746 (rs6000_hard_regno_mode_ok): New function.
16747 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16748 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16749 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16750 (riscv_hard_regno_mode_ok): ...this and make static.
16751 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16752 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16753 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16754 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16755 (rl78_hard_regno_mode_ok): Make static and return bool.
16756 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16757 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16759 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16760 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16761 (rs6000_hard_regno_mode_ok): Rename to...
16762 (rs6000_hard_regno_mode_ok_uncached): ...this.
16763 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16764 (rs6000_hard_regno_mode_ok): New function.
16765 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16766 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16767 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16768 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16769 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16770 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16771 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16772 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16773 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16774 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16775 (sh_hard_regno_mode_ok): Make static.
16776 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16777 instead of HARD_REGNO_MODE_OK.
16778 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16779 (sparc_mode_class): Delete.
16780 (HARD_REGNO_MODE_OK): Delete.
16781 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16782 (hard_regno_mode_classes): Make static.
16783 (sparc_mode_class): Likewise.
16784 (sparc_hard_regno_mode_ok): New function.
16785 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16786 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16787 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16789 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16790 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16791 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16792 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16793 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16794 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16795 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16796 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16797 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16798 (visium_hard_regno_mode_ok): New function.
16799 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16800 instead of HARD_REGNO_MODE_OK.
16801 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16802 (HARD_REGNO_MODE_OK): Delete.
16803 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16804 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16805 (xtensa_option_override): Update accordingly.
16806 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16807 (xtensa_hard_regno_mode_ok): New function.
16808 * system.h (HARD_REGNO_MODE_OK): Poison.
16810 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16811 Alan Hayward <alan.hayward@arm.com>
16812 David Sherwood <david.sherwood@arm.com>
16814 * target.def (hard_regno_call_part_clobbered): New hook.
16815 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16816 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16817 * doc/tm.texi: Regenerate.
16818 * hooks.h (hook_bool_uint_mode_false): Declare.
16819 * hooks.c (hook_bool_uint_mode_false): New function.
16820 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16821 * cselib.c (cselib_process_insn): Use
16822 targetm.hard_regno_call_part_clobbered instead of
16823 HARD_REGNO_CALL_PART_CLOBBERED.
16824 * ira-conflicts.c (ira_build_conflicts): Likewise.
16825 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16826 * lra-constraints.c (need_for_call_save_p): Likewise.
16827 * lra-lives.c: Include target.h.
16828 (check_pseudos_live_through_calls): Use
16829 targetm.hard_regno_call_part_clobbered instead of
16830 HARD_REGNO_CALL_PART_CLOBBERED.
16831 * regcprop.c: Include target.h.
16832 (copyprop_hardreg_forward_1): Use
16833 targetm.hard_regno_call_part_clobbered instead of
16834 HARD_REGNO_CALL_PART_CLOBBERED.
16835 * reginfo.c (choose_hard_reg_mode): Likewise.
16836 * regrename.c (check_new_reg_p): Likewise.
16837 * reload.c (find_equiv_reg): Likewise.
16838 * reload1.c (emit_reload_insns): Likewise.
16839 * sched-deps.c (deps_analyze_insn): Likewise.
16840 * sel-sched.c (init_regs_for_mode): Likewise.
16841 (mark_unavailable_hard_regs): Likewise.
16842 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16843 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16844 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16846 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16847 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16848 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16850 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16852 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16853 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16854 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16856 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16857 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16858 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16860 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16861 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16863 * config/powerpcspe/powerpcspe.c
16864 (rs6000_hard_regno_call_part_clobbered): New function.
16865 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16866 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16867 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16869 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16870 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16871 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16873 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16874 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16875 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16877 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16878 Alan Hayward <alan.hayward@arm.com>
16879 David Sherwood <david.sherwood@arm.com>
16881 * rtl.h (subreg_memory_offset): Declare.
16882 * emit-rtl.c (subreg_memory_offset): New function.
16883 * expmed.c (store_bit_field_1): Use it.
16884 * expr.c (undefined_operand_subword_p): Likewise.
16885 * simplify-rtx.c (simplify_subreg): Likewise.
16887 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16889 PR rtl-optimization/57448
16892 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16893 using atomic_load pattern.
16894 (expand_atomic_store): Likewise.
16896 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16899 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16900 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16901 handling. Use replace_call_with_value with NULL instead of
16902 gsi_replace, unlink_stmt_vdef and release_defs.
16904 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16907 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16909 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16912 * rtl.h (emit_library_call_value_1): Declare.
16913 (emit_library_call): Replace declaration with a series of overloads.
16914 Remove the parameter count argument.
16915 (emit_library_call_value): Likewise.
16916 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16917 with an "rtx_mode_t *".
16918 (emit_library_call_value): Delete.
16919 (emit_library_call): Likewise.
16920 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16921 (asan_emit_allocas_unpoison): Likewise.
16922 * builtins.c (expand_builtin_powi): Likewise.
16923 (expand_asan_emit_allocas_unpoison): Likewise.
16924 * cfgexpand.c (expand_main_function): Likewise.
16925 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16926 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16927 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16928 * config/arm/arm.c (arm_trampoline_init): Likewise.
16929 (arm_call_tls_get_addr): Likewise.
16930 (arm_expand_divmod_libfunc): Likewise.
16931 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16932 (smulsi3_highpart): Likewise.
16933 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16934 (c6x_expand_compare): Likewise.
16935 (c6x_expand_movmem): Likewise.
16936 * config/frv/frv.c (frv_trampoline_init): Likewise.
16937 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16938 (ix86_expand_divmod_libfunc): Likewise.
16939 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16940 (ia64_expand_compare): Likewise.
16941 (ia64_profile_hook): Likewise.
16942 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16943 (nonlocal_goto): Likewise.
16944 (restore_stack_nonlocal): Likewise.
16945 * config/m32r/m32r.c (block_move_call): Likewise.
16946 (m32r_trampoline_init): Likewise.
16947 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16948 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16949 (m68k_call_m68k_read_tp): Likewise.
16950 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16951 (microblaze_expand_divide): Likewise.
16952 * config/mips/mips.h (mips_args): Likewise.
16953 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16954 (MIPS_ICACHE_SYNC): Likewise.
16955 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16956 (nios2_trampoline_init): Likewise.
16957 * config/pa/pa.c (hppa_tls_call): Likewise.
16958 (pa_trampoline_init): Likewise.
16959 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16960 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16961 (expand_strn_compare): Likewise.
16962 (rs6000_generate_compare): Likewise.
16963 (rs6000_expand_float128_convert): Likewise.
16964 (output_profile_hook): Likewise.
16965 (rs6000_trampoline_init): Likewise.
16966 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16967 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16968 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16969 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16970 (rs6000_generate_compare): Likewise.
16971 (rs6000_expand_float128_convert): Likewise.
16972 (output_profile_hook): Likewise.
16973 (rs6000_trampoline_init): Likewise.
16974 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16975 * config/sh/sh.c (sh_trampoline_init): Likewise.
16976 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16977 (sparc_emit_float_lib_cmp): Likewise.
16978 (sparc32_initialize_trampoline): Likewise.
16979 (sparc64_initialize_trampoline): Likewise.
16980 (sparc_profile_hook): Likewise.
16981 * config/spu/spu.c (ea_load_store): Likewise.
16982 * config/spu/spu.md (floatunssidf2): Likewise.
16983 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16984 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16985 * config/visium/visium.c (expand_block_move_4): Likewise.
16986 (expand_block_move_2): Likewise.
16987 (expand_block_move_1): Likewise.
16988 (expand_block_set_4): Likewise.
16989 (expand_block_set_2): Likewise.
16990 (expand_block_set_1): Likewise.
16991 (visium_trampoline_init): Likewise.
16992 (visium_profile_hook): Likewise.
16993 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16994 (xtensa_setup_frame_addresses): Likewise.
16995 (xtensa_trampoline_init): Likewise.
16996 * except.c (sjlj_emit_function_enter): Likewise.
16997 (sjlj_emit_function_exit): Likewise.
16998 * explow.c (allocate_dynamic_stack_space): Likewise.
16999 (probe_stack_range): Likewise.
17000 * expr.c (convert_mode_scalar): Likewise.
17001 * optabs.c (expand_binop): Likewise.
17002 (expand_twoval_binop_libfunc): Likewise.
17003 (expand_unop): Likewise.
17004 (prepare_cmp_insn): Likewise.
17005 (prepare_float_lib_cmp): Likewise.
17006 (expand_float): Likewise.
17007 (expand_fix): Likewise.
17008 (expand_fixed_convert): Likewise.
17009 (maybe_emit_sync_lock_test_and_set): Likewise.
17010 (expand_atomic_compare_and_swap): Likewise.
17011 (expand_mem_thread_fence): Likewise.
17012 (expand_atomic_fetch_op): Likewise.
17014 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17016 * doc/generic.texi (OpenACC): Adjust URL.
17017 * doc/invoke.texi (C Dialect Options): Ditto.
17019 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17021 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17022 predicate for operand 1. Add (m,<S>) constraint.
17023 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17024 Prevent memory operand 1 with register operand 2.
17026 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17028 PR rtl-optimization/82024
17029 * combine.c (try_combine): If the combination result is a PARALLEL,
17030 and we only need to retain the SET in there that would be placed
17031 at I2, check that we can place that at I3 instead, before doing so.
17033 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17036 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17038 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17039 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17042 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17043 Jeff Law <law@redhat.com>
17045 * varasm.c (bss_initializer_p): Do not put constants into .bss
17046 (categorize_decl_for_section): Handle bss_initializer_p returning
17047 false when DECL_INITIAL is NULL.
17049 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17052 * config/s390/s390.c (s390_can_inline_p): New function.
17054 2017-09-01 Jeff Law <law@redhat.com>
17056 PR tree-optimization/82052
17057 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17058 Always initialize the returned slot after a hash table miss
17059 when INSERT is true.
17061 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17063 * config/s390/s390.md (mem_signal_fence): Remove.
17064 * doc/md.texi (mem_signal_fence): Remove.
17065 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17067 * target-insns.def (mem_signal_fence): Remove.
17069 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17072 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17075 * asan.c (create_odr_indicator): Strip name encoding from assembler
17076 name before appending it after __odr_asan_.
17078 2017-09-01 Martin Liska <mliska@suse.cz>
17080 PR tree-optimization/82059
17081 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17082 frequency only when an edge is redirected.
17084 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17086 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17087 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17088 (arc_conditional_register_usage): Remove ARC600 lp_count
17090 (arc_file_start): Emit Tag_ARC_CPU_variation.
17091 (arc_can_use_doloop_p): New conditions to use ZOLs.
17092 (hwloop_fail): New function.
17093 (hwloop_optimize): Likewise.
17094 (hwloop_pattern_reg): Likewise.
17095 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17096 (arc_reorg_loops): New function, calls reorg_loops.
17097 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17098 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17100 (arc_loop_hazard): Remove function, functionality moved into
17102 (arc_hazard): Remove arc_loop_hazard call.
17103 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17104 into hwloop_optimize.
17105 (arc_label_align): Remove ZOL handling.
17106 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17107 * config/arc/arc.md (doloop_begin): Remove pattern.
17108 (doloop_begin_i): Likewise.
17109 (doloop_end_i): Likewise.
17110 (doloop_fallback): Likewise.
17111 (doloop_fallback_m): Likewise.
17112 (doloop_end): Reimplement expand.
17113 (arc_lp): New pattern for LP instruction.
17114 (loop_end): New pattern.
17115 (loop_fail): Likewise.
17116 (decrement_and_branch_until_zero): Likewise.
17117 * config/arc/arc.opt (mlpc-width): New option.
17118 * doc/invoke.texi (mlpc-width): Document option.
17120 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17122 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17123 (arc_ccfsm_advance): Fix checking for delay slots.
17124 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17126 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17128 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17130 (movhi_insn): Update store instruction constraint which are saving
17131 6-bit short immediates.
17132 (movsi_insn): Consider also short scaled load operations.
17133 (zero_extendhisi2_i): Use Usd constraint instead of T.
17134 (extendhisi2_i): Add q constraint.
17135 (arc_clzsi2): Add type and length attributes.
17136 (arc_ctzsi2): Likewise.
17137 * config/arc/constraints.md (Usc): Update constraint, the
17138 assembler can parse two relocations for a single instruction.
17140 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17142 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17143 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17145 2017-08-31 Olivier Hainque <hainque@adacore.com>
17147 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17148 match as powerpc-wrs-vxworks*.
17150 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17152 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17153 register constraint for by-element operand.
17154 (aarch64_mls_elt_merge<mode>): Likewise.
17156 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17158 * config/arc/arc.c (arc_can_follow_jump): Check for short
17161 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17163 * config.gcc: Use g.opt for arc.
17164 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17165 functionality moved to ...
17166 (legitimate_scaled_address_p): New function, ...here.
17167 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17168 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17169 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17171 (arc_override_options): Handle G option.
17172 (arc_output_pic_addr_const): Correct function definition.
17173 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17174 (arc_decl_anon_ns_mem_p): Delete.
17175 (arc_in_small_data_p): Overhaul this function to take into
17176 consideration the value given via G option.
17177 (arc_rewrite_small_data_1): Renamed and corrected old
17178 arc_rewrite_small_data function.
17179 (arc_rewrite_small_data): New function.
17180 (small_data_pattern): Don't use pic_offset_table_rtx.
17181 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17182 * config/arc/simdext.md (movmisalignv2hi): Use
17183 prepare_move_operands function.
17185 (movmisalign*): Likewise.
17186 * doc/invoke.texi (ARC options): Document -G option.
17188 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17190 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17192 * config/arc/arc.c (arc_print_operand): Output scalled address for
17193 sdata whenever is possible.
17194 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17195 load/stores are available.
17196 (compact_sda_memory_operand): Check for the alignment required by
17197 code density instructions.
17198 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17200 * config/arc/constraints.md (Usd): Update constraint.
17201 (Us0): New constraint.
17202 (Usc): Update constraint.
17204 2017-08-31 Richard Biener <rguenther@suse.de>
17206 PR middle-end/82054
17207 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17208 function only once.
17210 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17212 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17213 Resize type_signature.
17215 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17216 Alan Hayward <alan.hayward@arm.com>
17217 David Sherwood <david.sherwood@arm.com>
17219 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17220 subregs whose inner modes can be stored in GPRs.
17221 (aarch64_classify_index): Likewise.
17223 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17224 Alan Hayward <alan.hayward@arm.com>
17225 David Sherwood <david.sherwood@arm.com>
17227 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17228 (V_INT_EQUIV): ...this.
17229 (v_cmp_result): Rename to...
17230 (v_int_equiv): ...this.
17231 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17232 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17233 (copysign<mode>3): Likewise.
17234 (aarch64_simd_bsl<mode>_internal): Likewise.
17235 (aarch64_simd_bsl<mode>): Likewise.
17236 (vec_cmp<mode><mode>): Likewise.
17237 (vcond<mode><mode>): Likewise.
17238 (vcond<v_cmp_mixed><mode>): Likewise.
17239 (vcondu<mode><v_cmp_mixed>): Likewise.
17240 (aarch64_cm<optab><mode>): Likewise.
17241 (aarch64_cmtst<mode>): Likewise.
17242 (aarch64_fac<optab><mode>): Likewise.
17243 (vec_perm_const<mode>): Likewise.
17244 (vcond_mask_<mode><v_cmp_result>): Rename to...
17245 (vcond_mask_<mode><v_int_equiv>): ...this.
17246 (vec_cmp<mode><v_cmp_result>): Rename to...
17247 (vec_cmp<mode><v_int_equiv>): ...this.
17249 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17250 Alan Hayward <alan.hayward@arm.com>
17251 David Sherwood <david.sherwood@arm.com>
17253 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17255 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17256 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17257 (UNSPEC_LD4_DREG): New unspecs.
17258 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17259 (aarch64_ld2<mode>_dreg_be): Replace with...
17260 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17262 (aarch64_ld3<mode>_dreg_le)
17263 (aarch64_ld3<mode>_dreg_be): Replace with...
17264 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17266 (aarch64_ld4<mode>_dreg_le)
17267 (aarch64_ld4<mode>_dreg_be): Replace with...
17268 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17271 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17273 PR tree-optimization/81987
17274 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17275 insert an initializer in a location not dominated by the stride
17278 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17280 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17281 on the entire header of the finally block in the fallthru case.
17283 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17285 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17287 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17289 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17290 rs6000_emit_move_from_cr and call renamed function.
17291 (rs6000_emit_prologue): Call renamed functions.
17292 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17293 movesi_from_cr, remove volatile CRs.
17295 2017-08-30 Jon Beniston <jon@beniston.com>
17296 Richard Biener <rguenther@suse.de>
17298 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17299 of VECTOR_MODE_P check.
17300 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17301 element vector types.
17303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17305 * df.h (df_read_modify_subreg_p): Remove in favor of...
17306 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17307 const_rtx instead of an rtx.
17308 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17309 * df-problems.c (df_word_lr_mark_ref): Likewise.
17310 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17311 (mark_pseudo_reg_dead): Likewise.
17312 (mark_ref_dead): Likewise.
17313 * reginfo.c (init_subregs_of_mode): Likewise.
17314 * sched-deps.c (sched_analyze_1): Likewise.
17315 * df-scan.c (df_def_record_1): Likewise.
17316 (df_uses_record): Likewise.
17317 (df_read_modify_subreg_p): Remove in favor of...
17318 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17319 const_rtx instead of an rtx.
17321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17322 Alan Hayward <alan.hayward@arm.com>
17323 David Sherwood <david.sherwood@arm.com>
17325 * rtl.h (partial_subreg_p): New function.
17326 * caller-save.c (save_call_clobbered_regs): Use it.
17327 * calls.c (expand_call): Likewise.
17328 * combine.c (combinable_i3pat): Likewise.
17329 (simplify_set): Likewise.
17330 (make_extraction): Likewise.
17331 (make_compound_operation_int): Likewise.
17332 (gen_lowpart_or_truncate): Likewise.
17333 (force_to_mode): Likewise.
17334 (make_field_assignment): Likewise.
17335 (reg_truncated_to_mode): Likewise.
17336 (record_truncated_value): Likewise.
17337 (move_deaths): Likewise.
17338 * cse.c (record_jump_cond): Likewise.
17339 (cse_insn): Likewise.
17340 * cselib.c (cselib_lookup_1): Likewise.
17341 * expmed.c (extract_bit_field_using_extv): Likewise.
17342 * function.c (assign_parm_setup_reg): Likewise.
17343 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17344 * ira-build.c (create_insn_allocnos): Likewise.
17345 * lra-coalesce.c (merge_pseudos): Likewise.
17346 * lra-constraints.c (match_reload): Likewise.
17347 (simplify_operand_subreg): Likewise.
17348 (curr_insn_transform): Likewise.
17349 * lra-lives.c (process_bb_lives): Likewise.
17350 * lra.c (new_insn_reg): Likewise.
17351 (lra_substitute_pseudo): Likewise.
17352 * regcprop.c (mode_change_ok): Likewise.
17353 (maybe_mode_change): Likewise.
17354 (copyprop_hardreg_forward_1): Likewise.
17355 * reload.c (push_reload): Likewise.
17356 (find_reloads): Likewise.
17357 (find_reloads_subreg_address): Likewise.
17358 * reload1.c (alter_reg): Likewise.
17359 (eliminate_regs_1): Likewise.
17360 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17362 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17364 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17365 back to if statements, including unpack.
17367 2017-08-30 Martin Liska <mliska@suse.cz>
17369 PR inline-asm/82001
17370 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17372 (func_checker::compare_asm_inputs_outputs): ... this function.
17373 (func_checker::compare_gimple_asm): Use the function to compare
17374 also ASM constrains.
17375 * ipa-icf-gimple.h: Rename the function.
17377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17378 Alan Hayward <alan.hayward@arm.com>
17379 David Sherwood <david.sherwood@arm.com>
17381 * coretypes.h (complex_mode): New type.
17382 * gdbhooks.py (build_pretty_printer): Handle it.
17383 * machmode.h (complex_mode): New class.
17384 (complex_mode::includes_p): New function.
17385 (is_complex_int_mode): Likewise.
17386 (is_complex_float_mode): Likewise.
17387 * genmodes.c (get_mode_class): Handle complex mode classes.
17388 * function.c (expand_function_end): Use is_complex_int_mode.
17390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17391 Alan Hayward <alan.hayward@arm.com>
17392 David Sherwood <david.sherwood@arm.com>
17394 * coretypes.h (scalar_mode_pod): New typedef.
17395 * gdbhooks.py (build_pretty_printer): Handle it.
17396 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17397 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17398 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17399 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17400 as_a <scalar_mode>.
17402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17403 Alan Hayward <alan.hayward@arm.com>
17404 David Sherwood <david.sherwood@arm.com>
17406 * machmode.h (mode_for_vector): Take a scalar_mode instead
17408 * stor-layout.c (mode_for_vector): Likewise.
17409 * explow.c (promote_mode): Use as_a <scalar_mode>.
17410 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17413 Alan Hayward <alan.hayward@arm.com>
17414 David Sherwood <david.sherwood@arm.com>
17416 * target.def (preferred_simd_mode): Take a scalar_mode
17417 instead of a machine_mode.
17418 * targhooks.h (default_preferred_simd_mode): Likewise.
17419 * targhooks.c (default_preferred_simd_mode): Likewise.
17420 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17421 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17422 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17423 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17424 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17425 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17426 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17427 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17429 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17430 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17431 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17432 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17433 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17434 * doc/tm.texi: Regenerate.
17435 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17439 Alan Hayward <alan.hayward@arm.com>
17440 David Sherwood <david.sherwood@arm.com>
17442 * target.def (scalar_mode_supported_p): Take a scalar_mode
17443 instead of a machine_mode.
17444 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17445 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17446 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17447 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17448 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17449 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17450 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17451 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17452 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17453 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17454 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17455 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17456 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17457 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17459 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17460 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17461 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17462 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17463 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17465 * doc/tm.texi: Regenerate.
17467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17468 Alan Hayward <alan.hayward@arm.com>
17469 David Sherwood <david.sherwood@arm.com>
17471 * coretypes.h (opt_scalar_mode): New typedef.
17472 * gdbhooks.py (build_pretty_printers): Handle it.
17473 * machmode.h (mode_iterator::get_2xwider): Add overload for
17475 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17477 * expr.c (convert_mode_scalar): Likewise.
17478 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17479 * optabs.c (expand_float): Likewise.
17480 (expand_fix): Likewise.
17481 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17484 Alan Hayward <alan.hayward@arm.com>
17485 David Sherwood <david.sherwood@arm.com>
17487 * optabs.c (expand_float): Explicitly check for scalars before
17488 using a branching expansion.
17489 (expand_fix): Likewise.
17491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17492 Alan Hayward <alan.hayward@arm.com>
17493 David Sherwood <david.sherwood@arm.com>
17495 * expr.c (convert_mode): Split scalar handling out into...
17496 (convert_mode_scalar): ...this new function. Treat the modes
17499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17500 Alan Hayward <alan.hayward@arm.com>
17501 David Sherwood <david.sherwood@arm.com>
17503 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17505 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17508 Alan Hayward <alan.hayward@arm.com>
17509 David Sherwood <david.sherwood@arm.com>
17511 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17512 rather than a machine_mode.
17513 (fixed_from_string): Likewise.
17514 (fixed_convert): Likewise.
17515 (fixed_convert_from_int): Likewise.
17516 (fixed_convert_from_real): Likewise.
17517 (real_convert_from_fixed): Likewise.
17518 * fixed-value.c (fixed_from_double_int): Likewise.
17519 (fixed_from_string): Likewise.
17520 (fixed_convert): Likewise.
17521 (fixed_convert_from_int): Likewise.
17522 (fixed_convert_from_real): Likewise.
17523 (real_convert_from_fixed): Likewise.
17524 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17527 Alan Hayward <alan.hayward@arm.com>
17528 David Sherwood <david.sherwood@arm.com>
17530 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17531 of separate mode class checks. Do not allow vector modes here.
17532 (immed_wide_int_const): Use as_a <scalar_mode>.
17533 * explow.c (trunc_int_for_mode): Likewise.
17534 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17535 (wi::shwi): Likewise.
17536 (wi::min_value): Likewise.
17537 (wi::max_value): Likewise.
17538 * dwarf2out.c (loc_descriptor): Likewise.
17539 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17540 for CONST_WIDE_INT.
17542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17543 Alan Hayward <alan.hayward@arm.com>
17544 David Sherwood <david.sherwood@arm.com>
17546 * tree.h (SCALAR_TYPE_MODE): New macro.
17547 * expr.c (expand_expr_addr_expr_1): Use it.
17548 (expand_expr_real_2): Likewise.
17549 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17550 (fold_convert_const_fixed_from_int): Likewise.
17551 (fold_convert_const_fixed_from_real): Likewise.
17552 (native_encode_fixed): Likewise
17553 (native_encode_complex): Likewise
17554 (native_encode_vector): Likewise.
17555 (native_interpret_fixed): Likewise.
17556 (native_interpret_real): Likewise.
17557 (native_interpret_complex): Likewise.
17558 (native_interpret_vector): Likewise.
17559 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17560 (simd_clone_adjust_argument_types): Likewise.
17561 (simd_clone_init_simd_arrays): Likewise.
17562 (simd_clone_adjust): Likewise.
17563 * stor-layout.c (layout_type): Likewise.
17564 * tree.c (build_minus_one_cst): Likewise.
17565 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17566 * tree-inline.c (estimate_move_cost): Likewise.
17567 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17568 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17569 (vectorizable_reduction): Likewise.
17570 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17571 (vect_recog_mixed_size_cond_pattern): Likewise.
17572 (check_bool_pattern): Likewise.
17573 (adjust_bool_pattern): Likewise.
17574 (search_type_for_mask_1): Likewise.
17575 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17576 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17577 (vectorizable_load): Likewise.
17578 (vectorizable_store): Likewise.
17579 * ubsan.c (ubsan_encode_value): Likewise.
17580 * varasm.c (output_constant): Likewise.
17582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17583 Alan Hayward <alan.hayward@arm.com>
17584 David Sherwood <david.sherwood@arm.com>
17586 * coretypes.h (scalar_mode): New class.
17587 * machmode.h (scalar_mode): Likewise.
17588 (scalar_mode::includes_p): New function.
17589 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17590 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17591 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17592 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17593 * expmed.c (store_bit_field_1): Likewise.
17594 (extract_bit_field_1): Likewise.
17595 * expr.c (write_complex_part): Likewise.
17596 (read_complex_part): Likewise.
17597 (emit_move_complex_push): Likewise.
17598 (expand_expr_real_2): Likewise.
17599 * function.c (assign_parm_setup_reg): Likewise.
17600 (assign_parms_unsplit_complex): Likewise.
17601 * optabs.c (expand_binop): Likewise.
17602 * rtlanal.c (subreg_get_info): Likewise.
17603 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17604 * varasm.c (output_constant_pool_2): Likewise.
17606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17607 Alan Hayward <alan.hayward@arm.com>
17608 David Sherwood <david.sherwood@arm.com>
17610 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17612 (expmed_mult_highpart_optab): Likewise.
17613 (expmed_mult_highpart): Likewise.
17615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17616 Alan Hayward <alan.hayward@arm.com>
17617 David Sherwood <david.sherwood@arm.com>
17619 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17620 instead of a machine_mode.
17621 (builtin_memset_read_str): Likewise.
17622 * builtins.c (c_readstr): Likewise.
17623 (builtin_memcpy_read_str): Likewise.
17624 (builtin_strncpy_read_str): Likewise.
17625 (builtin_memset_read_str): Likewise.
17626 (builtin_memset_gen_str): Likewise.
17627 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17628 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17629 instead of a machine_mode.
17630 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17632 (make_extraction): Likewise.
17633 (try_widen_shift_mode): Take and return scalar_int_modes instead
17635 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17636 a scalar_int_mode instead of a machine_mode.
17637 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17638 (avr_addr_space_pointer_mode): Likewise.
17639 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17640 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17641 (msp430_unwind_word_mode): Likewise.
17642 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17643 (spu_addr_space_pointer_mode): Likewise.
17644 (spu_addr_space_address_mode): Likewise.
17645 (spu_libgcc_cmp_return_mode): Likewise.
17646 (spu_libgcc_shift_count_mode): Likewise.
17647 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17648 (rl78_addr_space_pointer_mode): Likewise.
17649 (fl78_unwind_word_mode): Likewise.
17650 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17652 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17653 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17654 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17655 (mips_valid_pointer_mode): Likewise.
17656 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17657 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17658 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17660 (ft32_addr_space_address_mode): Likewise.
17661 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17662 scalar_int_mode instead of a machine_mode.
17663 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17665 (m32c_addr_space_address_mode): Likewise.
17666 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17667 (rs6000_eh_return_filter_mode): Likewise.
17668 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17669 (rs6000_eh_return_filter_mode): Likewise.
17670 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17671 (s390_libgcc_shift_count_mode): Likewise.
17672 (s390_unwind_word_mode): Likewise.
17673 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17675 * target.def (mode_rep_extended): Likewise.
17676 (valid_pointer_mode): Likewise.
17677 (addr_space.valid_pointer_mode): Likewise.
17678 (eh_return_filter_mode): Return a scalar_int_mode rather than
17680 (libgcc_cmp_return_mode): Likewise.
17681 (libgcc_shift_count_mode): Likewise.
17682 (unwind_word_mode): Likewise.
17683 (addr_space.pointer_mode): Likewise.
17684 (addr_space.address_mode): Likewise.
17685 * doc/tm.texi: Regenerate.
17686 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17688 (do_jump): Use scalar_int_mode for local variables.
17689 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17690 rather than a machine_mode.
17691 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17692 (scompare_loc_descriptor_wide): Likewise.
17693 (scompare_loc_descriptor_narrow): Likewise.
17694 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17696 * except.c (sjlj_emit_dispatch_table): Likewise.
17697 (expand_builtin_eh_copy_values): Likewise.
17698 * explow.c (convert_memory_address_addr_space_1): Likewise.
17699 Take a scalar_int_mode rather than a machine_mode.
17700 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17701 than a machine_mode.
17702 (memory_address_addr_space): Use scalar_int_mode for local variables.
17703 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17704 rather than a machine_mode.
17705 * expmed.c (mask_rtx): Likewise.
17706 (init_expmed_one_conv): Likewise.
17707 (expand_mult_highpart_adjust): Likewise.
17708 (extract_high_half): Likewise.
17709 (expmed_mult_highpart_optab): Likewise.
17710 (expmed_mult_highpart): Likewise.
17711 (expand_smod_pow2): Likewise.
17712 (expand_sdiv_pow2): Likewise.
17713 (emit_store_flag_int): Likewise.
17714 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17716 (extract_low_bits): Likewise.
17717 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17719 * expr.c (pieces_addr::adjust): Likewise.
17720 (can_store_by_pieces): Likewise.
17721 (store_by_pieces): Likewise.
17722 (clear_by_pieces_1): Likewise.
17723 (expand_expr_addr_expr_1): Likewise.
17724 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17725 (expand_expr_real_1): Likewise.
17726 (try_casesi): Likewise.
17727 * final.c (shorten_branches): Likewise.
17728 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17729 type of "mode" to machine_mode.
17730 * internal-fn.c (expand_arith_overflow_result_store): Take a
17731 scalar_int_mode rather than a machine_mode.
17732 (expand_mul_overflow): Use scalar_int_mode for local variables.
17733 * loop-doloop.c (doloop_modify): Likewise.
17734 (doloop_optimize): Likewise.
17735 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17736 than a machine_mode.
17737 (expand_doubleword_shift_condmove): Likewise.
17738 (expand_doubleword_shift): Likewise.
17739 (expand_doubleword_clz): Likewise.
17740 (expand_doubleword_popcount): Likewise.
17741 (expand_doubleword_parity): Likewise.
17742 (expand_absneg_bit): Use scalar_int_mode for local variables.
17743 (prepare_float_lib_cmp): Likewise.
17744 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17745 rather than a machine_mode.
17746 (convert_memory_address_addr_space): Likewise.
17747 (get_mode_bounds): Likewise.
17748 (get_address_mode): Return a scalar_int_mode rather than a
17750 * rtlanal.c (get_address_mode): Likewise.
17751 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17752 than a machine_mode.
17753 * targhooks.c (default_mode_rep_extended): Likewise.
17754 (default_valid_pointer_mode): Likewise.
17755 (default_addr_space_valid_pointer_mode): Likewise.
17756 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17757 than a machine_mode.
17758 (default_libgcc_cmp_return_mode): Likewise.
17759 (default_libgcc_shift_count_mode): Likewise.
17760 (default_unwind_word_mode): Likewise.
17761 (default_addr_space_pointer_mode): Likewise.
17762 (default_addr_space_address_mode): Likewise.
17763 * targhooks.h (default_eh_return_filter_mode): Likewise.
17764 (default_libgcc_cmp_return_mode): Likewise.
17765 (default_libgcc_shift_count_mode): Likewise.
17766 (default_unwind_word_mode): Likewise.
17767 (default_addr_space_pointer_mode): Likewise.
17768 (default_addr_space_address_mode): Likewise.
17769 (default_mode_rep_extended): Take a scalar_int_mode rather than
17771 (default_valid_pointer_mode): Likewise.
17772 (default_addr_space_valid_pointer_mode): Likewise.
17773 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17775 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17776 rather than a machine_mode.
17777 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17778 for local variables.
17779 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17780 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17781 than a machine_mode.
17783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17784 Alan Hayward <alan.hayward@arm.com>
17785 David Sherwood <david.sherwood@arm.com>
17787 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17788 the mode argument to scalar_int_mode.
17789 (do_jump_by_parts_zero_rtx): Likewise.
17790 (do_jump_by_parts_equality_rtx): Likewise.
17791 (do_jump_by_parts_greater): Take a mode argument.
17792 (do_jump_by_parts_equality): Likewise.
17793 (do_jump_1): Update calls accordingly.
17795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17796 Alan Hayward <alan.hayward@arm.com>
17797 David Sherwood <david.sherwood@arm.com>
17799 * is-a.h (safe_dyn_cast): New function.
17800 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17801 (jump_table_for_label): Likewise.
17802 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17803 instead of an rtx_insn *.
17804 (shorten_branches): Use dyn_cast instead of LABEL_P and
17805 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17806 rtx_jump_table_data::get_data_mode.
17807 (final_scan_insn): Likewise.
17809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17810 Alan Hayward <alan.hayward@arm.com>
17811 David Sherwood <david.sherwood@arm.com>
17813 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17814 trying to combine a full-register integer set with a subreg
17817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17818 Alan Hayward <alan.hayward@arm.com>
17819 David Sherwood <david.sherwood@arm.com>
17821 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17822 that is always either address_mode or pointer_mode.
17824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17825 Alan Hayward <alan.hayward@arm.com>
17826 David Sherwood <david.sherwood@arm.com>
17828 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17829 when the two are known to be equal.
17831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17832 Alan Hayward <alan.hayward@arm.com>
17833 David Sherwood <david.sherwood@arm.com>
17835 * simplify-rtx.c (simplify_const_unary_operation): Use
17836 is_a <scalar_int_mode> instead of checking for a nonzero
17837 precision. Forcibly convert op_mode to a scalar_int_mode
17838 in that case. More clearly differentiate the operand and
17839 result modes and use the former when deciding what the value
17840 of a count-bits operation should be. Use is_int_mode instead
17841 of checking for a MODE_INT. Remove redundant check for whether
17842 this mode has a zero precision.
17844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17845 Alan Hayward <alan.hayward@arm.com>
17846 David Sherwood <david.sherwood@arm.com>
17848 * optabs.c (widen_leading): Change the type of the mode argument
17849 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17850 (widen_bswap): Likewise.
17851 (expand_parity): Likewise.
17852 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17853 (expand_ffs): Likewise.
17854 (epand_unop): Check for scalar integer modes before calling the
17857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17858 Alan Hayward <alan.hayward@arm.com>
17859 David Sherwood <david.sherwood@arm.com>
17861 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17863 (expand_expr_real_1): Update call accordingly.
17865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17866 Alan Hayward <alan.hayward@arm.com>
17867 David Sherwood <david.sherwood@arm.com>
17869 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17870 value_mode arguments. Use scalar_int_mode internally.
17871 (store_bit_field_1): Rename the new integer mode from imode
17872 to op0_mode and use it instead of GET_MODE (op0). Update calls
17873 to store_split_bit_field, store_bit_field_using_insv and
17874 store_fixed_bit_field.
17875 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17876 Use scalar_int_mode internally. Use a bit count rather than a mode
17877 when calculating the largest bit size for get_best_mode.
17878 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17879 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17880 Remove assertion that OP0 has a scalar integer mode.
17881 (store_split_bit_field): Add op0_mode and value_mode arguments.
17882 Update calls to extract_fixed_bit_field.
17883 (extract_bit_field_using_extv): Add an op0_mode argument.
17884 Use scalar_int_mode internally.
17885 (extract_bit_field_1): Rename the new integer mode from imode to
17886 op0_mode and use it instead of GET_MODE (op0). Update calls to
17887 extract_split_bit_field, extract_bit_field_using_extv and
17888 extract_fixed_bit_field.
17889 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17890 to extract_split_bit_field and extract_fixed_bit_field_1.
17891 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17892 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17893 on the target mode.
17894 (extract_split_bit_field): Add an op0_mode argument. Update call
17895 to extract_fixed_bit_field.
17897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17898 Alan Hayward <alan.hayward@arm.com>
17899 David Sherwood <david.sherwood@arm.com>
17901 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17902 * explow.c (hard_function_value): Likewise.
17903 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17904 convert_to_mode call outside the loop.
17905 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17906 for the mode iterator. Require the mode specified by max_pieces
17908 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17910 (copy_blkmode_to_reg): Likewise.
17911 (set_storage_via_setmem): Likewise.
17912 * optabs.c (prepare_cmp_insn): Likewise.
17913 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17914 * stor-layout.c (finish_bitfield_representative): Likewise.
17916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17917 Alan Hayward <alan.hayward@arm.com>
17918 David Sherwood <david.sherwood@arm.com>
17920 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17921 * expr.c (convert_move): Use them.
17922 (convert_modes): Likewise.
17923 (store_expr_with_bounds): Likewise.
17925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17926 Alan Hayward <alan.hayward@arm.com>
17927 David Sherwood <david.sherwood@arm.com>
17929 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17930 parameter for the mode of "x". Remove the "known_x", "known_mode"
17931 and "known_ret" arguments. Change the type of the mode argument
17932 to scalar_int_mode.
17933 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17934 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17935 (reg_num_sign_bit_copies_for_combine): Likewise.
17936 * rtlanal.c (nonzero_bits1): Likewise.
17937 (num_sign_bit_copies1): Likewise.
17938 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17939 (reg_num_sign_bit_copies_general): Likewise.
17940 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17941 (reg_nonzero_bits_general): Likewise.
17943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17944 Alan Hayward <alan.hayward@arm.com>
17945 David Sherwood <david.sherwood@arm.com>
17947 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17948 than in subroutines. Return 1 for non-integer modes.
17949 (cached_num_sign_bit_copies): Change the type of the mode parameter
17950 to scalar_int_mode.
17951 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17952 classes. Handle CONST_INT_P first and then check whether X also
17953 has a scalar integer mode. Check the same thing for inner registers
17954 of a SUBREG and for values that are being extended or truncated.
17956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17957 Alan Hayward <alan.hayward@arm.com>
17958 David Sherwood <david.sherwood@arm.com>
17960 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17961 in subroutines. Return the mode mask for non-integer modes.
17962 (cached_nonzero_bits): Change the type of the mode parameter
17963 to scalar_int_mode.
17964 (nonzero_bits1): Likewise. Remove early exit for other mode
17965 classes. Handle CONST_INT_P first and then check whether X
17966 also has a scalar integer mode.
17968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17969 Alan Hayward <alan.hayward@arm.com>
17970 David Sherwood <david.sherwood@arm.com>
17972 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17973 Return a scalar_int_mode and assert that the size is greater than
17975 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17976 (op_by_pieces_d::op_by_pieces_d): Likewise.
17977 (op_by_pieces_d::run): Likewise.
17978 (can_store_by_pieces): Likewise.
17980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17981 Alan Hayward <alan.hayward@arm.com>
17982 David Sherwood <david.sherwood@arm.com>
17984 * combine.c (extract_left_shift): Add a mode argument and update
17986 (make_compound_operation_int): Change the type of the mode parameter
17987 to scalar_int_mode and update the call to extract_left_shift.
17989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17990 Alan Hayward <alan.hayward@arm.com>
17991 David Sherwood <david.sherwood@arm.com>
17993 * combine.c (simplify_and_const_int): Change the type of the mode
17994 parameter to scalar_int_mode.
17995 (simplify_and_const_int_1): Likewise. Update recursive call.
17997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17998 Alan Hayward <alan.hayward@arm.com>
17999 David Sherwood <david.sherwood@arm.com>
18001 * combine.c (simplify_compare_const): Check that the mode is a
18002 scalar_int_mode (rather than VOIDmode) before testing its
18004 (simplify_comparison): Move COMPARISON_P handling out of the
18005 loop and restrict the latter part of the loop to scalar_int_modes.
18006 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18007 and when considering SUBREG_REGs. Use is_int_mode instead of
18008 checking GET_MODE_CLASS against MODE_INT.
18010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18011 Alan Hayward <alan.hayward@arm.com>
18012 David Sherwood <david.sherwood@arm.com>
18014 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18015 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18016 shift_unit_mode and for modes involved in scalar shifts.
18018 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18019 Alan Hayward <alan.hayward@arm.com>
18020 David Sherwood <david.sherwood@arm.com>
18022 * combine.c (force_int_to_mode): New function, split out from...
18023 (force_to_mode): ...here. Keep xmode up-to-date and use it
18024 instead of GET_MODE (x).
18026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18027 Alan Hayward <alan.hayward@arm.com>
18028 David Sherwood <david.sherwood@arm.com>
18030 * optabs-query.h (extraction_insn::struct_mode): Change type to
18031 opt_scalar_int_mode and update comment.
18032 (extraction_insn::field_mode): Change type to scalar_int_mode.
18033 (extraction_insn::pos_mode): Likewise.
18034 * combine.c (make_extraction): Update accordingly.
18035 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18036 (get_optab_extraction_insn): Likewise.
18037 * recog.c (simplify_while_replacing): Likewise.
18038 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18039 parameter to opt_scalar_int_mode.
18041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18042 Alan Hayward <alan.hayward@arm.com>
18043 David Sherwood <david.sherwood@arm.com>
18045 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18046 to a scalar_int_mode instead of a machine_mode.
18047 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18048 (get_best_mode): Return a boolean and use a pointer argument to store
18049 the selected mode. Replace the limit mode parameter with a bit limit.
18050 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18051 for the values returned by bit_field_mode_iterator::next_mode.
18052 (store_bit_field): Update call to get_best_mode.
18053 (store_fixed_bit_field): Likewise.
18054 (extract_fixed_bit_field): Likewise.
18055 * expr.c (optimize_bitfield_assignment_op): Likewise.
18056 * fold-const.c (optimize_bit_field_compare): Likewise.
18057 (fold_truth_andor_1): Likewise.
18058 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18059 Update for new type of m_mode.
18060 (get_best_mode): As above.
18062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18063 Alan Hayward <alan.hayward@arm.com>
18064 David Sherwood <david.sherwood@arm.com>
18066 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18067 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18068 (store_bit_field): Check is_a <scalar_int_mode> before calling
18069 strict_volatile_bitfield_p.
18070 (extract_bit_field): Likewise.
18072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18073 Alan Hayward <alan.hayward@arm.com>
18074 David Sherwood <david.sherwood@arm.com>
18076 * target.def (cstore_mode): Return a scalar_int_mode.
18077 * doc/tm.texi: Regenerate.
18078 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18079 * targhooks.h (default_cstore_mode): Likewise.
18080 * targhooks.c (default_cstore_mode): Likewise, using a forced
18082 * expmed.c (emit_cstore): Expect the target of the cstore to be
18085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18086 Alan Hayward <alan.hayward@arm.com>
18087 David Sherwood <david.sherwood@arm.com>
18089 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18091 (niter_desc): Likewise mode.
18092 (iv_analyze): Add a mode parameter.
18094 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18095 and change its type to scalar_int_mode.
18096 * loop-iv.c: Update commentary at head of file.
18097 (iv_constant): Pass the mode paraeter before the rtx it describes
18098 and change its type to scalar_int_mode. Remove VOIDmode handling.
18099 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18100 (iv_extend): Likewise.
18101 (shorten_into_mode): Likewise.
18102 (iv_add): Use scalar_int_mode.
18103 (iv_mult): Likewise.
18104 (iv_shift): Likewise.
18105 (canonicalize_iv_subregs): Likewise.
18106 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18107 it describes and change its mode to scalar_int_mode. Also change
18108 the type of the returned inner_mode to scalar_int_mode.
18109 (get_biv_step): Likewise, turning outer_mode from a pointer
18110 into a direct parameter. Update call to get_biv_step_1.
18111 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18112 iv_constant and get_biv_step.
18113 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18114 and change its type to scalar_int_mode. Don't initialise iv->mode
18115 to VOIDmode and remove later checks for its still being VOIDmode.
18116 Update calls to iv_analyze_op and iv_analyze_expr. Check
18117 is_a <scalar_int_mode> when changing the mode under consideration.
18118 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18119 Update call to iv_analyze_expr.
18120 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18121 inner register is not also a scalar_int_mode. Update call to
18123 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18124 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18125 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18126 separate mode class checks. Update calls to iv_analyze. Remove
18127 fix-up of VOIDmodes after iv_analyze_biv.
18128 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18129 don't have a scalar_int_mode. Update call to biv_p.
18131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18132 Alan Hayward <alan.hayward@arm.com>
18133 David Sherwood <david.sherwood@arm.com>
18135 * cfgexpand.c (convert_debug_memory_address): Use
18136 as_a <scalar_int_mode>.
18137 * combine.c (expand_compound_operation): Likewise.
18138 (make_extraction): Likewise.
18139 (change_zero_ext): Likewise.
18140 (simplify_comparison): Likewise.
18141 * cse.c (cse_insn): Likewise.
18142 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18143 (mem_loc_descriptor): Likewise.
18144 (loc_descriptor): Likewise.
18145 * expmed.c (init_expmed_one_mode): Likewise.
18146 (synth_mult): Likewise.
18147 (emit_store_flag_1): Likewise.
18148 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18149 of a comparison with size.
18150 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18151 (reduce_to_bit_field_precision): Likewise.
18152 * function.c (expand_function_end): Likewise.
18153 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18154 * loop-doloop.c (doloop_modify): Likewise.
18155 * optabs.c (expand_binop): Likewise.
18156 (expand_unop): Likewise.
18157 (expand_copysign_absneg): Likewise.
18158 (prepare_cmp_insn): Likewise.
18159 (maybe_legitimize_operand): Likewise.
18160 * recog.c (const_scalar_int_operand): Likewise.
18161 * rtlanal.c (get_address_mode): Likewise.
18162 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18163 (simplify_cond_clz_ctz): Likewise.
18164 * tree-nested.c (get_nl_goto_field): Likewise.
18165 * tree.c (build_vector_type_for_mode): Likewise.
18166 * var-tracking.c (use_narrower_mode): Likewise.
18168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18169 Alan Hayward <alan.hayward@arm.com>
18170 David Sherwood <david.sherwood@arm.com>
18172 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18173 * builtins.c (expand_builtin_signbit): Use it.
18174 * cfgexpand.c (expand_debug_expr): Likewise.
18175 * dojump.c (do_jump): Likewise.
18176 (do_compare_and_jump): Likewise.
18177 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18178 * expmed.c (make_tree): Likewise.
18179 * expr.c (expand_expr_real_2): Likewise.
18180 (expand_expr_real_1): Likewise.
18181 (try_casesi): Likewise.
18182 * fold-const-call.c (fold_const_call_ss): Likewise.
18183 * fold-const.c (unextend): Likewise.
18184 (extract_muldiv_1): Likewise.
18185 (fold_single_bit_test): Likewise.
18186 (native_encode_int): Likewise.
18187 (native_encode_string): Likewise.
18188 (native_interpret_int): Likewise.
18189 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18190 * internal-fn.c (expand_addsub_overflow): Likewise.
18191 (expand_neg_overflow): Likewise.
18192 (expand_mul_overflow): Likewise.
18193 (expand_arith_overflow): Likewise.
18194 * match.pd: Likewise.
18195 * stor-layout.c (layout_type): Likewise.
18196 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18197 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18198 * tree-ssanames.c (get_range_info): Likewise.
18199 * tree-switch-conversion.c (array_value_type) Likewise.
18200 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18201 (vect_recog_divmod_pattern): Likewise.
18202 (vect_recog_mixed_size_cond_pattern): Likewise.
18203 * tree-vrp.c (extract_range_basic): Likewise.
18204 (simplify_float_conversion_using_ranges): Likewise.
18205 * tree.c (int_fits_type_p): Likewise.
18206 * ubsan.c (instrument_bool_enum_load): Likewise.
18207 * varasm.c (mergeable_string_section): Likewise.
18208 (narrowing_initializer_constant_valid_p): Likewise.
18209 (output_constant): Likewise.
18211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18212 Alan Hayward <alan.hayward@arm.com>
18213 David Sherwood <david.sherwood@arm.com>
18215 * machmode.h (NARROWEST_INT_MODE): New macro.
18216 * expr.c (alignment_for_piecewise_move): Use it instead of
18217 GET_CLASS_NARROWEST_MODE (MODE_INT).
18218 (push_block): Likewise.
18219 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18221 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18224 Alan Hayward <alan.hayward@arm.com>
18225 David Sherwood <david.sherwood@arm.com>
18227 * postreload.c (move2add_valid_value_p): Change the type of the
18228 mode parameter to scalar_int_mode.
18229 (move2add_use_add2_insn): Add a mode parameter and use it instead
18231 (move2add_use_add3_insn): Likewise.
18232 (reload_cse_move2add): Update accordingly.
18234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18235 Alan Hayward <alan.hayward@arm.com>
18236 David Sherwood <david.sherwood@arm.com>
18238 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18240 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18241 * optabs.c (expand_unop): Likewise.
18243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18244 Alan Hayward <alan.hayward@arm.com>
18245 David Sherwood <david.sherwood@arm.com>
18247 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18248 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18249 (popcount_loc_descriptor): Likewise.
18250 (bswap_loc_descriptor): Likewise.
18251 (rotate_loc_descriptor): Likewise.
18252 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18253 calling the functions above.
18255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18256 Alan Hayward <alan.hayward@arm.com>
18257 David Sherwood <david.sherwood@arm.com>
18259 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18261 (try_combine): Likewise.
18262 (simplify_if_then_else): Likewise.
18263 * cse.c (cse_insn): Likewise.
18264 * dwarf2out.c (mem_loc_descriptor): Likewise.
18265 * emit-rtl.c (gen_lowpart_common): Likewise.
18266 * simplify-rtx.c (simplify_truncation): Likewise.
18267 (simplify_binary_operation_1): Likewise.
18268 (simplify_const_relational_operation): Likewise.
18269 (simplify_ternary_operation): Likewise.
18270 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18273 Alan Hayward <alan.hayward@arm.com>
18274 David Sherwood <david.sherwood@arm.com>
18276 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18277 * reload.c (push_reload): Likewise.
18278 (find_reloads): Likewise.
18280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18281 Alan Hayward <alan.hayward@arm.com>
18282 David Sherwood <david.sherwood@arm.com>
18284 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18285 (make_compound_operation_int): Likewise.
18286 (change_zero_ext): Likewise.
18287 * expr.c (convert_move): Likewise.
18288 (convert_modes): Likewise.
18289 * fwprop.c (forward_propagate_subreg): Likewise.
18290 * loop-iv.c (get_biv_step_1): Likewise.
18291 * optabs.c (widen_operand): Likewise.
18292 * postreload.c (move2add_valid_value_p): Likewise.
18293 * recog.c (simplify_while_replacing): Likewise.
18294 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18295 (simplify_binary_operation_1): Likewise. Remove redundant
18296 mode equality check.
18298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18299 Alan Hayward <alan.hayward@arm.com>
18300 David Sherwood <david.sherwood@arm.com>
18302 * combine.c (combine_simplify_rtx): Add checks for
18303 is_a <scalar_int_mode>.
18304 (simplify_if_then_else): Likewise.
18305 (make_field_assignment): Likewise.
18306 (simplify_comparison): Likewise.
18307 * ifcvt.c (noce_try_bitop): Likewise.
18308 * loop-invariant.c (canonicalize_address_mult): Likewise.
18309 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18312 Alan Hayward <alan.hayward@arm.com>
18313 David Sherwood <david.sherwood@arm.com>
18315 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18316 is_a <scalar_int_mode> instead of != BLKmode.
18318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18319 Alan Hayward <alan.hayward@arm.com>
18320 David Sherwood <david.sherwood@arm.com>
18322 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18323 instead of != VOIDmode.
18324 * combine.c (if_then_else_cond): Likewise.
18325 (change_zero_ext): Likewise.
18326 * dwarf2out.c (mem_loc_descriptor): Likewise.
18327 (loc_descriptor): Likewise.
18328 * rtlanal.c (canonicalize_condition): Likewise.
18329 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18332 Alan Hayward <alan.hayward@arm.com>
18333 David Sherwood <david.sherwood@arm.com>
18335 * simplify-rtx.c (simplify_binary_operation_1): Use
18336 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18339 Alan Hayward <alan.hayward@arm.com>
18340 David Sherwood <david.sherwood@arm.com>
18342 * wide-int.h (int_traits<unsigned char>) New class.
18343 (int_traits<unsigned short>) Likewise.
18344 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18345 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18347 * combine.c (set_nonzero_bits_and_sign_copies): Use
18348 is_a <scalar_int_mode>.
18349 (find_split_point): Likewise.
18350 (combine_simplify_rtx): Likewise.
18351 (simplify_logical): Likewise.
18352 (expand_compound_operation): Likewise.
18353 (expand_field_assignment): Likewise.
18354 (make_compound_operation): Likewise.
18355 (extended_count): Likewise.
18356 (change_zero_ext): Likewise.
18357 (simplify_comparison): Likewise.
18358 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18359 (ucompare_loc_descriptor): Likewise.
18360 (minmax_loc_descriptor): Likewise.
18361 (mem_loc_descriptor): Likewise.
18362 (loc_descriptor): Likewise.
18363 * expmed.c (init_expmed_one_mode): Likewise.
18364 * lra-constraints.c (lra_constraint_offset): Likewise.
18365 * optabs.c (prepare_libcall_arg): Likewise.
18366 * postreload.c (move2add_note_store): Likewise.
18367 * reload.c (operands_match_p): Likewise.
18368 * rtl.h (load_extend_op): Likewise.
18369 * rtlhooks.c (gen_lowpart_general): Likewise.
18370 * simplify-rtx.c (simplify_truncation): Likewise.
18371 (simplify_unary_operation_1): Likewise.
18372 (simplify_binary_operation_1): Likewise.
18373 (simplify_const_binary_operation): Likewise.
18374 (simplify_const_relational_operation): Likewise.
18375 (simplify_subreg): Likewise.
18376 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18377 * var-tracking.c (adjust_mems): Likewise.
18378 (prepare_call_arguments): Likewise.
18380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18381 Alan Hayward <alan.hayward@arm.com>
18382 David Sherwood <david.sherwood@arm.com>
18384 * machmode.h (is_int_mode): New fuction.
18385 * combine.c (find_split_point): Use it.
18386 (combine_simplify_rtx): Likewise.
18387 (simplify_if_then_else): Likewise.
18388 (simplify_set): Likewise.
18389 (simplify_shift_const_1): Likewise.
18390 (simplify_comparison): Likewise.
18391 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18392 * cse.c (notreg_cost): Likewise.
18393 (cse_insn): Likewise.
18394 * cselib.c (cselib_lookup_1): Likewise.
18395 * dojump.c (do_jump_1): Likewise.
18396 (do_compare_rtx_and_jump): Likewise.
18397 * dse.c (get_call_args): Likewise.
18398 * dwarf2out.c (rtl_for_decl_init): Likewise.
18399 (native_encode_initializer): Likewise.
18400 * expmed.c (emit_store_flag_1): Likewise.
18401 (emit_store_flag): Likewise.
18402 * expr.c (convert_modes): Likewise.
18403 (store_field): Likewise.
18404 (expand_expr_real_1): Likewise.
18405 * fold-const.c (fold_read_from_constant_string): Likewise.
18406 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18407 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18408 * optabs.c (expand_binop): Likewise.
18409 (expand_unop): Likewise.
18410 (expand_abs_nojump): Likewise.
18411 (expand_one_cmpl_abs_nojump): Likewise.
18412 * simplify-rtx.c (mode_signbit_p): Likewise.
18413 (val_signbit_p): Likewise.
18414 (val_signbit_known_set_p): Likewise.
18415 (val_signbit_known_clear_p): Likewise.
18416 (simplify_relational_operation_1): Likewise.
18417 * tree.c (vector_type_mode): Likewise.
18419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18420 Alan Hayward <alan.hayward@arm.com>
18421 David Sherwood <david.sherwood@arm.com>
18423 * machmode.h (smallest_mode_for_size): Fix formatting.
18424 (smallest_int_mode_for_size): New function.
18425 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18426 instead of smallest_mode_for_size.
18427 * combine.c (make_extraction): Likewise.
18428 * config/arc/arc.c (arc_expand_movmem): Likewise.
18429 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18430 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18431 * config/s390/s390.c (s390_expand_insv): Likewise.
18432 * config/sparc/sparc.c (assign_int_registers): Likewise.
18433 * config/spu/spu.c (spu_function_value): Likewise.
18434 (spu_function_arg): Likewise.
18435 * coverage.c (get_gcov_type): Likewise.
18436 (get_gcov_unsigned_t): Likewise.
18437 * dse.c (find_shift_sequence): Likewise.
18438 * expmed.c (store_bit_field_1): Likewise.
18439 * expr.c (convert_move): Likewise.
18440 (store_field): Likewise.
18441 * internal-fn.c (expand_arith_overflow): Likewise.
18442 * optabs-query.c (get_best_extraction_insn): Likewise.
18443 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18444 * stor-layout.c (layout_type): Likewise.
18445 (initialize_sizetypes): Likewise.
18446 * targhooks.c (default_get_mask_mode): Likewise.
18447 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18450 Alan Hayward <alan.hayward@arm.com>
18451 David Sherwood <david.sherwood@arm.com>
18453 * machmode.h (opt_mode::else_blk): New function.
18454 (int_mode_for_mode): Declare.
18455 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18456 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18458 * cfgexpand.c (expand_debug_expr): Likewise.
18459 * combine.c (gen_lowpart_or_truncate): Likewise.
18460 (gen_lowpart_for_combine): Likewise.
18461 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18462 * config/avr/avr.c (avr_to_int_mode): Likewise.
18463 (avr_out_plus_1): Likewise.
18464 (avr_out_plus): Likewise.
18465 (avr_out_round): Likewise.
18466 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18467 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18468 (s390_expand_vcond): Likewise.
18469 * config/spu/spu.c (spu_split_immediate): Likewise.
18470 (spu_expand_mov): Likewise.
18471 * dse.c (get_stored_val): Likewise.
18472 * expmed.c (store_bit_field_1): Likewise.
18473 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18475 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18476 (extract_low_bits): Likewise.
18477 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18478 handling rather than repeating the check.
18479 (emit_group_store): Likewise.
18480 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18481 * optabs.c (expand_absneg_bit): Likewise.
18482 (expand_copysign_absneg): Likewise.
18483 (expand_copysign_bit): Likewise.
18484 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18485 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18486 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18487 * var-tracking.c (prepare_call_arguments): Likewise.
18488 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18489 int_mode_for_mode instead of mode_for_size.
18490 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18493 Alan Hayward <alan.hayward@arm.com>
18494 David Sherwood <david.sherwood@arm.com>
18496 * machmode.h (int_mode_for_size): New function.
18497 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18498 instead of mode_for_size.
18499 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18501 * combine.c (expand_field_assignment): Use int_mode_for_size
18502 instead of mode_for_size.
18503 (make_extraction): Likewise.
18504 (simplify_shift_const_1): Likewise.
18505 (simplify_comparison): Likewise.
18506 * dojump.c (do_jump): Likewise.
18507 * dwarf2out.c (mem_loc_descriptor): Likewise.
18508 * emit-rtl.c (init_derived_machine_modes): Likewise.
18509 * expmed.c (flip_storage_order): Likewise.
18510 (convert_extracted_bit_field): Likewise.
18511 * expr.c (copy_blkmode_from_reg): Likewise.
18512 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18513 * internal-fn.c (expand_mul_overflow): Likewise.
18514 * lower-subreg.c (simple_move): Likewise.
18515 * optabs-libfuncs.c (init_optabs): Likewise.
18516 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18517 * tree.c (vector_type_mode): Likewise.
18518 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18519 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18520 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18521 * tree-vect-stmts.c (vectorizable_load): Likewise.
18522 (vectorizable_store): Likewise.
18524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18525 Alan Hayward <alan.hayward@arm.com>
18526 David Sherwood <david.sherwood@arm.com>
18528 * coretypes.h (pod_mode): New type.
18529 (scalar_int_mode_pod): New typedef.
18530 * machmode.h (pod_mode): New class.
18531 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18532 * genmodes.c (emit_mode_int_n): Update accordingly.
18533 * lower-subreg.h (target_lower_subreg): Change type to
18534 scalar_int_mode_pod.
18535 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18536 scalar_int_mode_pod.
18538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18539 Alan Hayward <alan.hayward@arm.com>
18540 David Sherwood <david.sherwood@arm.com>
18542 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18543 machine_mode to scalar_int_mode.
18544 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18545 (rs6000_option_override_internal): Remove cast to int.
18546 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18547 machine_mode to scalar_int_mode.
18548 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18549 (rs6000_option_override_internal): Remove cast to int.
18550 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18551 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18553 * config/s390/s390.c (s390_expand_builtin): Likewise.
18554 * coretypes.h (scalar_int_mode): New type.
18555 (opt_scalar_int_mode): New typedef.
18556 * machmode.h (scalar_int_mode): New class.
18557 (scalar_int_mode::includes_p): New function.
18558 (byte_mode): Change type to scalar_int_mode.
18559 (word_mode): Likewise.
18560 (ptr_mode): Likewise.
18561 * emit-rtl.c (byte_mode): Likewise.
18562 (word_mode): Likewise.
18563 (ptr_mode): Likewise.
18564 (init_derived_machine_modes): Update accordingly.
18565 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18566 and MODE_PARTIAL_INT.
18567 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18568 opt_scalar_int_mode.
18570 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18571 Alan Hayward <alan.hayward@arm.com>
18572 David Sherwood <david.sherwood@arm.com>
18574 * target.def (libgcc_floating_mode_supported_p): Take a
18576 * doc/tm.texi: Regenerate.
18577 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18579 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18580 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18583 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18584 Alan Hayward <alan.hayward@arm.com>
18585 David Sherwood <david.sherwood@arm.com>
18587 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18588 * doc/tm.texi: Regenerate.
18589 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18590 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18591 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18592 * targhooks.h (default_floatn_mode): Likewise.
18593 * targhooks.c (default_floatn_mode): Likewise.
18594 * tree.c (build_common_tree_nodes): Update accordingly.
18596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18597 Alan Hayward <alan.hayward@arm.com>
18598 David Sherwood <david.sherwood@arm.com>
18600 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18601 (mode_iterator::iterate_p): Likewise.
18602 (mode_iterator::get_wider): Likewise.
18603 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18606 Alan Hayward <alan.hayward@arm.com>
18607 David Sherwood <david.sherwood@arm.com>
18609 * coretypes.h (opt_scalar_float_mode): New typedef.
18610 * machmode.h (float_mode_for_size): New function.
18611 * emit-rtl.c (double_mode): Delete.
18612 (init_emit_once): Use float_mode_for_size.
18613 * stor-layout.c (layout_type): Likewise.
18614 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18617 Alan Hayward <alan.hayward@arm.com>
18618 David Sherwood <david.sherwood@arm.com>
18620 * output.h (assemble_real): Take a scalar_float_mode.
18621 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18622 * config/arm/arm.md (consttable_4): Likewise.
18623 (consttable_8): Likewise.
18624 (consttable_16): Likewise.
18625 * config/mips/mips.md (consttable_float): Likewise.
18626 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18627 * varasm.c (assemble_real): Take a scalar_float_mode.
18628 (output_constant_pool_2): Update accordingly.
18629 (output_constant): Likewise.
18631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18632 Alan Hayward <alan.hayward@arm.com>
18633 David Sherwood <david.sherwood@arm.com>
18635 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18636 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18637 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18638 (native_encode_real): Likewise.
18639 (native_interpret_real): Likewise.
18640 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18641 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18644 Alan Hayward <alan.hayward@arm.com>
18645 David Sherwood <david.sherwood@arm.com>
18647 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18648 <scalar_float_mode>. Simplify.
18649 (gen_extend_conv_libfunc): Likewise.
18651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18652 Alan Hayward <alan.hayward@arm.com>
18653 David Sherwood <david.sherwood@arm.com>
18655 * coretypes.h (scalar_float_mode): New type.
18656 * machmode.h (mode_traits::from_int): Use machine_mode if
18657 USE_ENUM_MODES is defined.
18658 (is_a): New function.
18660 (dyn_cast): Likewise.
18661 (scalar_float_mode): New class.
18662 (scalar_float_mode::includes_p): New function.
18663 (is_float_mode): Likewise.
18664 * gdbhooks.py (MachineModePrinter): New class.
18665 (build_pretty_printer): Use it for scalar_float_mode.
18666 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18667 (format_helper::format_helper): Turn into a template.
18668 * genmodes.c (get_mode_class): New function.
18669 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18670 or machine_mode if none.
18671 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18672 as_a <scalar_float_mode>.
18673 * dwarf2out.c (mem_loc_descriptor): Likewise.
18674 (insert_float): Likewise.
18675 (add_const_value_attribute): Likewise.
18676 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18677 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18678 (expand_unop): Update accordingly.
18679 (expand_abs_nojump): Likewise.
18680 (expand_copysign_absneg): Take a scalar_float_mode.
18681 (expand_copysign_bit): Likewise.
18682 (expand_copysign): Update accordingly.
18684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18685 Alan Hayward <alan.hayward@arm.com>
18686 David Sherwood <david.sherwood@arm.com>
18688 * coretypes.h (opt_mode): New class.
18689 * machmode.h (opt_mode): Likewise.
18690 (opt_mode::else_void): New function.
18691 (opt_mode::require): Likewise.
18692 (opt_mode::exists): Likewise.
18693 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18694 (GET_MODE_2XWIDER_MODE): Likewise.
18695 (mode_iterator::get_wider): Update accordingly.
18696 (mode_iterator::get_2xwider): Likewise.
18697 (mode_iterator::get_known_wider): Likewise, turning into a template.
18698 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18699 forcing a wider mode to exist.
18700 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18701 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18702 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18703 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18704 * lower-subreg.c (init_lower_subreg): Likewise.
18705 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18706 on the final iteration.
18707 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18708 a wider mode exists before asking for a move pattern.
18709 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18710 forcing a wider mode to exist.
18711 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18712 returning false if no such mode exists.
18713 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18714 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18715 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18716 Avoid checking for a MODE_INT if we already know the mode is not a
18718 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18719 forcing a wider mode to exist.
18720 (expmed_mult_highpart_optab): Likewise.
18721 (expmed_mult_highpart): Likewise.
18722 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18724 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18725 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18726 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18727 * internal-fn.c (expand_mul_overflow): Update use of
18728 GET_MODE_2XWIDER_MODE.
18729 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18730 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18731 GET_MODE_WIDER_MODE.
18732 (convert_plusminus_to_widen): Likewise.
18733 * tree-switch-conversion.c (array_value_type): Likewise.
18734 * var-tracking.c (emit_note_insn_var_location): Likewise.
18735 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18736 Return false inside rather than outside the loop if no wider mode
18738 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18739 and GET_MODE_2XWIDER_MODE
18740 (can_compare_p): Use else_void.
18741 * gdbhooks.py (OptMachineModePrinter): New class.
18742 (build_pretty_printer): Use it for opt_mode.
18744 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18745 Alan Hayward <alan.hayward@arm.com>
18746 David Sherwood <david.sherwood@arm.com>
18748 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18749 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18751 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18752 Alan Hayward <alan.hayward@arm.com>
18753 David Sherwood <david.sherwood@arm.com>
18755 * machmode.h (mode_traits): New structure.
18756 (get_narrowest_mode): New function.
18757 (mode_iterator::start): Likewise.
18758 (mode_iterator::iterate_p): Likewise.
18759 (mode_iterator::get_wider): Likewise.
18760 (mode_iterator::get_known_wider): Likewise.
18761 (mode_iterator::get_2xwider): Likewise.
18762 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18763 (FOR_EACH_MODE): Likewise.
18764 (FOR_EACH_MODE_FROM): Likewise.
18765 (FOR_EACH_MODE_UNTIL): Likewise.
18766 (FOR_EACH_WIDER_MODE): Likewise.
18767 (FOR_EACH_2XWIDER_MODE): Likewise.
18768 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18769 * combine.c (simplify_comparison): Likewise
18770 * config/i386/i386.c (type_natural_mode): Likewise.
18771 * cse.c (cse_insn): Likewise.
18772 * dse.c (find_shift_sequence): Likewise.
18773 * emit-rtl.c (init_derived_machine_modes): Likewise.
18774 (init_emit_once): Likewise.
18775 * explow.c (hard_function_value): Likewise.
18776 * expmed.c (extract_fixed_bit_field_1): Likewise.
18777 (extract_bit_field_1): Likewise.
18778 (expand_divmod): Likewise.
18779 (emit_store_flag_1): Likewise.
18780 * expr.c (init_expr_target): Likewise.
18781 (convert_move): Likewise.
18782 (alignment_for_piecewise_move): Likewise.
18783 (widest_int_mode_for_size): Likewise.
18784 (emit_block_move_via_movmem): Likewise.
18785 (copy_blkmode_to_reg): Likewise.
18786 (set_storage_via_setmem): Likewise.
18787 (compress_float_constant): Likewise.
18788 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18789 * optabs-query.c (get_best_extraction_insn): Likewise.
18790 * optabs.c (expand_binop): Likewise.
18791 (expand_twoval_unop): Likewise.
18792 (expand_twoval_binop): Likewise.
18793 (widen_leading): Likewise.
18794 (widen_bswap): Likewise.
18795 (expand_parity): Likewise.
18796 (expand_unop): Likewise.
18797 (prepare_cmp_insn): Likewise.
18798 (prepare_float_lib_cmp): Likewise.
18799 (expand_float): Likewise.
18800 (expand_fix): Likewise.
18801 (expand_sfix_optab): Likewise.
18802 * postreload.c (move2add_use_add2_insn): Likewise.
18803 * reg-stack.c (reg_to_stack): Likewise.
18804 * reginfo.c (choose_hard_reg_mode): Likewise.
18805 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18806 * stor-layout.c (mode_for_size): Likewise.
18807 (smallest_mode_for_size): Likewise.
18808 (mode_for_vector): Likewise.
18809 (finish_bitfield_representative): Likewise.
18810 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18811 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18812 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18813 * var-tracking.c (prepare_call_arguments): Likewise.
18815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18816 Alan Hayward <alan.hayward@arm.com>
18817 David Sherwood <david.sherwood@arm.com>
18819 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18820 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18821 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18822 * machmode.h (mode_size): Move earlier in file.
18823 (mode_precision): Likewise.
18824 (mode_inner): Likewise.
18825 (mode_nunits): Likewise.
18826 (mode_unit_size): Likewise.
18827 (unit_unit_precision): Likewise.
18828 (mode_wider): Likewise.
18829 (mode_2xwider): Likewise.
18830 (machine_mode): New class.
18831 (mode_to_bytes): New function.
18832 (mode_to_bits): Likewise.
18833 (mode_to_precision): Likewise.
18834 (mode_to_inner): Likewise.
18835 (mode_to_unit_size): Likewise.
18836 (mode_to_unit_precision): Likewise.
18837 (mode_to_nunits): Likewise.
18838 (GET_MODE_SIZE): Use mode_to_bytes.
18839 (GET_MODE_BITSIZE): Use mode_to_bits.
18840 (GET_MODE_PRECISION): Use mode_to_precision.
18841 (GET_MODE_INNER): Use mode_to_inner.
18842 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18843 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18844 (GET_MODE_NUNITS): Use mode_to_nunits.
18845 * system.h (ALWAYS_INLINE): New macro.
18846 * config/powerpcspe/powerpcspe-c.c
18847 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18848 int for arg1_mode and arg2_mode.
18850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18851 Alan Hayward <alan.hayward@arm.com>
18852 David Sherwood <david.sherwood@arm.com>
18854 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18855 Prefix mode names with E_ in case statements.
18856 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18857 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18858 (aarch64_split_simd_move): Likewise.
18859 (aarch64_gen_storewb_pair): Likewise.
18860 (aarch64_gen_loadwb_pair): Likewise.
18861 (aarch64_gen_store_pair): Likewise.
18862 (aarch64_gen_load_pair): Likewise.
18863 (aarch64_get_condition_code_1): Likewise.
18864 (aarch64_constant_pool_reload_icode): Likewise.
18865 (get_rsqrte_type): Likewise.
18866 (get_rsqrts_type): Likewise.
18867 (get_recpe_type): Likewise.
18868 (get_recps_type): Likewise.
18869 (aarch64_gimplify_va_arg_expr): Likewise.
18870 (aarch64_simd_container_mode): Likewise.
18871 (aarch64_emit_load_exclusive): Likewise.
18872 (aarch64_emit_store_exclusive): Likewise.
18873 (aarch64_expand_compare_and_swap): Likewise.
18874 (aarch64_gen_atomic_cas): Likewise.
18875 (aarch64_emit_bic): Likewise.
18876 (aarch64_emit_atomic_swap): Likewise.
18877 (aarch64_emit_atomic_load_op): Likewise.
18878 (aarch64_evpc_trn): Likewise.
18879 (aarch64_evpc_uzp): Likewise.
18880 (aarch64_evpc_zip): Likewise.
18881 (aarch64_evpc_ext): Likewise.
18882 (aarch64_evpc_rev): Likewise.
18883 (aarch64_evpc_dup): Likewise.
18884 (aarch64_gen_ccmp_first): Likewise.
18885 (aarch64_gen_ccmp_next): Likewise.
18886 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18887 (alpha_emit_xfloating_libcall): Likewise.
18888 (emit_insxl): Likewise.
18889 (alpha_arg_type): Likewise.
18890 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18891 (arc_preferred_simd_mode): Likewise.
18892 (arc_secondary_reload): Likewise.
18893 (get_arc_condition_code): Likewise.
18894 (arc_print_operand): Likewise.
18895 (arc_legitimate_constant_p): Likewise.
18896 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18897 * config/arc/arc.md (casesi_load): Likewise.
18898 (casesi_compact_jump): Likewise.
18899 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18900 (cc_use_register): Likewise.
18901 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18902 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18903 (arm_init_iwmmxt_builtins): Likewise.
18904 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18905 (neon_expand_vector_init): Likewise.
18906 (arm_attr_length_move_neon): Likewise.
18907 (maybe_get_arm_condition_code): Likewise.
18908 (arm_emit_vector_const): Likewise.
18909 (arm_preferred_simd_mode): Likewise.
18910 (arm_output_iwmmxt_tinsr): Likewise.
18911 (thumb1_output_casesi): Likewise.
18912 (thumb2_output_casesi): Likewise.
18913 (arm_emit_load_exclusive): Likewise.
18914 (arm_emit_store_exclusive): Likewise.
18915 (arm_expand_compare_and_swap): Likewise.
18916 (arm_evpc_neon_vuzp): Likewise.
18917 (arm_evpc_neon_vzip): Likewise.
18918 (arm_evpc_neon_vrev): Likewise.
18919 (arm_evpc_neon_vtrn): Likewise.
18920 (arm_evpc_neon_vext): Likewise.
18921 (arm_validize_comparison): Likewise.
18922 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18923 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18924 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18925 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18926 (c6x_preferred_simd_mode): Likewise.
18927 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18928 (epiphany_rtx_costs): Likewise.
18929 * config/epiphany/predicates.md (proper_comparison_operator):
18931 * config/frv/frv.c (condexec_memory_operand): Likewise.
18932 (frv_emit_move): Likewise.
18933 (output_move_single): Likewise.
18934 (output_condmove_single): Likewise.
18935 (frv_hard_regno_mode_ok): Likewise.
18936 (frv_matching_accg_mode): Likewise.
18937 * config/h8300/h8300.c (split_adds_subs): Likewise.
18938 (h8300_rtx_costs): Likewise.
18939 (h8300_print_operand): Likewise.
18940 (compute_mov_length): Likewise.
18941 (output_logical_op): Likewise.
18942 (compute_logical_op_length): Likewise.
18943 (compute_logical_op_cc): Likewise.
18944 (h8300_shift_needs_scratch_p): Likewise.
18945 (output_a_shift): Likewise.
18946 (compute_a_shift_length): Likewise.
18947 (compute_a_shift_cc): Likewise.
18948 (expand_a_rotate): Likewise.
18949 (output_a_rotate): Likewise.
18950 * config/i386/i386.c (classify_argument): Likewise.
18951 (function_arg_advance_32): Likewise.
18952 (function_arg_32): Likewise.
18953 (function_arg_64): Likewise.
18954 (function_value_64): Likewise.
18955 (ix86_gimplify_va_arg): Likewise.
18956 (ix86_legitimate_constant_p): Likewise.
18957 (put_condition_code): Likewise.
18958 (split_double_mode): Likewise.
18959 (ix86_avx256_split_vector_move_misalign): Likewise.
18960 (ix86_expand_vector_logical_operator): Likewise.
18961 (ix86_split_idivmod): Likewise.
18962 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18963 (ix86_build_const_vector): Likewise.
18964 (ix86_build_signbit_mask): Likewise.
18965 (ix86_match_ccmode): Likewise.
18966 (ix86_cc_modes_compatible): Likewise.
18967 (ix86_expand_branch): Likewise.
18968 (ix86_expand_sse_cmp): Likewise.
18969 (ix86_expand_sse_movcc): Likewise.
18970 (ix86_expand_int_sse_cmp): Likewise.
18971 (ix86_expand_vec_perm_vpermi2): Likewise.
18972 (ix86_expand_vec_perm): Likewise.
18973 (ix86_expand_sse_unpack): Likewise.
18974 (ix86_expand_int_addcc): Likewise.
18975 (ix86_split_to_parts): Likewise.
18976 (ix86_vectorize_builtin_gather): Likewise.
18977 (ix86_vectorize_builtin_scatter): Likewise.
18978 (avx_vpermilp_parallel): Likewise.
18979 (inline_memory_move_cost): Likewise.
18980 (ix86_tieable_integer_mode_p): Likewise.
18981 (x86_maybe_negate_const_int): Likewise.
18982 (ix86_expand_vector_init_duplicate): Likewise.
18983 (ix86_expand_vector_init_one_nonzero): Likewise.
18984 (ix86_expand_vector_init_one_var): Likewise.
18985 (ix86_expand_vector_init_concat): Likewise.
18986 (ix86_expand_vector_init_interleave): Likewise.
18987 (ix86_expand_vector_init_general): Likewise.
18988 (ix86_expand_vector_set): Likewise.
18989 (ix86_expand_vector_extract): Likewise.
18990 (emit_reduc_half): Likewise.
18991 (ix86_emit_i387_round): Likewise.
18992 (ix86_mangle_type): Likewise.
18993 (ix86_expand_round_sse4): Likewise.
18994 (expand_vec_perm_blend): Likewise.
18995 (canonicalize_vector_int_perm): Likewise.
18996 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18997 (expand_vec_perm_1): Likewise.
18998 (expand_vec_perm_interleave3): Likewise.
18999 (expand_vec_perm_even_odd_pack): Likewise.
19000 (expand_vec_perm_even_odd_1): Likewise.
19001 (expand_vec_perm_broadcast_1): Likewise.
19002 (ix86_vectorize_vec_perm_const_ok): Likewise.
19003 (ix86_expand_vecop_qihi): Likewise.
19004 (ix86_expand_mul_widen_hilo): Likewise.
19005 (ix86_expand_sse2_abs): Likewise.
19006 (ix86_expand_pextr): Likewise.
19007 (ix86_expand_pinsr): Likewise.
19008 (ix86_preferred_simd_mode): Likewise.
19009 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19010 * config/i386/sse.md (*andnot<mode>3): Likewise.
19011 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19012 (*<code><mode>3): Likewise.
19013 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19014 (ia64_expand_atomic_op): Likewise.
19015 (ia64_arg_type): Likewise.
19016 (ia64_mode_to_int): Likewise.
19017 (ia64_scalar_mode_supported_p): Likewise.
19018 (ia64_vector_mode_supported_p): Likewise.
19019 (expand_vec_perm_broadcast): Likewise.
19020 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19021 (iq2000_function_arg_advance): Likewise.
19022 (iq2000_function_arg): Likewise.
19023 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19024 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19025 (m68k_libcall_value): Likewise.
19026 (m68k_function_value): Likewise.
19027 (sched_attr_op_type): Likewise.
19028 * config/mcore/mcore.c (mcore_output_move): Likewise.
19029 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19031 (microblaze_function_arg): Likewise.
19032 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19033 (mips_print_operand): Likewise.
19034 (mips_mode_ok_for_mov_fmt_p): Likewise.
19035 (mips_vector_mode_supported_p): Likewise.
19036 (mips_preferred_simd_mode): Likewise.
19037 (mips_expand_vpc_loongson_even_odd): Likewise.
19038 (mips_expand_vec_unpack): Likewise.
19039 (mips_expand_vi_broadcast): Likewise.
19040 (mips_expand_vector_init): Likewise.
19041 (mips_expand_vec_reduc): Likewise.
19042 (mips_expand_msa_cmp): Likewise.
19043 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19044 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19045 (cc_flags_for_mode): Likewise.
19046 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19047 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19048 (nds32_output_casesi_pc_relative): Likewise.
19049 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19050 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19051 (nvptx_gen_unpack): Likewise.
19052 (nvptx_gen_pack): Likewise.
19053 (nvptx_gen_shuffle): Likewise.
19054 (nvptx_gen_wcast): Likewise.
19055 (nvptx_preferred_simd_mode): Likewise.
19056 * config/pa/pa.c (pa_secondary_reload): Likewise.
19057 * config/pa/predicates.md (base14_operand): Likewise.
19058 * config/powerpcspe/powerpcspe-c.c
19059 (altivec_resolve_overloaded_builtin): Likewise.
19060 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19062 (rs6000_preferred_simd_mode): Likewise.
19063 (output_vec_const_move): Likewise.
19064 (rs6000_expand_vector_extract): Likewise.
19065 (rs6000_split_vec_extract_var): Likewise.
19066 (reg_offset_addressing_ok_p): Likewise.
19067 (rs6000_legitimate_offset_address_p): Likewise.
19068 (rs6000_legitimize_address): Likewise.
19069 (rs6000_emit_set_const): Likewise.
19070 (rs6000_const_vec): Likewise.
19071 (rs6000_emit_move): Likewise.
19072 (spe_build_register_parallel): Likewise.
19073 (rs6000_darwin64_record_arg_recurse): Likewise.
19074 (swap_selector_for_mode): Likewise.
19075 (spe_init_builtins): Likewise.
19076 (paired_init_builtins): Likewise.
19077 (altivec_init_builtins): Likewise.
19078 (do_load_for_compare): Likewise.
19079 (rs6000_generate_compare): Likewise.
19080 (rs6000_expand_float128_convert): Likewise.
19081 (emit_load_locked): Likewise.
19082 (emit_store_conditional): Likewise.
19083 (rs6000_output_function_epilogue): Likewise.
19084 (rs6000_handle_altivec_attribute): Likewise.
19085 (rs6000_function_value): Likewise.
19086 (emit_fusion_gpr_load): Likewise.
19087 (emit_fusion_p9_load): Likewise.
19088 (emit_fusion_p9_store): Likewise.
19089 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19090 (fusion_gpr_mem_load): Likewise.
19091 (fusion_addis_mem_combo_load): Likewise.
19092 (fusion_addis_mem_combo_store): Likewise.
19093 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19094 (fusion_gpr_mem_load): Likewise.
19095 (fusion_addis_mem_combo_load): Likewise.
19096 (fusion_addis_mem_combo_store): Likewise.
19097 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19099 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19100 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19101 (rs6000_preferred_simd_mode): Likewise.
19102 (output_vec_const_move): Likewise.
19103 (rs6000_expand_vector_extract): Likewise.
19104 (rs6000_split_vec_extract_var): Likewise.
19105 (reg_offset_addressing_ok_p): Likewise.
19106 (rs6000_legitimate_offset_address_p): Likewise.
19107 (rs6000_legitimize_address): Likewise.
19108 (rs6000_emit_set_const): Likewise.
19109 (rs6000_const_vec): Likewise.
19110 (rs6000_emit_move): Likewise.
19111 (rs6000_darwin64_record_arg_recurse): Likewise.
19112 (swap_selector_for_mode): Likewise.
19113 (paired_init_builtins): Likewise.
19114 (altivec_init_builtins): Likewise.
19115 (rs6000_expand_float128_convert): Likewise.
19116 (emit_load_locked): Likewise.
19117 (emit_store_conditional): Likewise.
19118 (rs6000_output_function_epilogue): Likewise.
19119 (rs6000_handle_altivec_attribute): Likewise.
19120 (rs6000_function_value): Likewise.
19121 (emit_fusion_gpr_load): Likewise.
19122 (emit_fusion_p9_load): Likewise.
19123 (emit_fusion_p9_store): Likewise.
19124 * config/rx/rx.c (rx_gen_move_template): Likewise.
19125 (flags_from_mode): Likewise.
19126 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19127 (s390_slb_comparison): Likewise.
19128 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19129 (s390_vector_mode_supported_p): Likewise.
19130 (s390_cc_modes_compatible): Likewise.
19131 (s390_match_ccmode_set): Likewise.
19132 (s390_canonicalize_comparison): Likewise.
19133 (s390_emit_compare_and_swap): Likewise.
19134 (s390_branch_condition_mask): Likewise.
19135 (s390_rtx_costs): Likewise.
19136 (s390_secondary_reload): Likewise.
19137 (__SECONDARY_RELOAD_CASE): Likewise.
19138 (s390_expand_cs): Likewise.
19139 (s390_preferred_simd_mode): Likewise.
19140 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19141 * config/sh/sh.c (sh_print_operand): Likewise.
19142 (dump_table): Likewise.
19143 (sh_secondary_reload): Likewise.
19144 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19145 * config/sh/sh.md (casesi_worker_1): Likewise.
19146 (casesi_worker_2): Likewise.
19147 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19148 (fcc_comparison_operator): Likewise.
19149 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19150 (emit_soft_tfmode_cvt): Likewise.
19151 (sparc_preferred_simd_mode): Likewise.
19152 (output_cbranch): Likewise.
19153 (sparc_print_operand): Likewise.
19154 (sparc_expand_vec_perm_bmask): Likewise.
19155 (vector_init_bshuffle): Likewise.
19156 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19157 (spu_vector_mode_supported_p): Likewise.
19158 (spu_expand_insv): Likewise.
19159 (spu_emit_branch_or_set): Likewise.
19160 (spu_handle_vector_attribute): Likewise.
19161 (spu_builtin_splats): Likewise.
19162 (spu_builtin_extract): Likewise.
19163 (spu_builtin_promote): Likewise.
19164 (spu_expand_sign_extend): Likewise.
19165 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19166 (tilegx_simd_int): Likewise.
19167 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19168 (tilepro_simd_int): Likewise.
19169 * config/v850/v850.c (const_double_split): Likewise.
19170 (v850_print_operand): Likewise.
19171 (ep_memory_offset): Likewise.
19172 * config/vax/vax.c (vax_rtx_costs): Likewise.
19173 (vax_output_int_move): Likewise.
19174 (vax_output_int_add): Likewise.
19175 (vax_output_int_subtract): Likewise.
19176 * config/visium/predicates.md (visium_branch_operator): Likewise.
19177 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19178 (visium_print_operand_address): Likewise.
19179 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19180 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19181 (xtensa_expand_conditional_branch): Likewise.
19182 (xtensa_copy_incoming_a7): Likewise.
19183 (xtensa_output_literal): Likewise.
19184 * dfp.c (decimal_real_maxval): Likewise.
19185 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19188 Alan Hayward <alan.hayward@arm.com>
19189 David Sherwood <david.sherwood@arm.com>
19191 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19192 (mode_nunits_inline): Likewise.
19193 (mode_inner_inline): Likewise.
19194 (mode_unit_size_inline): Likewise.
19195 (mode_unit_precision_inline): Likewise.
19196 (emit_insn_modes_h): Likewise. Also emit a #define of the
19198 (emit_mode_wider): Add an E_ prefix to mode names.
19199 (emit_mode_complex): Likewise.
19200 (emit_mode_inner): Likewise.
19201 (emit_mode_adjustments): Likewise.
19202 (emit_mode_int_n): Likewise.
19203 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19204 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19205 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19206 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19207 (CRC32_BUILTIN, ENTRY): Likewise.
19208 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19209 (aarch64_pop_regs): Likewise.
19210 (aarch64_process_components): Likewise.
19211 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19212 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19213 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19214 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19215 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19216 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19218 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19219 (pcmpistr): Likewise.
19220 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19221 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19222 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19224 * config/rl78/rl78.c (mduc_regs): Likewise.
19225 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19226 (htm_expand_builtin): Likewise.
19227 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19228 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19229 * config/xtensa/xtensa.c (print_operand): Likewise.
19230 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19231 (NUM_MODE_VECTOR_INT): Likewise.
19232 * genoutput.c (null_operand): Likewise.
19233 (output_operand_data): Likewise.
19234 * genrecog.c (print_parameter_value): Likewise.
19235 * lra.c (debug_operand_data): Likewise.
19237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19239 * dwarf2out.c (scompare_loc_descriptor_wide)
19240 (scompare_loc_descriptor_narrow): New functions, split out from...
19241 (scompare_loc_descriptor): ...here.
19242 * expmed.c (emit_store_flag_int): New function, split out from...
19243 (emit_store_flag): ...here.
19245 2017-08-30 Richard Biener <rguenther@suse.de>
19247 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19248 (dwarf2out_early_finish): Move setting of AT_pubnames from
19249 early debug output to early finish.
19251 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19253 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19254 and -mdata-region to the assembler.
19256 2017-08-30 Richard Biener <rguenther@suse.de>
19258 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19260 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19261 (dwarf2out_early_global_decl): Only generate a DIE for the
19262 abstract origin if it doesn't already exist or is a declaration DIE.
19263 (resolve_addr): Do not add the linkage name twice when
19264 generating a stub DIE for the DW_TAG_GNU_call_site target.
19266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19268 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19269 Use machine_mode rather than int for arg1_mode.
19271 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19274 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19275 that the second argument of the built-in functions to unpack
19276 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19277 switch statement instead a lot of if statements.
19278 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19279 Allow 64-bit values to be in Altivec registers as well as
19280 traditional floating point registers.
19281 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19283 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19285 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19286 MAX_REGS_PER_ADDRESS == 1.
19288 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19290 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19291 * config/i386/i386.c (ix86_option_override_internal): Simplify
19292 setting of opts->x_flag_entry.
19294 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19295 Jakub Jelinek <jakub@redhat.com>
19296 Richard Biener <rguenther@suse.de>
19298 PR tree-optimization/81503
19299 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19300 folded constant fits in the target type; reorder tests for clarity.
19302 2017-08-29 Martin Liska <mliska@suse.cz>
19304 * passes.def: Include pass_lower_switch.
19305 * stmt.c (dump_case_nodes): Remove and move to
19306 tree-switch-conversion.
19307 (case_values_threshold): Likewise.
19308 (expand_switch_as_decision_tree_p): Likewise.
19309 (emit_case_decision_tree): Likewise.
19310 (expand_case): Likewise.
19311 (balance_case_nodes): Likewise.
19312 (node_has_low_bound): Likewise.
19313 (node_has_high_bound): Likewise.
19314 (node_is_bounded): Likewise.
19315 (emit_case_nodes): Likewise.
19316 (struct simple_case_node): New struct.
19317 (add_case_node): Remove.
19318 (emit_case_dispatch_table): Use vector instead of case_list.
19319 (reset_out_edges_aux): Remove.
19320 (compute_cases_per_edge): Likewise.
19321 (expand_case): Build list of simple_case_node.
19322 (expand_sjlj_dispatch_table): Use it.
19323 * tree-switch-conversion.c (struct case_node): Moved from
19324 stmt.c and adjusted.
19325 (emit_case_nodes): Likewise.
19326 (node_has_low_bound): Likewise.
19327 (node_has_high_bound): Likewise.
19328 (node_is_bounded): Likewise.
19329 (case_values_threshold): Likewise.
19330 (reset_out_edges_aux): Likewise.
19331 (compute_cases_per_edge): Likewise.
19332 (add_case_node): Likewise.
19333 (dump_case_nodes): Likewise.
19334 (balance_case_nodes): Likewise.
19335 (expand_switch_as_decision_tree_p): Likewise.
19336 (emit_jump): Likewise.
19337 (emit_case_decision_tree): Likewise.
19338 (try_switch_expansion): Likewise.
19339 (do_jump_if_equal): Likewise.
19340 (emit_cmp_and_jump_insns): Likewise.
19341 (fix_phi_operands_for_edge): New function.
19342 (record_phi_operand_mapping): Likewise.
19343 (class pass_lower_switch): New pass.
19344 (pass_lower_switch::execute): New function.
19345 (make_pass_lower_switch): Likewise.
19346 (conditional_probability):
19347 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19348 * tree-pass.h: Add make_pass_lower_switch.
19350 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19353 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19356 2017-08-29 Richard Biener <rguenther@suse.de>
19358 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19359 we do not add a DW_AT_inline attribute twice.
19360 (gen_subprogram_die): Remove code setting DW_AT_inline on
19361 DECL_ABSTRACT_P nodes.
19363 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19365 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19366 calls to internal functions.
19367 (gimplify_modify_expr): Likewise.
19368 * tree-call-cdce.c (use_internal_fn): Likewise.
19369 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19370 (convert_to_divmod): Set the nothrow flag.
19371 * tree-if-conv.c (predicate_mem_writes): Likewise.
19372 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19373 (vectorizable_call): Likewise.
19374 (vectorizable_store): Likewise.
19375 (vectorizable_load): Likewise.
19376 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19377 (vect_recog_mask_conversion_pattern): Likewise.
19379 2017-08-29 Martin Liska <mliska@suse.cz>
19382 * gcc.c (driver_handle_option): Add new argument.
19383 * opts-common.c (handle_option): Pass
19384 target_option_override_hook.
19385 * opts-global.c (lang_handle_option): Add new option.
19386 (set_default_handlers): Add new argument.
19387 (decode_options): Likewise.
19388 * opts.c (target_handle_option): Likewise.
19389 (common_handle_option): Call target_option_override_hook.
19390 * opts.h (struct cl_option_handler_func): Add hook for
19391 target option override.
19392 (struct cl_option_handlers): Likewise.
19393 (set_default_handlers): Add new argument.
19394 (decode_options): Likewise.
19395 (common_handle_option): Likewise.
19396 (target_handle_option): Likewise.
19397 * toplev.c (toplev::main): Pass targetm.target_option.override
19400 2017-08-29 Richard Biener <rguenther@suse.de>
19401 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19403 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19404 life to the active subtree.
19406 2017-08-28 Jeff Law <law@redhat.com>
19408 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19409 derive_equivalences.
19410 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19411 Code moved into....
19412 (edge_info::derive_equivalences): New private member function
19414 * tree-ssa-dom.c (class edge_info): Changed from a struct
19415 to a class. Add ctor/dtor, methods and data members.
19416 (edge_info::edge_info): Renamed from allocate_edge_info.
19417 Initialize additional members.
19418 (edge_info::~edge_info): New.
19419 (free_dom_edge_info): Delete the edge info.
19420 (record_edge_info): Use new class & associated member functions.
19421 Tighten forms for testing for edge equivalences.
19422 (record_temporary_equivalences): Iterate over the simple
19423 equivalences rather than assuming there's only one per edge.
19424 (cprop_into_successor_phis): Iterate over the simple
19425 equivalences rather than assuming there's only one per edge.
19426 (optimize_stmt): Use operand_equal_p rather than pointer
19427 equality for mini-DSE code.
19429 2017-08-28 Nathan Sidwell <nathan@acm.org>
19431 * gcc.c (execute): Fold SIGPIPE handling into switch
19432 statement. Adjust internal error message.
19434 2017-08-28 Richard Biener <rguenther@suse.de>
19437 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19438 Do nothing for removed DIEs.
19440 2017-08-28 Richard Biener <rguenther@suse.de>
19442 PR tree-optimization/81977
19443 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19446 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19449 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19450 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19451 using targetm.gen_mem_thread_fence.
19453 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19456 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19457 predicate to register_operand. Reorder operands.
19458 (*btr<mode>): Ditto.
19459 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19460 (*btr<mode>_mask): Ditto.
19462 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19464 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19465 * config/rs6000/xmmintrin.h: New file.
19466 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19468 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19471 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19472 parameter and_insn and return it.
19473 (recombine_lvx_pattern): Insert a copy to ensure availability of
19474 the base register of the copied masking operation at the point of
19475 the instruction replacement.
19476 (recombine_stvx_pattern): Likewise.
19478 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19480 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19481 undocumented switches.
19482 (-mpower9-dform-vector): Likewise.
19483 (-mpower9-dform): Likewise.
19484 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19485 comments to delete references to -mpower9-dform* switches.
19486 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19487 Delete reference to -mpower9-dform* switches, test for
19488 -mpower9-vector instead.
19489 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19490 (OTHER_P9_VECTOR_MASKS): Likewise.
19491 (POWERPC_MASKS): Likewise.
19492 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19493 tests against -mpower9-dform* to -mpower9-vector. Delete code
19494 that checked for -mpower9-dform* consistancy with other options.
19495 Add test for -mpower9-misc to enable other power9 switches.
19496 (rs6000_init_hard_regno_mode_ok): Likewise.
19497 (rs6000_option_override_internal): Likewise.
19498 (rs6000_emit_prologue): Likewise.
19499 (rs6000_emit_epilogue): Likewise.
19500 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19501 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19502 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19504 (emit_fusion_p9_store): Likewise.
19505 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19506 resetting these macros if the assembler does not support ISA 3.0
19508 (TARGET_P9_DFORM_VECTOR): Likewise.
19509 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19510 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19512 2017-08-25 Alan Modra <amodra@gmail.com>
19514 PR rtl-optimization/81747
19515 * cse.c (cse_extended_basic_block): Don't attempt to record
19516 equivalences for degenerate conditional branches that branch
19517 to their fall-through.
19519 2017-08-24 Martin Sebor <msebor@redhat.com>
19521 PR middle-end/81908
19522 * gimple-fold.c (size_must_be_zero_p): New function.
19523 (gimple_fold_builtin_memory_op): Call it.
19525 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19527 * config/rs6000/mm_malloc.h: New file.
19529 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19531 PR tree-optimization/81913
19532 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19533 analysis when either IVs in condition can wrap.
19535 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19537 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19538 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19540 2017-08-24 Richard Biener <rguenther@suse.de>
19543 * targhooks.c (default_target_can_inline_p): Properly
19544 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19545 is present and always compare.
19546 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19547 infer -mfpmath=sse from TARGET_SSE_P.
19548 (ix86_can_inline_p): Properly use target_option_default_node when
19549 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19551 2017-08-24 Richard Biener <rguenther@suse.de>
19554 * dwarf2out.c (output_die): Handle flag_generate_offload like
19556 (output_comp_unit): Likewise.
19557 (gen_array_type_die): Likewise.
19558 (dwarf2out_early_finish): Likewise.
19559 (note_variable_value_in_expr): Likewise.
19560 (dwarf2out_finish): Likewise. Adjust assert.
19561 * cgraphunit.c (symbol_table::compile): Move setting of
19562 flag_generate_offload earlier ...
19563 (symbol_table::finalize_compilation_unit): ... here, before
19564 early debug finalization.
19566 2017-08-24 Richard Biener <rguenther@suse.de>
19568 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19569 and ipa-fnsummary.h.
19570 (ix86_can_inline_p): When ix86_fpmath flags do not match
19571 check whether the callee uses FP math at all.
19573 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19575 PR middle-end/81931
19576 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19577 instead of TYPE_PRECISION.
19579 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19580 Alan Hayward <alan.hayward@arm.com>
19581 David Sherwood <david.sherwood@arm.com>
19583 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19585 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19587 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19589 (ix86_frame::stack_realign_allocate): New field.
19590 (struct machine_frame_state): Modify comments.
19591 (machine_frame_state::sp_realigned_fp_end): New field.
19592 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19593 layout calculation.
19594 (sp_valid_at): Add assertion to assure no attempt to access invalid
19595 offset of a realigned stack.
19596 (fp_valid_at): Likewise.
19597 (choose_baseaddr): Modify comments.
19598 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19599 ix86_expand_prologue.
19600 (ix86_expand_prologue): Modify stack realignment and allocation.
19601 (ix86_expand_epilogue): Modify comments.
19602 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19603 avx2_runtime, avx512f, and avx512f_runtime.
19605 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19607 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19608 (mstackrealign): Do not init to -1.
19609 * config/i386/i386.c (ix86_option_override_internal):
19610 Check opts_set, not opts when setting default value of
19611 opts->x_ix86_force_align_arg_pointer.
19613 2017-08-23 Richard Biener <rguenther@suse.de>
19615 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19616 lang_hooks.decl_printable_name.
19617 * print-rtl-function.c (print_rtx_function): Likewise.
19618 * tree-pretty-print.c (dump_function_header): Likewise.
19620 2017-08-23 Richard Biener <rguenther@suse.de>
19623 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19624 -g0 at compile-time.
19626 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19628 PR middle-end/19706
19629 * doc/sourcebuild.texi (Other hardware attributes):
19632 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19634 PR middle-end/19706
19635 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19636 Fix single-use check.
19638 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19640 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19642 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19644 * doc/install.texi: Modify to add more details on running selected
19647 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19649 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19650 is combined with -mabi=ms.
19651 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19654 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19656 PR tree-optimization/81488
19657 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19658 and cached_basis fields.
19659 (MAX_SPREAD): New constant.
19660 (alloc_cand_and_find_basis): Initialize new fields.
19661 (clear_visited): New function.
19662 (create_phi_basis_1): Rename from create_phi_basis, set visited
19663 and cached_basis fields.
19664 (create_phi_basis): New wrapper function.
19665 (phi_add_costs_1): Rename from phi_add_costs, add spread
19666 parameter, set visited field, short-circuit when limits reached.
19667 (phi_add_costs): New wrapper function.
19668 (record_phi_increments_1): Rename from record_phi_increments, set
19670 (record_phi_increments): New wrapper function.
19671 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19672 (phi_incr_cost): New wrapper function.
19673 (all_phi_incrs_profitable_1): Rename from
19674 all_phi_incrs_profitable, set visited field.
19675 (all_phi_incrs_profitable): New wrapper function.
19677 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19678 Alan Hayward <alan.hayward@arm.com>
19679 David Sherwood <david.sherwood@arm.com>
19681 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19682 that takes the outer and inner modes.
19683 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19684 comparison as the canonical test for a paradoxical subreg.
19685 * combine.c (simplify_set): Use paradoxical_subreg_p.
19686 (make_extraction): Likewise.
19687 (force_to_mode): Likewise.
19688 (rtx_equal_for_field_assignment_p): Likewise.
19689 (gen_lowpart_for_combine): Likewise.
19690 (simplify_comparison): Likewise.
19691 * cse.c (equiv_constant): Likewise.
19692 * expmed.c (store_bit_field_1): Likewise.
19693 * final.c (alter_subreg): Likewise.
19694 * fwprop.c (propagate_rtx): Likewise.
19695 (forward_propagate_subreg): Likewise.
19696 * ira-conflicts.c (ira_build_conflicts): Likewise.
19697 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19698 * lra-constraints.c (curr_insn_transform): Likewise.
19699 (split_reg): Likewise.
19700 * lra-eliminations.c (move_plus_up): Likewise.
19701 (lra_eliminate_regs_1): Likewise.
19702 * recog.c (general_operand): Likewise.
19703 * ree.c (combine_reaching_defs): Likewise.
19704 * reload.c (push_reload): Likewise.
19705 (find_reloads): Likewise.
19706 * reload1.c (elimination_effects): Likewise.
19707 (compute_reload_subreg_offset): Likewise.
19708 (choose_reload_regs): Likewise.
19709 * rtlanal.c (subreg_lsb_1): Likewise.
19710 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19711 (simplify_subreg): Likewise.
19712 * var-tracking.c (track_loc_p): Likewise.
19713 * emit-rtl.c (byte_lowpart_offset): Likewise.
19714 (paradoxical_subreg_p): Delete out-of-line definition.
19716 2017-08-22 Jeff Law <law@redhat.com>
19718 PR tree-optimization/81741
19719 PR tree-optimization/71947
19720 * tree-ssa-dom.c: Include tree-inline.h.
19721 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19722 equivalences if one is more expensive to compute than the other.
19723 * tree-ssa-scopedtables.h (class const_or_copies): Make
19724 record_const_or_copy_raw method private.
19725 (class avail_exprs_stack): New method simplify_binary_operation.
19726 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19727 avail_exprs_stack::simplify_binary_operation as needed.
19728 (avail_exprs_stack::simplify_binary_operation): New function.
19730 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19732 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19733 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19734 (DOT_SYMBOLS): Likewise.
19735 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19736 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19737 (RS6000_ABI_NAME): Likewise.
19738 (TARGET_CMODEL): Likewise.
19739 (TOC_SECTION_ASM_OP): Likewise.
19740 (SET_CMODEL): New macro.
19741 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19743 2017-08-22 Richard Biener <rguenther@suse.de>
19745 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19746 to allow for free-lang-data replacements similar to verify_type_variant.
19748 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19749 Alan Hayward <alan.hayward@arm.com>
19750 David Sherwood <david.sherwood@arm.com>
19752 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19753 VOIDmode for the LABEL_REF.
19755 2017-08-22 Richard Biener <rguenther@suse.de>
19757 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19759 2017-08-22 Richard Biener <rguenther@suse.de>
19761 * common.opt (feliminate-dwarf2-dups): Ignore.
19762 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19763 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19764 same_die_p_wrap, compute_section_prefix,
19765 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19766 (comdat_symbol_id, comdat_symbol_number): Likewise.
19767 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19769 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19770 (output_die): Mark unreachable path unreachable.
19771 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19772 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19773 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19774 (dwarf2out_early_finish): Likewise.
19776 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19778 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19780 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19783 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19784 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19785 (avr_attribute_table) <io, io_low, address>: Initialize
19786 .decl_required with true.
19788 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19790 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19791 undocumented debugging options.
19792 (-mvsx-scalar-double): Likewise.
19793 (-mallow-df-permute): Likewise.
19794 (-mvectorize-builtins): Likewise.
19795 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19796 (rs6000_builtin_vectorized_function): Likewise.
19797 (rs6000_builtin_md_vectorized_function): Likewise.
19798 (rs6000_opt_vars): Likewise.
19800 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19803 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19804 (*btrq_imm): Rename from *btrq.
19805 (*btcq_imm): Rename from *btcq.
19806 (btsc): New code attribute.
19807 (*<btsc><mode>): New insn pattern.
19808 (*btr<mode>): Ditto.
19809 (*<btsc><mode>_mask): New insn_and_split pattern.
19810 (*btr<mode>_mask): Ditto.
19812 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19813 Alan Hayward <alan.hayward@arm.com>
19814 David Sherwood <david.sherwood@arm.com>
19816 * function.c (pad_below): Simplify padding calculation.
19818 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19819 Alan Hayward <alan.hayward@arm.com>
19820 David Sherwood <david.sherwood@arm.com>
19822 * target.def (function_prologue): Remove frame size argument.
19823 (function_epilogue): Likewise.
19824 * doc/tm.texi: Regenerate.
19825 * final.c (final_start_function): Update call to function_prologue.
19826 (final_end_function): Update call to function_epilogue.
19827 (default_function_pro_epilogue): Remove frame size argument.
19828 * output.h (default_function_pro_epilogue): Likewise.
19829 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19830 (arm_output_function_prologue): Likewise.
19831 * config/frv/frv.c (frv_function_prologue): Likewise.
19832 (frv_function_epilogue): Likewise.
19833 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19834 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19835 (ia64_output_function_epilogue): Likewise.
19836 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19837 (m32r_output_function_epilogue): Likewise.
19838 * config/microblaze/microblaze.c (microblaze_function_prologue)
19839 (microblaze_function_epilogue): Likewise.
19840 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19841 (mips_output_function_epilogue): Likewise.
19842 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19843 (mmix_target_asm_function_epilogue): Likewise.
19844 * config/msp430/msp430.c (msp430_start_function): Likewise.
19845 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19846 (nds32_asm_function_epilogue): Likewise.
19847 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19848 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19849 (pa_output_function_epilogue): Likewise.
19850 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19851 (rs6000_output_function_epilogue): Likewise.
19852 * config/rl78/rl78.c (rl78_start_function): Likewise.
19853 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19854 (rs6000_output_function_epilogue): Likewise.
19855 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19856 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19857 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19858 (sparc_asm_function_epilogue): Likewise.
19860 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19862 * tree.h (type_has_mode_precision_p): New function.
19863 * convert.c (convert_to_integer_1): Use it.
19864 * expr.c (expand_expr_real_2): Likewise.
19865 (expand_expr_real_1): Likewise.
19866 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19867 * match.pd: Likewise.
19868 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19869 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19870 * tree-tailcall.c (process_assignment): Likewise.
19871 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19872 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19873 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19874 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19875 (vectorizable_assignment): Likewise.
19876 (vectorizable_shift): Likewise.
19877 (vectorizable_operation): Likewise.
19878 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19880 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19882 * match.pd: Add pow (C, x) simplification.
19884 2017-08-21 Richard Biener <rguenther@suse.de>
19886 PR tree-optimization/81900
19887 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19888 for blocks with abnormal predecessors.
19889 (compute_antic): Do not set visited flag prematurely.
19891 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19894 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19896 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19898 * stor-layout.h (vector_type_mode): Move to...
19899 * tree.h (vector_type_mode): ...here.
19900 * stor-layout.c (vector_type_mode): Move to...
19901 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19903 2017-08-21 Richard Biener <rguenther@suse.de>
19905 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19906 register_external_die hooks.
19907 (debug_false_tree_charstarstar_uhwistar): Declare.
19908 (debug_nothing_tree_charstar_uhwi): Likewise.
19909 * debug.c (do_nothing_debug_hooks): Adjust.
19910 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19911 (debug_nothing_tree_charstar_uhwi): Likewise.
19912 * dbxout.c (dbx_debug_hooks): Adjust.
19913 (xcoff_debug_hooks): Likewise.
19914 * sdbout.c (sdb_debug_hooks): Likewise.
19915 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19916 * dwarf2out.c (macinfo_label_base): New global.
19917 (dwarf2out_register_external_die): New function for the
19918 register_external_die hook.
19919 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19920 (dwarf2_debug_hooks): Use them.
19921 (dwarf2_lineno_debug_hooks): Adjust.
19922 (struct die_struct): Add with_offset flag.
19923 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19924 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19925 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19926 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19927 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19928 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19929 defining section names for the early LTO debug variants.
19930 (reset_indirect_string): New helper.
19931 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19932 (print_dw_val): Add support for offsetted symbol references.
19933 (get_ultimate_context): Split out from is_cxx.
19934 (is_cxx): Use get_ultimate_context.
19935 (is_fortran): Add decl overload.
19936 (compute_comp_unit_symbol): Split out worker from
19937 compute_section_prefix.
19938 (compute_section_prefix): Call compute_comp_unit_symbol and
19939 set comdat_type_p here.
19940 (output_die): Skip DIE symbol output for the LTO added one.
19941 Handle DIE symbol references with offset.
19942 (output_comp_unit): Guard section name mangling properly.
19943 For LTO debug sections emit a symbol at the section beginning
19944 which we use to refer to its DIEs.
19945 (add_abstract_origin_attribute): For DIEs registered via
19946 dwarf2out_register_external_die directly refer to the early
19947 DIE rather than indirectly through the shadow one we created.
19948 Remove obsolete call to dwarf2out_abstract_function for
19949 non-function/block origins.
19950 (gen_array_type_die): When generating early LTO debug do
19951 not emit DW_AT_string_length.
19952 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19953 late when in LTO. As suggested place a gcc_unreachable for
19954 the DECL_ABSTRACT_P case.
19955 (gen_subprogram_die): Avoid another specification DIE
19956 for early built declarations/definitions for the late LTO case.
19957 (gen_variable_die): Add type references for late duplicated VLA dies
19959 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19960 we have the abstract instance already.
19961 (process_scope_var): Adjust decl DIE contexts in LTO which
19962 first puts them in limbo.
19963 (gen_decl_die): Do not generate type DIEs late apart from
19964 types for VLAs or for decls we do not yet have a DIE. Do not
19965 call dwarf2out_abstract_function late.
19966 (dwarf2out_early_global_decl): Make sure to create DIEs
19967 for abstract instances of a decl first.
19968 (dwarf2out_late_global_decl): Adjust comment.
19969 (output_macinfo_op): With multiple macro sections use
19970 macinfo_label_base to distinguish labels.
19971 (output_macinfo): Likewise. Update macinfo_label_base.
19972 Pass in the line info label.
19973 (note_variable_value_in_expr): When generating LTO resolve
19974 all variable values here by generating DIEs as needed.
19975 (init_sections_and_labels): Add early LTO debug flag parameter
19976 and generate different sections and names if set. Add generation
19977 counter for the labels so we can have multiple of them.
19978 (reset_dies): Helper to allow DIEs to be output multiple times.
19979 (dwarf2out_finish): When outputting DIEs to the fat part of an
19980 LTO object first reset DIEs.
19981 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19982 (modified_type_die): Check for decl_ultimate_origin being self
19984 (gen_type_die_with_usage): Likewise.
19985 (gen_typedef_die): Allow decl_ultimate_origin being self.
19986 (set_decl_abstract_flags): Remove.
19987 (set_block_abstract_flags): Likewise.
19988 (dwarf2out_abstract_function): Treat the early generated DIEs
19989 as the abstract copy and only add DW_AT_inline and
19990 DW_AT_artificial here and call set_decl_origin_self.
19991 If the DIE has an abstract origin don't do anything.
19992 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19993 if we have none yet (Go fails to build one, PR78628).
19994 (variably_modified_type_p): Prevent endless recursion for Ada
19995 cyclic pointer types.
19996 * lto-streamer-in.c: Include debug.h.
19997 (dref_queue): New global.
19998 (lto_read_tree_1): Stream in DIE references.
19999 (lto_input_tree): Register DIE references.
20000 (input_function): Stream DECL_DEBUG_ARGS.
20001 * lto-streamer-out.c: Include debug.h.
20002 (lto_write_tree_1): Output DIE references.
20003 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20004 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20005 (output_function): Stream DECL_DEBUG_ARGS.
20006 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20007 Stream DECL_ABSTRACT_ORIGIN.
20008 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20009 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20010 DECL_CONTEXT for file-scope decls.
20011 * lto-streamer.h (struct dref_entry): Declare.
20012 (dref_queue): Likewise.
20013 * cfgexpand.c (pass_expand::execute): Do not call the
20014 outlining_inline_function hook here.
20015 * lto-wrapper.c (debug_obj): New global.
20016 (tool_cleanup): Unlink it if required.
20017 (debug_objcopy): New function.
20018 (run_gcc): Handle early debug sections in the IL files by
20019 extracting them to separate files, partially linkin them and
20020 feeding the result back as result to the linker.
20021 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20022 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20023 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20024 sections into a separate segment.
20025 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20027 (darwin_asm_dwarf_section): Likewise.
20028 (darwin_asm_output_dwarf_offset): Likewise.
20029 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20031 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20032 Alan Hayward <alan.hayward@arm.com>
20033 David Sherwood <david.sherwood@arm.com>
20035 * read-md.h (md_reader::record_potential_iterator_use): Replace
20036 pointer argument with an rtx and an index.
20037 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20038 (apply_mode_iterator): Likewise.
20039 (apply_code_iterator): Likewise.
20040 (apply_int_iterator): Likewise.
20041 (apply_subst_iterator): Likewise.
20042 (record_iterator_use): Likewise.
20043 (record_attribute_use): Likewise.
20044 (md_reader::record_potential_iterator_use): Likewise. Update calls
20045 to record_iterator_use and apply_iterator.
20046 (iterator_use): Replace ptr with x and index.
20047 (attribute_use): Likewise.
20048 (apply_attribute_uses): Update calls to apply_iterator.
20049 (apply_iterators): Likewise. Update initialization of iterator_use.
20050 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20051 and record_potential_iterator_use.
20052 (rtx_reader::read_rtx_operand): Likewise.
20054 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20055 Alan Hayward <alan.hayward@arm.com>
20056 David Sherwood <david.sherwood@arm.com>
20058 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20061 2017-08-21 Richard Biener <rguenther@suse.de>
20063 PR middle-end/81884
20064 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20065 at struct end conservatively when comparing common bases.
20067 2017-08-21 Richard Biener <rguenther@suse.de>
20069 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20070 (mem_ref_in_stmt): Remove.
20071 (determine_max_movement): Use ref index to get at the reference.
20072 (invariantness_dom_walker::before_dom_children): Deal with
20073 lim data already initialized.
20074 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20076 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20078 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20079 (TARGET_ISA_ROUND): Ditto.
20080 (TARGET_ROUND): Ditto.
20081 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20082 * config/i386/i386.md: Ditto.
20083 * config/i386/sse.md: Ditto.
20084 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20085 with OPTION_MASK_ISA_SSE4_1.
20087 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20090 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20091 __builtin_ia32_lzcnt_u16.
20093 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20096 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20097 (rs6000_set_current_function): Rewrite function to use it.
20099 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20102 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20103 and TYPE_WARN_IF_NOT_ALIGN.
20104 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20105 (handle_warn_if_not_align): New.
20106 (place_union_field): Call handle_warn_if_not_align.
20107 (place_field): Call handle_warn_if_not_align.
20108 Copy TYPE_WARN_IF_NOT_ALIGN.
20109 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20110 (layout_type): Likewise.
20111 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20113 (tree_decl_common): Add warn_if_not_align.
20114 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20115 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20116 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20117 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20118 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20119 * doc/extend.texi: Document warn_if_not_aligned attribute.
20120 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20122 2017-08-17 Martin Liska <mliska@suse.cz>
20125 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20126 (get_data_dependence): Use it as pointer type.
20127 (distribute_loop): Likewise.
20129 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20131 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20132 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20133 (p8_vmrgow_<mode>_direct): New define_insn.
20134 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20135 handle endianness for vmrgew and vmrgow permute patterns.
20137 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20139 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20140 * config/rs6000/rs6000-cpus.def: Remove comment.
20141 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20142 (POWERPC_MASKS): Likewise.
20143 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20144 use of TARGET_VSX_TIMODE.
20145 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20146 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20147 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20148 (rs6000_option_override_internal): Remove dead code.
20149 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20150 (rs6000_legitimize_reload_address): Likewise.
20151 (rs6000_legitimate_address_p): Likewise.
20152 (rs6000_opt_masks): Delete "vsx-timode".
20153 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20154 from function comment.
20155 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20156 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20157 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20159 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20160 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20161 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20162 (VSX_TI): Likewise.
20164 (define_peephole2): Likewise.
20166 2017-08-17 Martin Sebor <msebor@redhat.com>
20169 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20170 past the end of an array.
20171 (test_pp_format): Add test cases.
20173 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20175 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20176 missing ECF_NOTHROW flags.
20178 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20181 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20182 operands residing in integer registers.
20183 (*vsx_le_perm_load_<mode>): Likewise.
20184 (*vsx_le_perm_store_<mode>): Likewise.
20185 (define_peephole2): Add peepholes to optimize the above.
20187 2017-08-17 Marek Polacek <polacek@redhat.com>
20189 PR middle-end/81814
20190 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20191 to mimic what shorten_compare did. Change the return type to bool.
20192 (fold_cond_expr_with_comparison): Update call to
20193 operand_equal_for_comparison_p.
20194 (fold_ternary_loc): Likewise.
20196 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20198 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20200 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20202 2017-08-17 Richard Biener <rguenther@suse.de>
20204 * tree-ssa-structalias.c (solve_graph): When propagating
20205 to successors update the graphs succ edges and avoid duplicate work.
20207 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20210 * config/i386/i386.c (ix86_option_override_internal): Save target
20211 specific options after ix86_stack_protector_guard_reg was changed.
20213 2017-08-17 Richard Biener <rguenther@suse.de>
20215 PR tree-optimization/81827
20216 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20217 (new_var_info): Initialize it conservatively.
20218 (get_call_vi): Mark register vars.
20219 (new_scalar_tmp_constraint_exp): Likewise.
20220 (handle_rhs_call): Likewise.
20221 (handle_const_call): Likewise.
20222 (create_function_info_for): Likewise.
20223 (solve_constraints): Sort varinfos to separate register from
20224 non-register vars to pack points-to solution bitmaps during
20227 2017-08-17 Marek Polacek <polacek@redhat.com>
20229 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20231 2017-08-17 Richard Biener <rguenther@suse.de>
20233 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20234 to true when overflow is undefined and we saturated the result.
20236 2017-08-17 Alan Modra <amodra@gmail.com>
20239 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20240 Don't use store multiple if only one reg needs saving.
20241 (interesting_frame_related_regno): New function.
20242 (rs6000_frame_related): Don't emit frame info for regs that
20244 (rs6000_emit_epilogue): Likewise.
20246 2017-08-16 Nathan Sidwell <nathan@acm.org>
20248 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20249 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20250 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20251 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20252 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20253 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20254 (verify_type): Adjust for TYPE_BINFO move.
20255 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20256 process TYPE_BINFO directly.
20257 (hash_tree): Likewise.
20258 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20260 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20263 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20265 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20267 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20270 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20271 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20272 Add ix86_binary_operator_ok to insn constraint.
20273 (*iordi_1_bts): Ditto.
20274 (*xordi_1_btc): Ditto.
20275 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20276 Update corresponding peephole2 pattern.
20280 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20282 PR tree-optimization/81832
20283 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20284 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20286 2017-08-16 Marek Polacek <polacek@redhat.com>
20289 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20290 regarding size_zero_node.
20292 2017-08-16 Martin Liska <mliska@suse.cz>
20295 * config.gcc: Respect previously set extra_objs in case
20298 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20300 PR tree-optimization/81835
20301 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20302 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20303 not depend on the phi.
20305 2017-08-16 Alan Modra <amodra@gmail.com>
20307 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20310 2017-08-16 Alan Modra <amodra@gmail.com>
20312 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20313 (save_reg_p): ..into this. Update all callers.
20314 (first_reg_to_save): Simplify.
20316 2017-08-16 Alan Modra <amodra@gmail.com>
20318 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20321 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20325 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20326 constant count if that count is less than 32.
20328 2017-08-15 Nathan Sidwell <nathan@acm.org>
20330 * gcc.c (execute): Emit friendlier message if inferior is killed
20331 by an external cause.
20333 2017-08-15 Richard Biener <rguenther@suse.de>
20335 PR tree-optimization/81790
20336 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20337 CONSTRUCTORs from simplifying and VN.
20339 2017-08-14 Martin Sebor <msebor@redhat.com>
20341 * builtin-attrs.def: Add comments.
20343 2017-08-14 Martin Sebor <msebor@redhat.com>
20346 * doc/extend.texi (attribute nonstring): Document new attribute.
20348 2017-08-14 Martin Sebor <msebor@redhat.com>
20351 * tree-diagnostic.c (default_tree_printer): Handle %G.
20352 * gimple-pretty-print.h (percent_G_format): Declare new function.
20353 * gimple-pretty-print.c (percent_G_format): Define.
20354 * tree-pretty-print.c (percent_K_format): Add argument.
20356 2017-08-14 Martin Sebor <msebor@redhat.com>
20358 PR translation/79998
20359 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20360 Remove a stray space.
20362 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20365 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20366 (*iordi_1_bts): Ditto.
20367 (*xordi_1_btc): Ditto.
20369 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20372 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20373 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20375 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20376 quoted strings, and make more translator-friendly.
20377 (darwin_rs6000_override_options): Likewise.
20378 (rs6000_option_override_internal): Likewise.
20379 (rs6000_return_in_memory): Fix overlong line.
20380 (init_cmulative_args): Use quoted strings, and make more
20381 translator-friendly.
20382 (rs6000_pass_by_reference): Fix overlong line.
20383 (def_builtin): Use quoted strings.
20384 (altivec_expand_predicate_builtin): Use quoted strings, and make
20385 more translator-friendly.
20386 (htm_expand_builtin): Use quoted strings.
20387 (cpu_expand_builtin): Use quoted strings, and make more
20388 translator-friendly.
20389 (altivec_expand_builtin): Likewise.
20390 (paired_expand_predicate_builtin): Likewise.
20391 (rs6000_invalid_builtin): Likewise.
20392 (builtin_function_type): Use quoted strings.
20393 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20394 more translator-friendly.
20395 (rs6000_trampoline_init): Likewise.
20396 (rs6000_handle_altivec_attribute): Likewise.
20397 (rs6000_inner_target_options): Use quoted strings.
20398 (rs6000_disable_incompatible_switches): Likewise.
20399 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20400 strings, and make more translator-friendly.
20401 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20403 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20405 PR tree-optimization/81799
20406 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20407 cond_expr to simple gimple operand.
20409 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20411 PR middle-end/46932
20412 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20414 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20418 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20419 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20420 TARGET_GASISR_PROLOGUES.
20421 * config/avr/avr.c (avr_option_override): Same.
20422 (avr_pass_pre_proep::execute): Same.
20424 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20427 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20428 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20430 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20432 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20433 define_insn_and_split. Split to a memory load from 0 in
20434 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20435 using PTR mode iterator.
20436 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20437 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20438 (*add_tp_<mode>): Redefine as define_insn_and_split.
20439 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20440 address space. Merge with *add_tp_x32 using PTR mode iterator.
20441 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20442 Split to an add with a memory load from 0 in
20443 DEFAULT_TLS_SEG_REG address space.
20445 2017-08-12 Andrew Pinski <apinski@cavium.com>
20447 * config/aarch64/aarch64-option-extensions.def (rdma):
20448 Fix feature string to what Linux prints out in /proc/cpuinfo.
20450 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20453 * dwarf2out.c (modified_type_die): For C typedef types that have
20454 an ultimate origin, process the ultimate origin instead of the
20456 (gen_typedef_die): Assert that input DECLs have no ultimate
20458 (gen_type_die_with_usage): For typedef variants that have an
20459 ultimate origin, just call gen_decl_die on the original DECL.
20460 (process_scope_var): Avoid creating DIEs for local typedefs and
20461 concrete static variables.
20463 2017-08-12 Alan Modra <amodra@gmail.com>
20467 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20468 match gnu-user.h startfile.
20469 (ENDFILE_LINUX_SPEC): Similarly.
20471 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20474 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20476 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20478 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20479 * config/aarch64/aarch64.md (mov<mode>): Change.
20480 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20481 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20482 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20484 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20486 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20487 for storage order barriers.
20489 2017-08-11 Martin Liska <mliska@suse.cz>
20492 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20493 variables of void type.
20495 2017-08-11 Martin Liska <mliska@suse.cz>
20497 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20498 TARGET_SUPPORTS_ALIASES.
20499 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20500 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20501 (optimize_weakref): Likewise.
20502 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20503 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20504 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20506 2017-08-11 Martin Liska <mliska@suse.cz>
20509 * config/i386/i386.c (make_resolver_func): Do complete
20510 refactoring of the function.
20512 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20515 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20516 * config/i386/i386.c (ix86_stack_protect_guard): Use
20517 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20518 * doc/invoke.texi (x86 Options): Document
20519 -mstack-protector-guard-symbol= option.
20521 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20523 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20524 * config/i386/i386.c (ix86_split_stack_guard): New function.
20525 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20526 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20527 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20528 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20529 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20530 (split_stack_space_check): Call ix86_split_stack_guard.
20532 2017-08-10 Martin Sebor <msebor@redhat.com>
20534 * print-tree.c (print_node): Print location using the established
20536 Replace spaces with colons.
20537 (debug_raw, debug): Ditto.
20539 2017-08-10 Martin Sebor <msebor@redhat.com>
20542 * pretty-print.c (pp_format): Correct the handling of %s precision.
20544 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20547 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20549 (ix86_finalize_stack_frame_flags): This. Also clear
20550 frame_pointer_needed if -fno-omit-frame-pointer is used without
20552 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20553 with ix86_finalize_stack_frame_flags.
20554 (ix86_expand_epilogue): Likewise.
20555 (ix86_expand_split_stack_prologue): Likewise.
20556 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20558 2017-08-10 Martin Liska <mliska@suse.cz>
20561 * c-attribs.c (handle_target_attribute):
20562 Report warning for an empty string argument of target attribute.
20564 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20567 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20569 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20570 or DECL_NONLOCAL labels.
20571 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20572 or DECL_NONLOCAL labels here.
20574 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20576 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20577 to indicate when early gimple folding has been disabled.
20578 (rs6000_gimple_fold_builtin): Add debug content.
20579 (rs6000_invalid_builtin): Fix whitespace.
20580 (rs6000_expand_builtin): Fix whitespace.
20581 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20583 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20586 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20587 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20589 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20591 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20592 (qdf24xx): Likewise.
20593 * config/aarch64/aarch64.md: Include falkor.md.
20594 * config/aarch64/falkor.md: New.
20596 2017-08-09 Marek Polacek <polacek@redhat.com>
20599 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20600 * diagnostic.c (emit_diagnostic): Add a comment.
20601 (emit_diagnostic_valist): New function.
20603 2017-08-09 Marek Polacek <polacek@redhat.com>
20606 * input.c (make_location): New overload.
20607 * input.h (make_location): Declare.
20609 2017-08-08 Alan Modra <amodra@gmail.com>
20610 H.J. Lu <hongjiu.lu@intel.com>
20613 * gcc.c (NO_PIE_SPEC): Delete.
20614 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20616 (LINK_PIE_SPEC): ..to here.
20617 (LINK_COMMAND_SPEC): Support -no-pie.
20618 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20619 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20621 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20622 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20623 (ENDFILE_CRTEND_SPEC): Similarly.
20625 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20628 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20629 (mstack-protector-guard-offset=): Ditto.
20630 * config/i386/i386.c (ix86_option_override): Handle
20631 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20633 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20634 ix86_stack_protect_guard_offset variables.
20635 (TARGET_STACK_PROTECT_GUARD): Always define.
20636 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20637 and -mstack-protector-guard-offset= options.
20639 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20641 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20642 boundary case for the last candidate.
20644 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20646 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20647 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20649 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20651 PR middle-end/19706
20652 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20653 * config/aarch64/aarch64-builtins.c
20654 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20655 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20656 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20658 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20659 Andrew Pinski <pinskia@gmail.com>
20661 PR middle-end/19706
20662 * internal-fn.def (XORSIGN): New.
20663 * optabs.def (xorsign_optab): New.
20664 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20665 (convert_expand_mult_copysign): New.
20666 (pass_optimize_widening_mul::execute): Call
20667 convert_expand_mult_copysign.
20669 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20671 PR tree-optimization/81354
20672 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20673 Insert on edges rather than explicitly creating landing pads.
20674 (analyze_candidates_and_replace): Commit edge inserts.
20676 2017-08-08 Richard Biener <rguenther@suse.de>
20678 PR middle-end/81719
20679 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20680 (expand_simple_operations): Also look through ADDR_EXPRs with
20681 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20683 2017-08-08 Richard Biener <rguenther@suse.de>
20685 PR tree-optimization/81723
20686 * tree-vect-slp.c (struct bst_traits): New hash traits.
20687 (bst_fail): New global.
20688 (vect_build_slp_tree_2): New worker, split out from ...
20689 (vect_build_slp_tree): ... this now wrapping it with using
20690 bst_fail set to cache SLP tree build fails. Properly handle
20692 (vect_analyze_slp_instance): Allocate and free bst_fail.
20694 2017-08-08 Martin Liska <mliska@suse.cz>
20697 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20698 LABEL_DECLs that can be from a different function.
20700 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20702 PR tree-optimization/81744
20703 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20704 loop's number of iterations.
20706 2017-08-08 Martin Liska <mliska@suse.cz>
20708 * asan.c: Include header files.
20709 * attribs.c (build_decl_attribute_variant): New function moved
20711 (build_type_attribute_qual_variant): Likewise.
20712 (cmp_attrib_identifiers): Likewise.
20713 (simple_cst_list_equal): Likewise.
20714 (omp_declare_simd_clauses_equal): Likewise.
20715 (attribute_value_equal): Likewise.
20716 (comp_type_attributes): Likewise.
20717 (build_type_attribute_variant): Likewise.
20718 (lookup_ident_attribute): Likewise.
20719 (remove_attribute): Likewise.
20720 (merge_attributes): Likewise.
20721 (merge_type_attributes): Likewise.
20722 (merge_decl_attributes): Likewise.
20723 (merge_dllimport_decl_attributes): Likewise.
20724 (handle_dll_attribute): Likewise.
20725 (attribute_list_equal): Likewise.
20726 (attribute_list_contained): Likewise.
20727 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20728 (lookup_attribute_by_prefix): Likewise.
20729 * bb-reorder.c: Include header files.
20730 * builtins.c: Likewise.
20731 * calls.c: Likewise.
20732 * cfgexpand.c: Likewise.
20733 * cgraph.c: Likewise.
20734 * cgraphunit.c: Likewise.
20735 * convert.c: Likewise.
20736 * dwarf2out.c: Likewise.
20737 * final.c: Likewise.
20738 * fold-const.c: Likewise.
20739 * function.c: Likewise.
20740 * gimple-expr.c: Likewise.
20741 * gimple-fold.c: Likewise.
20742 * gimple-pretty-print.c: Likewise.
20743 * gimple.c: Likewise.
20744 * gimplify.c: Likewise.
20745 * hsa-common.c: Likewise.
20746 * hsa-gen.c: Likewise.
20747 * internal-fn.c: Likewise.
20748 * ipa-chkp.c: Likewise.
20749 * ipa-cp.c: Likewise.
20750 * ipa-devirt.c: Likewise.
20751 * ipa-fnsummary.c: Likewise.
20752 * ipa-inline.c: Likewise.
20753 * ipa-visibility.c: Likewise.
20755 * lto-cgraph.c: Likewise.
20756 * omp-expand.c: Likewise.
20757 * omp-general.c: Likewise.
20758 * omp-low.c: Likewise.
20759 * omp-offload.c: Likewise.
20760 * omp-simd-clone.c: Likewise.
20761 * opts-global.c: Likewise.
20762 * passes.c: Likewise.
20763 * predict.c: Likewise.
20764 * sancov.c: Likewise.
20765 * sanopt.c: Likewise.
20766 * symtab.c: Likewise.
20767 * toplev.c: Likewise.
20768 * trans-mem.c: Likewise.
20769 * tree-chkp.c: Likewise.
20770 * tree-eh.c: Likewise.
20771 * tree-into-ssa.c: Likewise.
20772 * tree-object-size.c: Likewise.
20773 * tree-parloops.c: Likewise.
20774 * tree-profile.c: Likewise.
20775 * tree-ssa-ccp.c: Likewise.
20776 * tree-ssa-live.c: Likewise.
20777 * tree-ssa-loop.c: Likewise.
20778 * tree-ssa-sccvn.c: Likewise.
20779 * tree-ssa-structalias.c: Likewise.
20780 * tree-ssa.c: Likewise.
20781 * tree-streamer-in.c: Likewise.
20782 * tree-vectorizer.c: Likewise.
20783 * tree-vrp.c: Likewise.
20784 * tsan.c: Likewise.
20785 * ubsan.c: Likewise.
20786 * varasm.c: Likewise.
20787 * varpool.c: Likewise.
20788 * tree.c: Remove functions moved to attribs.[ch].
20789 * tree.h: Likewise.
20790 * config/aarch64/aarch64.c: Add attrs.h header file.
20791 * config/alpha/alpha.c: Likewise.
20792 * config/arc/arc.c: Likewise.
20793 * config/arm/arm.c: Likewise.
20794 * config/avr/avr.c: Likewise.
20795 * config/bfin/bfin.c: Likewise.
20796 * config/c6x/c6x.c: Likewise.
20797 * config/cr16/cr16.c: Likewise.
20798 * config/cris/cris.c: Likewise.
20799 * config/darwin.c: Likewise.
20800 * config/epiphany/epiphany.c: Likewise.
20801 * config/fr30/fr30.c: Likewise.
20802 * config/frv/frv.c: Likewise.
20803 * config/ft32/ft32.c: Likewise.
20804 * config/h8300/h8300.c: Likewise.
20805 * config/i386/winnt.c: Likewise.
20806 * config/ia64/ia64.c: Likewise.
20807 * config/iq2000/iq2000.c: Likewise.
20808 * config/lm32/lm32.c: Likewise.
20809 * config/m32c/m32c.c: Likewise.
20810 * config/m32r/m32r.c: Likewise.
20811 * config/m68k/m68k.c: Likewise.
20812 * config/mcore/mcore.c: Likewise.
20813 * config/microblaze/microblaze.c: Likewise.
20814 * config/mips/mips.c: Likewise.
20815 * config/mmix/mmix.c: Likewise.
20816 * config/mn10300/mn10300.c: Likewise.
20817 * config/moxie/moxie.c: Likewise.
20818 * config/msp430/msp430.c: Likewise.
20819 * config/nds32/nds32-isr.c: Likewise.
20820 * config/nds32/nds32.c: Likewise.
20821 * config/nios2/nios2.c: Likewise.
20822 * config/nvptx/nvptx.c: Likewise.
20823 * config/pa/pa.c: Likewise.
20824 * config/pdp11/pdp11.c: Likewise.
20825 * config/powerpcspe/powerpcspe.c: Likewise.
20826 * config/riscv/riscv.c: Likewise.
20827 * config/rl78/rl78.c: Likewise.
20828 * config/rx/rx.c: Likewise.
20829 * config/s390/s390.c: Likewise.
20830 * config/sh/sh.c: Likewise.
20831 * config/sol2.c: Likewise.
20832 * config/sparc/sparc.c: Likewise.
20833 * config/spu/spu.c: Likewise.
20834 * config/stormy16/stormy16.c: Likewise.
20835 * config/tilegx/tilegx.c: Likewise.
20836 * config/tilepro/tilepro.c: Likewise.
20837 * config/v850/v850.c: Likewise.
20838 * config/vax/vax.c: Likewise.
20839 * config/visium/visium.c: Likewise.
20840 * config/xtensa/xtensa.c: Likewise.
20842 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20845 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20846 constraints since the -mupper-regs-* switches have been
20848 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20849 into a vector from a double word element that was extracted from
20850 another vector, and eliminate extra XXPERMDI instructions.
20851 (vsx_concat_<mode>_2): Likewise.
20852 (vsx_concat_<mode>_3): Likewise.
20853 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20854 concat to allow optimizing inserts from previous extracts.
20856 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20858 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20859 memory reference to a SSP offset in TLS address space.
20860 (ix86_print_operand) <case '@'>: Remove.
20861 (ix86_print_operand_punct_valid_p): Remove '@' code.
20862 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20863 UNSPEC_SP_TLS_TEST.
20864 (stack_tls_protect_set_<mode>): Remove.
20865 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20866 (stack_tls_protect_test_<mode>): Remove.
20867 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20869 2017-08-07 Olivier Hainque <hainque@adacore.com>
20872 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20874 2017-08-07 Douglas Rupp <rupp@adacore.com>
20876 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20877 variable was referenced as multidir in command.
20879 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20882 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20885 2017-08-07 Martin Liska <mliska@suse.cz>
20887 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20888 * config/rl78/rl78.c: Add include of attribs.h.
20889 * config/sh/sh.c: Likewise.
20890 * config/v850/v850.c: Likewise.
20892 2017-08-07 Tom de Vries <tom@codesourcery.com>
20894 PR middle-end/78266
20895 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20897 2017-08-07 Martin Liska <mliska@suse.cz>
20899 * config/mips/mips.c: Include attribs.h.
20901 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20904 * doc/invoke.texi: Document change in behvaior for -Ofast for
20907 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20909 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20911 (aarch64_pop_regs): Likewise.
20912 (aarch64_gen_load_pair): Likewise.
20913 (aarch64_save_callee_saves): Likewise.
20914 (aarch64_restore_callee_saves): Likewise.
20916 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20918 * config/i386/i386.c: Revert the last change.
20920 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20923 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20925 (ix86_finalize_stack_frame_flags): This. Also clear
20926 frame_pointer_needed if -fno-omit-frame-pointer is used without
20928 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20929 with ix86_finalize_stack_frame_flags.
20930 (ix86_expand_epilogue): Likewise.
20931 (ix86_expand_split_stack_prologue): Likewise.
20933 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20936 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20937 to P_AES for Westmere.
20939 2017-08-07 Jonathan Yong <10walls@gmail.com>
20941 * config/i386/mingw.opt (fset-stack-executable): Removed.
20942 * config/i386/cygming.opt (fset-stack-executable): Moved
20944 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20946 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20948 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20950 2017-08-07 Marek Polacek <polacek@redhat.com>
20952 PR middle-end/81737
20953 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20955 2017-08-07 Martin Liska <mliska@suse.cz>
20957 * attribs.h (canonicalize_attr_name): New function.
20958 (cmp_attribs): Move from c-format.c and adjusted.
20959 (is_attribute_p): Moved from tree.h.
20960 * tree-inline.c: Add new includes.
20961 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20962 (private_is_attribute_p): Remove.
20963 (private_lookup_attribute): Likewise.
20964 (private_lookup_attribute_by_prefix): Simplify.
20965 (remove_attribute): Use is_attribute_p.
20966 * tree.h: Remove removed declarations.
20968 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20970 PR middle-end/81698
20971 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20972 instead of computing it in the function. Formatting fix.
20973 (expand_case): Don't rely on default_edge being the first edge,
20974 clear it if removing it, pass default_edge to
20975 emit_case_dispatch_table.
20976 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20979 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20981 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20982 insn in the function, emit NOP after the insn.
20984 2017-08-06 Tom de Vries <tom@codesourcery.com>
20986 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20989 2017-08-06 Tom de Vries <tom@codesourcery.com>
20991 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20994 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20996 PR tree-optimization/57371
20997 * match.pd: New pattern.
20999 2017-08-04 Marek Polacek <polacek@redhat.com>
21001 PR middle-end/81695
21002 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21003 perform the computation in offset_int.
21005 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21007 PR tree-optimization/81136
21008 * tree-vectorizer.h: Include tree-hash-traits.h.
21009 (vec_base_alignments): New typedef.
21010 (vec_info): Add a base_alignments field.
21011 (vect_record_base_alignments): Declare.
21012 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21014 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21015 (create_data_ref): Add an is_conditional_in_stmt argument.
21016 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21017 the is_conditional_in_stmt field.
21018 (data_ref_loc): Add an is_conditional_in_stmt field.
21019 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21020 (find_data_references_in_stmt): Update call to create_data_ref.
21021 (graphite_find_data_references_in_stmt): Likewise.
21022 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21023 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21024 (vect_record_base_alignment): New function.
21025 (vect_record_base_alignments): Likewise.
21026 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21027 for nested statements even if we fail to compute a misalignment.
21028 Use pooled base alignments for unconditional references.
21029 (vect_find_same_alignment_drs): Compare base addresses instead
21031 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21032 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21034 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21036 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21037 Add an explicit name for the enum. Use auto_vec for slp_instances
21038 and grouped_stores.
21039 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21041 (_bb_vec_info): Add a constructor and destructor.
21042 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21043 (destroy_loop_vec_info): Delete.
21044 (vect_destroy_datarefs): Likewise.
21045 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21046 (vec_info::vec_info): New function.
21047 (vec_info::~vec_info): Likewise.
21048 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21049 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21050 destroy_loop_vec_info.
21051 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21052 (_loop_vec_info::_loop_vec_info): ...this.
21053 (destroy_loop_vec_info): Replace with...
21054 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21055 the stmt_vec_infos. Leave handling of vec_info information to its
21056 destructor. Remove explicit vector releases.
21057 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21058 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21059 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21060 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21061 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21062 (destroy_bb_vec_info): Replace with...
21063 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21064 information to its destructor.
21065 (vect_slp_analyze_bb_1): Use new and delete instead of
21066 new_bb_vec_info and destroy_bb_vec_info.
21067 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21070 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21072 * tree-data-ref.h (subscript): Add access_fn field.
21073 (data_dependence_relation): Add could_be_independent_p.
21074 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21075 (same_access_functions): Move to tree-data-ref.c.
21076 * tree-data-ref.c (ref_contains_union_access_p): New function.
21077 (access_fn_component_p): Likewise.
21078 (access_fn_components_comparable_p): Likewise.
21079 (dr_analyze_indices): Add a reference to access_fn_component_p.
21080 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21082 (constant_access_functions): Likewise.
21083 (add_other_self_distances): Likewise.
21084 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21085 (initialize_data_dependence_relation): Use XCNEW and remove
21086 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21087 of access functions that have the same type. Allow the
21088 subsequence to end with different bases in some circumstances.
21089 Record the chosen access functions in SUB_ACCESS_FN.
21090 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21091 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21092 (subscript_dependence_tester_1): Likewise dra and drb.
21093 (build_classic_dist_vector): Update calls accordingly.
21094 (subscript_dependence_tester): Likewise.
21095 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21096 DDR_COULD_BE_INDEPENDENT_P.
21097 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21098 comp_alias_ddrs instead of may_alias_ddrs.
21099 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21101 (vect_analyze_data_ref_dependence): Use it if
21102 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21103 distance vectors if that fails.
21104 (dependence_distance_ge_vf): New function.
21105 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21106 LOOP_VINFO_MAY_ALIAS_DDRS.
21108 2017-08-04 Richard Biener <rguenther@suse.de>
21110 PR middle-end/81705
21111 * fold-const.c (fold_binary_loc): Properly restrict
21112 minus_var0 && minus_var1 case when associating undefined overflow
21115 2017-08-04 Tom de Vries <tom@codesourcery.com>
21117 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21119 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21121 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21122 Don't start diagnostic messages with a capital letter.
21123 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21125 (rs6000_invalid_builtin): Likewise.
21126 (rs6000_trampoline_init): Likewise.
21128 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21131 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21132 after setting changeable df flags.
21134 2017-08-03 Richard Biener <rguenther@suse.de>
21136 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21137 up if the use is in USE - X.
21139 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21141 * toplev.c (dumpfile.h): New include.
21142 (internal_error_reentered): New static function. Use it...
21143 (internal_error_function): ...here to handle reentered internal_error.
21145 2017-08-03 Richard Biener <rguenther@suse.de>
21147 PR middle-end/81148
21148 * fold-const.c (split_tree): Add minus_var and minus_con
21149 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21150 here but always use minus_*.
21151 (associate_trees): Assert we never associate with MINUS_EXPR
21152 and NULL first operand. Do not recurse for PLUS_EXPR operands
21153 when associating as MINUS_EXPR either.
21154 (fold_binary_loc): Track minus_var and minus_con.
21156 2017-08-03 Tom de Vries <tom@codesourcery.com>
21159 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21160 ACCEL_COMPILER, apply finish_options on
21161 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21163 2017-08-03 Tom de Vries <tom@codesourcery.com>
21166 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21167 function_entry_patch_area_size > 0.
21169 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21172 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21173 instead of 10??LU, perform unit multiplication in wide_int,
21174 don't change alloc_object_size_limit if the limit is larger
21177 PR tree-optimization/81655
21178 PR tree-optimization/81588
21179 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21180 the case when ranges[i].low and high are 1 for unsigned type with
21183 PR middle-end/81052
21184 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21185 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21187 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21189 * tree-vrp.h: Add include guard.
21191 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21194 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21195 (ud2): New insn pattern.
21196 * config/i386/i386.c (ix86_expand_epilogue):
21197 For naked functions, generate ud2 instead of trap insn.
21199 2017-08-02 Marek Polacek <polacek@redhat.com>
21202 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21204 2017-08-02 Tom de Vries <tom@codesourcery.com>
21205 Cesar Philippidis <cesar@codesourcery.com>
21207 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21208 Add missing edge probabilities.
21210 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21212 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21213 Correct endianness.
21215 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21217 PR middle-end/79499
21218 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21219 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21220 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21221 prologue_seq sequences - if any.
21223 2017-08-02 Richard Biener <rguenther@suse.de>
21225 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21226 via vectors if supported, integer extracts via punning if supported
21227 or otherwise vector extracts.
21229 2017-08-02 Richard Biener <rguenther@suse.de>
21231 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21233 (bitmap_insert_into_set): ... this.
21235 2017-08-02 Richard Biener <rguenther@suse.de>
21237 PR tree-optimization/81633
21239 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21241 PR tree-optimization/71752
21242 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21244 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21246 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21247 (machine_function::call_ms2sysv_pad_out): Remove field.
21248 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21249 (ix86_compute_frame_layout): Likewise.
21251 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21254 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21255 attribute with interrupt attribute.
21257 2017-08-01 Andrew Pinski <apinski@cavium.com>
21259 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21260 BIT_INSERT_EXPR's operand 1
21261 to see if the types precision matches.
21263 2017-08-01 Martin Liska <mliska@suse.cz>
21265 PR middle-end/70140
21266 * builtins.c (expand_builtin_memcpy_args): Remove.
21267 (expand_builtin_memcpy): Call newly added function
21268 expand_builtin_memory_copy_args.
21269 (expand_builtin_memcpy_with_bounds): Likewise.
21270 (expand_builtin_mempcpy): Remove last argument.
21271 (expand_builtin_mempcpy_with_bounds): Likewise.
21272 (expand_builtin_memory_copy_args): New function created from
21273 expand_builtin_mempcpy_args with small modifications.
21274 (expand_builtin_mempcpy_args): Remove.
21275 (expand_builtin_stpcpy): Remove unused argument.
21276 (expand_builtin): Likewise.
21277 (expand_builtin_with_bounds): Likewise.
21279 2017-08-01 Martin Liska <mliska@suse.cz>
21282 Make mempcpy more optimal (PR middle-end/70140).
21284 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21287 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21288 __builtin_vec_cmpne verify both arguments are compatible vectors
21289 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21290 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21291 move computation of aligned to after checking the argument types.
21295 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21296 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21299 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21300 Doug Rupp <rupp@adacore.com>
21301 Olivier Hainque <hainque@adacore.com>
21303 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21304 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21306 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21307 for TARGET_OS_CPP_BUILTIN.
21308 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21309 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21311 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21312 passing required abi options to the assembler for EABI configurations.
21313 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21314 of .text.hot and .text.unlikely sections for kernel modules when
21315 using ARM style exceptions.
21316 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21317 options. Add EXTRA_CC1_SPEC.
21318 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21320 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21322 (ARM_TARGET2_DWARF_FORMAT): Define.
21323 * config/arm/t-vxworks: Adjust multilib control to removal of the
21324 Diab command line options.
21326 2017-08-01 Martin Liska <mliska@suse.cz>
21328 PR gcov-profile/81561
21329 * gcov.c (unblock): Make unblocking safe as we need to preserve
21330 index correspondence of blocks and block_lists.
21332 2017-08-01 Richard Biener <rguenther@suse.de>
21334 PR tree-optimization/81181
21335 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21336 (compute_antic): ... end of iteration here.
21338 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21340 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21341 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21342 (ftree-slp-vectorize): Likewise.
21343 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21344 can no longer be set independent of flag_tree_loop_vectorize.
21345 * omp-general.c (emp_max_vf): Likewise.
21346 * opts.c (enable_fdo_optimizations): Remove references to
21347 flag_tree_vectorize, these are now implicit.
21348 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21349 and leave it for the options machinery.
21351 2017-08-01 Martin Liska <mliska@suse.cz>
21353 PR middle-end/70140
21354 * builtins.c (expand_builtin_memcpy_args): Remove.
21355 (expand_builtin_memcpy): Call newly added function
21356 expand_builtin_memory_copy_args.
21357 (expand_builtin_memcpy_with_bounds): Likewise.
21358 (expand_builtin_mempcpy): Remove last argument.
21359 (expand_builtin_mempcpy_with_bounds): Likewise.
21360 (expand_builtin_memory_copy_args): New function created from
21361 expand_builtin_mempcpy_args with small modifications.
21362 (expand_builtin_mempcpy_args): Remove.
21363 (expand_builtin_stpcpy): Remove unused argument.
21364 (expand_builtin): Likewise.
21365 (expand_builtin_with_bounds): Likewise.
21367 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21370 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21371 print "ds:" only for immediates in generic address space.
21373 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21376 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21377 (ix86_function_ok_for_sibcall): Return false for naked functions.
21379 2017-08-01 Richard Biener <rguenther@suse.de>
21381 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21382 (compute_antic): Seed worklist with exit block predecessors.
21383 * cfganal.c (dfs_find_deadend): For a cycle return the source
21384 of the edge closing it.
21386 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21388 * config/aarch64/aarch64.c
21389 (aarch64_can_const_movi_rtx_p): Move 0 check.
21391 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21393 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21394 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21396 * match.pd: Ditto in address comparison pattern.
21398 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21400 PR tree-optimization/81627
21401 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21402 closed ssa form for store-store chain.
21404 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21406 PR tree-optimization/81620
21407 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21408 for store-store chain.
21410 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21412 PR tree-optimization/81588
21413 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21414 ranges[i].in_p, invert comparison code ccode. For >/>=,
21415 swap rhs1 and rhs2 and comparison code unconditionally,
21416 for </<= don't do that. Don't swap rhs1/rhs2 again if
21417 ranges[i].in_p, instead invert comparison code ccode if
21418 opcode or oe->rank is BIT_IOR_EXPR.
21421 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21422 a direct optab to conversion optab.
21423 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21424 with GET_MODE_INNER as last argument instead of optab_handler.
21425 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21426 vector extraction if possible and optab is available.
21427 * expr.c (store_constructor): Use convert_optab_handler instead
21428 of optab_handler. Use vector initialization from smaller
21429 vectors if possible and optab is available.
21430 * tree-vect-stmts.c (vectorizable_load): Likewise.
21431 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21432 now have two modes.
21433 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21434 of vec_init from half-sized vectors with the same element mode.
21435 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21436 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21437 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21438 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21439 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21440 after mode in gen_vec_extract* calls.
21441 (vec_extract<mode>): Renamed to ...
21442 (vec_extract<mode><ssescalarmodelower>): ... this.
21443 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21444 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21445 element mode after mode in gen_vec_init* calls.
21446 (VEC_INIT_HALF_MODE): New mode iterator.
21447 (vec_init<mode>): Renamed to ...
21448 (vec_init<mode><ssescalarmodelower>): ... this.
21449 (vec_init<mode><ssehalfvecmodelower>): New expander.
21450 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21451 (vec_extractv2sfsf): ... this.
21452 (vec_initv2sf): Renamed to ...
21453 (vec_initv2sfsf): ... this.
21454 (vec_extractv2si): Renamed to ...
21455 (vec_extractv2sisi): ... this.
21456 (vec_initv2si): Renamed to ...
21457 (vec_initv2sisi): ... this.
21458 (vec_extractv4hi): Renamed to ...
21459 (vec_extractv4hihi): ... this.
21460 (vec_initv4hi): Renamed to ...
21461 (vec_initv4hihi): ... this.
21462 (vec_extractv8qi): Renamed to ...
21463 (vec_extractv8qiqi): ... this.
21464 (vec_initv8qi): Renamed to ...
21465 (vec_initv8qiqi): ... this.
21466 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21467 (vec_init<mode>): Renamed to ...
21468 (vec_init<mode><VEC_base_l>): ... this.
21469 (vec_extract<mode>): Renamed to ...
21470 (vec_extract<mode><VEC_base_l>): ... this.
21471 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21472 (vec_initv2sfsf): ... this.
21473 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21474 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21475 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21476 element mode after mode in gen_vec_init* calls.
21477 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21478 (vec_init<mode><Vel>): ... this.
21479 (vec_extract<mode>): Renamed to ...
21480 (vec_extract<mode><Vel>): ... this.
21481 * config/aarch64/iterators.md (Vel): New mode attribute.
21482 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21483 Add element mode after mode in gen_vec_extract* calls.
21484 * config/s390/vector.md (non_vec_l): New mode attribute.
21485 (vec_extract<mode>): Renamed to ...
21486 (vec_extract<mode><non_vec_l>): ... this.
21487 (vec_init<mode>): Renamed to ...
21488 (vec_init<mode><non_vec_l>): ... this.
21489 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21490 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21492 * config/arm/iterators.md (V_elem_l): New mode attribute.
21493 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21494 (vec_extract<mode><V_elem_l>): ... this.
21495 (vec_extractv2di): Renamed to ...
21496 (vec_extractv2didi): ... this.
21497 (vec_init<mode>): Renamed to ...
21498 (vec_init<mode><V_elem_l>): ... this.
21499 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21500 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21501 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21502 Add element mode after gen_vec_extract* calls.
21503 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21504 (vec_init<mode><unitmode>): ... this.
21505 (vec_extract<mode>): Renamed to ...
21506 (vec_extract<mode><unitmode>): ... this.
21507 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21508 (vec_init<mode><unitmode>): ... this.
21509 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21510 (vec_initv2sfsf): ... this.
21511 (vec_extractv2sf): Renamed to ...
21512 (vec_extractv2sfsf): ... this.
21513 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21514 Add element mode after gen_vec_extract* calls.
21515 * config/mips/mips.md (unitmode): New mode iterator.
21516 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21517 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21518 * config/spu/spu.md (inner_l): New mode attribute.
21519 (vec_init<mode>): Renamed to ...
21520 (vec_init<mode><inner_l>): ... this.
21521 (vec_extract<mode>): Renamed to ...
21522 (vec_extract<mode><inner_l>): ... this.
21523 * config/sparc/sparc.md (veltmode): New mode iterator.
21524 (vec_init<VMALL:mode>): Renamed to ...
21525 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21526 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21527 (vec_initv2sisi): ... this.
21528 (vec_initv2sf): Renamed to ...
21529 (vec_initv2sfsf): ... this.
21530 (vec_extractv2sf): Renamed to ...
21531 (vec_extractv2sfsf): ... this.
21532 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21533 (vec_init<mode>): Renamed to ...
21534 (vec_init<mode><VEC_base_l>): ... this.
21535 (vec_extract<mode>): Renamed to ...
21536 (vec_extract<mode><VEC_base_l>): ... this.
21537 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21538 (vec_initv2sfsf): ... this.
21539 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21540 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21541 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21542 gen_vec_init* calls.
21544 2017-08-01 Richard Biener <rguenther@suse.de>
21546 PR tree-optimization/81297
21547 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21548 TREE_OVERFLOW from INTEGER_CSTs.
21550 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21552 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21554 2017-07-31 Carl Love <cel@us.ibm.com>
21556 * config/rs6000/rs6000-c: Add support for built-in functions
21557 vector signed char vec_xl_be (signed long long, signed char *);
21558 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21559 vector signed int vec_xl_be (signed long long, signed int *);
21560 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21561 vector signed long long vec_xl_be (signed long long, signed long long *);
21562 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21563 vector signed short vec_xl_be (signed long long, signed short *);
21564 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21565 vector double vec_xl_be (signed long long, double *);
21566 vector float vec_xl_be (signed long long, float *);
21567 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21568 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21569 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21571 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21572 (altivec_expand_builtin): Add switch statement to call
21573 altivec_expand_xl_be for each builtin.
21574 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21575 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21576 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21577 __builtin_vsx_le_be_v16qi.
21578 * doc/extend.texi: Update the built-in documentation file for the
21579 new built-in functions.
21581 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21584 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21586 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21588 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21590 * config.gcc: Add z14.
21591 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21592 CPU model numbers for z13s and z14.
21593 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21595 * config/s390/s390-opts.h (enum processor_type): Rename
21596 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21597 * config/s390/s390.c (processor_table): Add field for CPU name to
21598 be passed to Binutils.
21599 (s390_asm_output_machine_for_arch): Use the new field in
21600 processor_table for Binutils.
21601 (s390_expand_builtin): Replace arch12 with z14.
21602 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21603 (s390_get_sched_attrmask): Likewise.
21604 (s390_get_unit_mask): Likewise.
21605 * config/s390/s390.opt: Add z14 to processor_type enum.
21607 2017-07-31 Martin Jambor <mjambor@suse.cz>
21610 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21611 regardless of optimization level.
21613 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21614 Martin Liska <mliska@suse.cz>
21616 * predict.def: Remove old comment and adjust probability.
21617 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21618 PREDICT statements.
21620 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21623 * config/i386/i386.c (ix86_function_naked): New function.
21624 (ix86_can_use_return_insn_p): Return false for naked functions.
21625 (ix86_expand_prologue): Skip prologue for naked functions.
21626 (ix86_expand_epilogue): Skip epilogue for naked functions
21627 and emit trap instruction.
21628 (ix86_warn_func_return): New function.
21629 (ix86_attribute_table): Add "naked" attribute specification.
21630 (TARGET_WARN_FUNC_RETURN): Define.
21631 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21633 2017-07-31 Martin Liska <mliska@suse.cz>
21635 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21636 (dump_gimple_bb_header): Always dump BB info.
21637 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21639 2017-07-31 Martin Liska <mliska@suse.cz>
21642 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21643 also with current_function_decl non-null equality.
21645 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21648 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21649 change type to the element type, instead add eltype variable and
21650 use it where we are interested in the element type.
21652 PR tree-optimization/81603
21653 * ipa-polymorphic-call.c
21654 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21655 offset arithmetic in offset_int, bail out if the resulting bit offset
21656 doesn't fit into shwi.
21658 2017-07-31 Martin Liska <mliska@suse.cz>
21660 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21661 (gimplify_save_expr): Fix comment.
21663 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21666 * config/i386/i386.c (ix86_function_arg): Update arguments for
21668 (ix86_compute_frame_layout): Set the initial stack offset to
21669 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21670 INCOMING_FRAME_SP_OFFSET.
21671 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21672 stack before exception handler returns.
21673 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21674 the 'ERROR_CODE' for exception handler.
21676 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21678 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21679 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21680 (ASM_OUTPUT_REG_POP): Ditto.
21681 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21682 instead of asm_fprintf to output pure string.
21684 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21686 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21687 to imported_module_or_decl hook.
21688 (debug_nothing_tree_tree_tree_bool): Remove.
21689 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21690 * debug.c (do_nothing_debug_hooks): Use
21691 debug_nothing_tree_tree_tree_bool_bool instead of
21692 debug_nothing_tree_tree_tree_bool.
21693 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21694 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21695 * sdbout.c (sdb_debug_hooks): Likewise.
21696 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21697 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21699 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21700 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21701 attribute, don't add anything.
21703 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21705 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21706 (fold_build2_stat_loc): Likewise.
21707 (fold_build3_stat_loc): Likewise.
21708 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21709 (fold_build1_loc): Remove macro.
21710 (fold_build2_loc): Likewise.
21711 (fold_build3_loc): Likewise.
21713 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21715 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21716 (gimple_build_debug_bind_source_stat): Likewise.
21717 * gimple.h (gimple_build_debug_bind): Remove macro.
21718 (gimple_build_debug_bind_source): Likewise.
21720 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21722 * bitmap.c (bitmap_alloc): Adjust.
21723 (bitmap_gc_alloc): Likewise.
21724 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21726 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21728 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21729 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21730 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21731 (bitmap_gc_alloc_stat): Likewise.
21732 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21734 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21736 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21737 * rtl.h (shallow_copy_rtx): Remove macro.
21739 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21741 * emit-rtl.c (gen_raw_REG): Adjust.
21742 * gengenrtl.c (gendef): Likewise.
21743 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21744 * rtl.h (rtx_alloc): Remove macro.
21746 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21748 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21749 (build_tree_list_stat): Likewise.
21750 * tree.h (build_tree_list): Remove macro.
21751 (build_tree_list_vec): Likewise.
21753 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21755 * tree.c (make_vector_stat): Remove _stat from name.
21756 (build_vector_stat): Likewise.
21757 * tree.h (make_vector_stat): Remove macro.
21758 (build_vector_stat): Likewise.
21760 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21762 * tree.h (build_var_debug_value): Remove prototype.
21764 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21766 * tree.c (tree_cons_stat): Remove _stat from name.
21767 * tree.h (tree_cons): Remove macro.
21769 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21771 * tree.c (build_vl_exp_stat): Remove _stat from name.
21772 * tree.h (build_vl_exp): Remove macro.
21774 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21776 * tree.c (build_decl_stat): Remove _stat from name.
21777 * tree.h (build_decl): Remove macro.
21779 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21781 * gimple.c (gimple_build_with_ops_stat): Adjust.
21782 (gimple_alloc_stat): Remove _stat from name.
21783 * gimple.h (gimple_alloc): Remove macro.
21785 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21787 * tree.c (make_tree_vec_stat): Remove _stat from name.
21788 (grow_tree_vec_stat): Likewise.
21789 * tree.h (make_tree_vec_stat): Adjust prototype.
21790 (grow_tree_vec_stat): Likewise.
21791 (make_tree_vec): Remove macro.
21792 (grow_tree_vec): Likewise.
21794 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21796 * fold-const.c (fold_build1_stat_loc): Adjust.
21797 (fold_build2_stat_loc): Likewise.
21798 (fold_build3_stat_loc): Likewise.
21799 * tree.c (build0_stat): Remove _stat from name.
21800 (build1_stat): Likewise.
21801 (build2_stat): Likewise.
21802 (build3_stat): Likewise.
21803 (build4_stat): Likewise.
21804 (build5_stat): Likewise.
21805 * tree.h (build1_loc): Remove macro, and rename _stat function
21807 (build2_loc): Likewise.
21808 (build3_loc): Likewise.
21809 (build4_loc): Likewise.
21810 (build5_loc): Likewise.
21812 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21814 * tree.c (make_int_cst_stat): Remove _stat from name.
21815 * tree.h (make_int_cst_stat): Adjust prototype.
21816 (make_int_cst): Remove macro.
21818 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21820 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21821 * tree.h (make_tree_binfo_stat): Adjust prototype.
21822 (make_tree_binfo): Remove.
21824 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21826 * tree.c (copy_node_stat): Rename to copy_node.
21827 (build_distinct_type_copy): Adjust.
21828 * tree.h (copy_node_stat): Adjust prototype.
21829 (copy_node): Remove macro.
21831 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21833 * tree.c (make_node_stat): rename to make_node.
21834 (build_tree_list_stat): Adjust.
21835 (build0_stat): Likewise.
21836 (build2_stat): Likewise.
21837 (build3_stat): Likewise.
21838 (build4_stat): Likewise.
21839 (build5_stat): Likewise.
21840 (build_decl_stat): Likewise.
21841 * tree.h (make_node_stat): Adjust prototype.
21842 (make_node): remove macro.
21844 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21846 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21847 (PPC_FEATURE2_SCV): Likewise.
21848 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21850 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21852 * config/aarch64/aarch64.c
21853 (aarch64_internal_mov_immediate): Add new special pattern.
21854 * config/aarch64/aarch64.md (*movdi_aarch64):
21855 Add reg/32bit const mov case.
21857 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21858 Richard Sandiford <richard.sandiford@linaro.org>
21860 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21861 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21862 Add integer and movi cases.
21863 (movi-split-hf-df-sf split, fp16): New.
21864 (enabled): Added TARGET_FP_F16INST.
21865 * config/aarch64/iterators.md (GPF_HF): New.
21866 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21868 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21870 * config/aarch64/aarch64.c
21871 (aarch64_simd_container_mode): Add prototype.
21872 (aarch64_expand_mov_immediate): Add HI support.
21873 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21874 (aarch64_can_const_movi_rtx_p): New.
21875 (aarch64_preferred_reload_class):
21876 Remove restrictions of using FP registers for certain SIMD operations.
21877 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21878 (aarch64_valid_floating_const): Add integer move validation.
21879 (aarch64_simd_imm_scalar_p): Remove.
21880 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21881 (aarch64_legitimate_constant_p): Expand list of supported cases.
21882 * config/aarch64/aarch64-protos.h
21883 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21884 (aarch64_reinterpret_float_as_int): New.
21885 (aarch64_simd_imm_scalar_p): Remove.
21886 * config/aarch64/constraints.md (Uvi): New.
21887 (Dd): Split into Ds and new Dd.
21888 * config/aarch64/aarch64.md (*movsi_aarch64):
21890 (*movdi_aarch64): Add SIMD mov case.
21892 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21894 * tree-predcom.c: (struct chain): Handle store-store chain in which
21895 stores for elimination only store loop invariant values.
21896 (execute_pred_commoning_chain): Ditto.
21897 (prepare_initializers_chain_store_elim): Ditto.
21898 (prepare_finalizers): Ditto.
21899 (is_inv_store_elimination_chain): New function.
21900 (initialize_root_vars_store_elim_1): New function.
21902 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21904 * tree-predcom.c: Revise general description of the pass.
21905 (enum chain_type): New enum type for store elimination.
21906 (struct chain): New field supporting store elimination.
21907 (struct component): Ditto.
21908 (dump_chain): Dump store-stores chain.
21909 (release_chain): Release resources.
21910 (split_data_refs_to_components): Compute and create component
21911 contains only stores for elimination.
21912 (get_chain_last_ref_at): New function.
21913 (make_invariant_chain): Initialization.
21914 (make_rooted_chain): Specify chain type in parameter and record it.
21915 (add_looparound_copies): Skip for store-stores chain.
21916 (determine_roots_comp): Compute type of chain and pass it to
21918 (initialize_root_vars_store_elim_2): New function.
21919 (finalize_eliminated_stores): New function.
21920 (remove_stmt): Handle store for elimination.
21921 (execute_pred_commoning_chain): Execute predictive commoning on
21922 store-store chains.
21923 (determine_unroll_factor): Skip unroll for store-stores chain.
21924 (prepare_initializers_chain_store_elim): New function.
21925 (prepare_initializers_chain): Hanlde store-store chain.
21926 (prepare_finalizers_chain, prepare_finalizers): New function.
21927 (tree_predictive_commoning_loop): Return integer value indicating
21928 if loop is unrolled or lcssa form is corrupted.
21929 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21931 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21933 * tree-predcom.c (initialize_root): Delete.
21934 (execute_pred_commoning_chain): Initialize root vars and replace
21935 reference of non-combined chain directly, rather than call above
21938 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21940 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21941 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21943 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21945 * tree-predcom.c (struct chain): New field init_seq.
21946 (release_chain): Release init_seq.
21947 (prepare_initializers_chain): Record intialization stmts in above
21949 (insert_init_seqs): New function.
21950 (tree_predictive_commoning_loop): Call insert_init_seqs.
21952 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21954 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21956 2017-07-28 Richard Biener <rguenther@suse.de>
21958 * match.pd: Remove superfluous :c.
21959 * genmatch.c (simplify::id): Add member.
21960 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21962 (current_id): New global.
21963 (dt_node::parent): Move from ...
21964 (dt_operand::parent): ... here. Add for_id member.
21965 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21966 (decision_tree::find_node): Relax order requirement when
21967 merging DT_TRUE nodes to ones inbetween the current simplify
21968 and the one we try to merge with. Add diagnostic whenever
21969 we need to enforce pattern order by not merging.
21970 (decision_tree::insert): Set current_id.
21971 (decision_tree::print_node): Dump parent node and for_id.
21972 (parser::last_id): Add member.
21973 (parser::push_simplify): Assign unique id.
21974 (parser::parser): Initialize last_id.
21976 2017-07-28 Martin Liska <mliska@suse.cz>
21979 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21980 gimple_build_debug_bind.
21982 2017-07-28 Richard Biener <rguenther@suse.de>
21984 PR tree-optimization/81502
21985 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21987 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21989 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21990 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21991 for BIT_FIELD_REF args.
21992 * fold-const.c (make_bit_field_ref): Likewise.
21993 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21995 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21998 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21999 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22000 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22001 Or it into SANITIZER_UNDEFINED.
22002 * ubsan.c: Include gimple-fold.h and varasm.h.
22003 (ubsan_expand_ptr_ifn): New function.
22004 (instrument_pointer_overflow): New function.
22005 (maybe_instrument_pointer_overflow): New function.
22006 (instrument_object_size): Formatting fix.
22007 (pass_ubsan::execute): Call instrument_pointer_overflow
22008 and maybe_instrument_pointer_overflow.
22009 * internal-fn.c (expand_UBSAN_PTR): New function.
22010 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22011 * sanitizer.def (__ubsan_handle_pointer_overflow,
22012 __ubsan_handle_pointer_overflow_abort): New builtins.
22013 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22014 * internal-fn.def (UBSAN_PTR): New internal function.
22015 * opts.c (sanitizer_opts): Add pointer-overflow.
22016 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22017 * fold-const.c (build_range_check): Compute pointer range check in
22018 integral type if pointer arithmetics would be needed. Formatting
22021 2017-07-28 Martin Liska <mliska@suse.cz>
22024 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22025 parameters that are of a variable-length.
22027 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22029 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22031 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22032 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22033 (CRT_CALL_STATIC_FUNCTION): Likewise.
22034 (ASM_DEFAULT_SPEC): New define.
22035 (ASM_SPEC32): Likewise.
22036 (ASM_SPEC64): Likewise.
22037 (ASM_SPEC_COMMON): Likewise.
22038 (ASM_SPEC): Likewise.
22039 (INVALID_64BIT): Likewise.
22040 (LINK_OS_DEFAULT_SPEC): Likewise.
22041 (LINK_OS_SPEC32): Likewise.
22042 (LINK_OS_SPEC64): Likewise.
22043 (POWERPC_LINUX): Likewise.
22044 (PTRDIFF_TYPE): Likewise.
22045 (RESTORE_FP_PREFIX): Likewise.
22046 (RESTORE_FP_SUFFIX): Likewise.
22047 (SAVE_FP_PREFIX): Likewise.
22048 (SAVE_FP_SUFFIX): Likewise.
22049 (SIZE_TYPE): Likewise.
22050 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22051 (TARGET_64BIT): Likewise.
22052 (TARGET_64BIT): Likewise.
22053 (TARGET_AIX): Likewise.
22054 (WCHAR_TYPE_SIZE): Likewise.
22055 (WCHAR_TYPE): Undefine.
22056 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22057 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22058 (CPP_OS_RTEMS_SPEC): Delete.
22059 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22060 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22062 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22064 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22066 PR tree-optimization/81578
22067 * tree-parloops.c (build_new_reduction): Bail out if
22068 reduction_code isn't one of the standard OpenMP reductions.
22069 Move the details printing after that decision.
22071 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22073 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22074 related to reload_in_progress.
22075 (splat_input_operand): Likewise.
22076 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22078 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22080 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22081 (TARGET_INSTANTIATE_DECLS): Likewise.
22082 (legitimate_indexed_address_p): Delete reload_in_progress code.
22083 (rs6000_debug_legitimate_address_p): Likewise.
22084 (rs6000_eliminate_indexed_memrefs): Likewise.
22085 (rs6000_emit_le_vsx_store): Likewise.
22086 (rs6000_emit_move_si_sf_subreg): Likewise.
22087 (rs6000_emit_move): Likewise.
22088 (register_to_reg_type): Likewise.
22089 (rs6000_pre_atomic_barrier): Likewise.
22090 (rs6000_machopic_legitimize_pic_address): Likewise.
22091 (rs6000_allocate_stack_temp): Likewise.
22092 (rs6000_address_for_fpconvert): Likewise.
22093 (rs6000_address_for_altivec): Likewise.
22094 (rs6000_secondary_memory_needed_rtx): Delete function.
22095 (rs6000_check_sdmode): Likewise.
22096 (rs6000_alloc_sdmode_stack_slot): Likewise.
22097 (rs6000_instantiate_decls): Likewise.
22098 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22099 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22100 Delete reload_in_progress.
22101 (*vec_reload_and_plus_<mptrsize>): Likewise.
22102 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22103 (vsx_div_v2di): Likewise.
22104 (vsx_udiv_v2di): Likewise.
22106 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22108 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22109 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22110 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22111 (rs6000_debug_reg_global): Delete print of LRA status.
22112 (rs6000_option_override_internal): Delete dead LRA related code.
22113 (rs6000_lra_p): Delete function.
22114 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22116 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22118 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22119 * config/riscv/rtems.h: New file.
22121 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22122 Sudakshina Das <sudi.das@arm.com>
22124 * config/aarch64/aarch64.md
22125 (define_split for and<mode>3nr_compare): Move
22126 non aarch64_logical_operand to a register.
22127 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22128 register immediate operand to a register.
22129 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22131 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22133 PR middle-end/81564
22134 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22136 2017-07-27 Richard Biener <rguenther@suse.de>
22138 PR tree-optimization/81573
22139 PR tree-optimization/81494
22140 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22141 multi defuse cycle case.
22143 2017-07-27 Richard Biener <rguenther@suse.de>
22145 PR tree-optimization/81571
22146 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22149 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22151 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22152 earlier and only if MASK_FPU is set. Adjust formatting.
22154 2017-07-27 Martin Liska <mliska@suse.cz>
22156 * opt-functions.awk: Add validation of value of Init.
22157 * optc-gen.awk: Pass new argument.
22159 2017-07-27 Martin Liska <mliska@suse.cz>
22161 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22162 Fix wrong condition.
22164 2017-07-27 Martin Liska <mliska@suse.cz>
22166 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22167 BBs and edges seen by autoFDO.
22169 2017-07-27 Richard Biener <rguenther@suse.de>
22171 PR tree-optimization/81502
22172 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22173 with incompatible but same sized type.
22174 (execute_update_addresses_taken): Likewise.
22176 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22178 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22179 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22181 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22184 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22185 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22186 Change s_operand to memory_operand.
22188 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22190 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22191 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22192 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22193 Emit instructions rather than returning an expression. Handle TFmode
22194 and KFmode by casting to TImode.
22195 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22196 (rs6000_emit_le_vsx_store): Likewise.
22197 * config/rs6000/vsx.md (VSX_TI): New iterator.
22198 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22199 (*vsx_le_undo_permute_<mode>): Likewise.
22200 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22201 emit the split sequence.
22202 (*vsx_le_perm_store_<mode>): Likewise.
22204 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22206 PR tree-optimization/81555
22207 PR tree-optimization/81556
22208 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22209 if true, force CHANGED for the recursive invocation.
22210 (reassociate_bb): Remember original length of ops array, pass
22211 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22213 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22214 attributes for noipa attribute. For naked attribute use
22215 lookup_attribute first before lookup_attribute_spec.
22216 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22218 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22219 for functions with noipa attribute.
22220 (cgraph_externally_visible_p): Return true for functions with noipa
22222 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22223 for functions with noipa attribute.
22224 * doc/extend.texi: Document noipa function attribute.
22225 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22226 also for functions with noipa attribute.
22227 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22229 2017-07-26 Andrew Pinski <apinski@cavium.com>
22231 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22232 vec_unalign_load_cost and vec_unalign_store_cost.
22234 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22236 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22237 -mvsx-small-integer option.
22238 (ISA_3_0_MASKS_IEEE): Likewise.
22239 (OTHER_VSX_VECTOR_MASKS): Likewise.
22240 (POWERPC_MASKS): Likewise.
22241 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22242 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22243 code, only testing for DImode being allowed in non-VSX floating
22245 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22246 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22248 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22249 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22250 TARGET_P8_VECTOR test.
22251 (rs6000_secondary_reload_simple_move): Likewise.
22252 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22253 since TARGET_P9_VECTOR was already tested.
22254 (rs6000_opt_masks): Remove -mvsx-small-integer.
22255 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22256 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22258 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22259 test for TARGET_VEXTRACTUB was used, and that uses
22261 (p9 extract splitter): Likewise.
22262 (vsx_extract_<mode>_di_p9): Likewise.
22263 (vsx_extract_<mode>_store_p9): Likewise.
22264 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22265 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22266 the elimination of TARGET_VSX_SMALL_INTEGER.
22267 (vsx_extract_<mode>_p8): Likewise.
22268 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22269 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22270 (vsx_set_<mode>_p9): Likewise.
22271 (vsx_set_v4sf_p9): Likewise.
22272 (vsx_set_v4sf_p9_zero): Likewise.
22273 (vsx_insert_extract_v4sf_p9): Likewise.
22274 (vsx_insert_extract_v4sf_p9_2): Likewise.
22275 * config/rs6000/rs6000.md (sign extend splitter): Change
22276 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22277 (floatsi<mode>2_lfiwax_mem): Likewise.
22278 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22279 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22280 since a test for TARGET_P9_VECTOR was used.
22281 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22282 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22283 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22284 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22285 TARGET_P8_VECTOR test.
22286 (fix_trunc<mode>si2_stfiwx): Likewise.
22287 (fix_trunc<mode>si2_internal): Likewise.
22288 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22289 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22291 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22292 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22293 TARGET_P8_VECTOR test.
22294 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22295 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22296 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22298 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22299 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22300 since a test for TARGET_P9_VECTOR was used.
22301 (splitter for loading small constants): Likewise.
22303 2017-07-26 Andrew Pinski <apinski@cavium.com>
22305 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22308 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22311 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22312 (fp_valid_at): Likewise.
22314 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22316 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22317 (qdf24xx_addrcost_table): Likewise.
22318 (cortexa57_tunings): Update to use generic_branch_cost.
22319 (cortexa72_tunings): Likewise.
22320 (cortexa73_tunings): Likewise.
22321 (qdf24xx_tunings): Likewise.
22323 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22325 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22326 (thunderx2t99_branch_cost): Likewise.
22327 (cortexa35_tunings): Update to use generic_branch_cost.
22328 (cortexa53_tunings): Likewise.
22329 (cortexa57_tunings): Likewise.
22330 (cortexa72_tunings): Likewise.
22331 (cortexa73_tunings): Likewise.
22332 (thunderx2t99_tunings): Likewise.
22334 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22336 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22337 (sparc_option_override): Honour MASK_FSMULD.
22338 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22339 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22340 * config/sparc/sparc.opt (mfsmuld): New option.
22341 * doc/invoke.texi (mfsmuld): Document option.
22343 2017-07-26 Marek Polacek <polacek@redhat.com>
22345 PR middle-end/70992
22346 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22348 2017-07-26 Richard Biener <rguenther@suse.de>
22350 * gimple-match-head.c (do_valueize): Return OP if valueize
22352 (get_def): New helper to get at the def stmt of a SSA name
22353 if valueize allows.
22354 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22355 do_valueize to get at the def stmt.
22356 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22358 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22360 PR middle-end/46932
22361 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22363 2017-07-26 Martin Liska <mliska@suse.cz>
22366 * function.c (expand_function_start): Make expansion of
22367 nonlocal_goto_save_area after parm_birth_insn.
22369 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22371 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22372 from all CPU target flags enable members.
22374 2017-07-26 Richard Biener <rguenther@suse.de>
22376 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22377 (decision_tree::gen): Make 'type' const.
22378 (write_predicate): Likewise.
22380 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22382 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22383 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22384 (rs6000_option_override_internal): Likewise.
22385 (rs6000_expand_vector_set): Likewise.
22386 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22387 (TARGET_UPPER_REGS_SF): Likewise.
22388 (TARGET_UPPER_REGS_DI): Likewise.
22389 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22390 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22391 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22392 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22393 (Splitters for DI constants in Altivec registers): Likewise.
22394 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22395 (vsx_set_v4sf_p9): Likewise.
22396 (vsx_set_v4sf_p9_zero): Likewise.
22397 (vsx_insert_extract_v4sf_p9): Likewise.
22398 (vsx_insert_extract_v4sf_p9_2): Likewise.
22400 2017-07-25 Carl Love <cel@us.ibm.com>
22402 * doc/extend.texi: Update the built-in documentation file for the
22403 existing built-in functions
22404 vector signed char vec_cnttz (vector signed char);
22405 vector unsigned char vec_cnttz (vector unsigned char);
22406 vector signed short vec_cnttz (vector signed short);
22407 vector unsigned short vec_cnttz (vector unsigned short);
22408 vector signed int vec_cnttz (vector signed int);
22409 vector unsigned int vec_cnttz (vector unsigned int);
22410 vector signed long long vec_cnttz (vector signed long long);
22411 vector unsigned long long vec_cnttz (vector unsigned long long);
22413 2017-07-25 Andrew Pinski <apinski@cavium.com>
22415 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22416 accesses where the use is for the first operand of a BIT_INSERT.
22418 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22421 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22422 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22424 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22426 * config/i386/gstabs.h: Delete.
22427 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22429 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22431 * config/i386/i386.c (ix86_decompose_address): Do not check for
22432 register RTX when looking at index_reg or base_reg.
22433 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22435 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22437 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22438 to update EH info here.
22440 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22442 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22444 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22446 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22448 2017-07-25 Torsten Duwe <duwe@suse.de>
22450 * common.opt: Introduce -fpatchable-function-entry
22451 command line option, and its variables function_entry_patch_area_size
22452 and function_entry_patch_area_start.
22453 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22454 including a two-value parser.
22455 * target.def (print_patchable_function_entry): New target hook.
22456 * targhooks.h (default_print_patchable_function_entry): New function.
22457 * targhooks.c (default_print_patchable_function_entry): Likewise.
22458 * toplev.c (process_options): Switch off IPA-RA if
22459 patchable function entries are being generated.
22460 * varasm.c (assemble_start_function): Look at the
22461 patchable-function-entry command line switch and current
22462 function attributes and maybe generate NOP instructions by
22463 calling the print_patchable_function_entry hook.
22464 * doc/extend.texi: Document patchable_function_entry attribute.
22465 * doc/invoke.texi: Document -fpatchable_function_entry
22466 command line option.
22467 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22469 * doc/tm.texi: Re-generate.
22471 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22474 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22475 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22477 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22479 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22481 2017-07-25 Richard Biener <rguenther@suse.de>
22483 PR tree-optimization/81455
22484 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22485 not walk in cycles when looking for guards.
22487 2017-07-25 Richard Biener <rguenther@suse.de>
22489 PR tree-optimization/81529
22490 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22491 when optimizing backedge uses.
22493 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22495 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22497 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22498 to dl_section_ref. On AIX, append an expression to subtract
22499 the size of the section length to dl_section_ref.
22501 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22503 * configure.ac: If any of the config.* scripts fail, exit 1.
22504 * configure: Regenerate.
22506 2017-07-25 Richard Biener <rguenther@suse.de>
22508 PR middle-end/81546
22509 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22510 of immediate uses, be more verbose on errors.
22512 2017-07-25 Richard Biener <rguenther@suse.de>
22514 PR tree-optimization/81510
22515 * tree-vect-loop.c (vect_is_simple_reduction): When the
22516 reduction stmt is not inside the loop bail out.
22518 2017-07-25 Richard Biener <rguenther@suse.de>
22520 PR tree-optimization/81303
22521 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22522 profitability check against LOOP_VINFO_NITERSM1.
22524 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22526 * domwalk.c (cmp_bb_postorder): Simplify.
22527 (sort_bbs_postorder): New function. Use it...
22528 (dom_walker::walk): ...here to optimize common cases.
22530 2017-07-25 Martin Liska <mliska@suse.cz>
22533 * ipa-visibility.c (function_and_variable_visibility): Make the
22534 redirection just on target that supports aliasing.
22535 Fix GNU coding style.
22537 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22540 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22542 * config/arm/rtems.h: Likewise.
22543 * config/bfin/rtems.h: Likewise.
22544 * config/i386/rtemself.h: Likewise.
22545 * config/lm32/rtems.h: Likewise.
22546 * config/m32c/rtems.h: Likewise.
22547 * config/m68k/rtemself.h: Likewise.
22548 * config/microblaze/rtems.h: Likewise.
22549 * config/mips/rtems.h: Likewise.
22550 * config/moxie/rtems.h: Likewise.
22551 * config/nios2/rtems.h: Likewise.
22552 * config/powerpcspe/rtems.h: Likewise.
22553 * config/rs6000/rtems.h: Likewise.
22554 * config/rtems.h: Likewise.
22555 * config/sh/rtems.h: Likewise.
22556 * config/sh/rtemself.h: Likewise.
22557 * config/sparc/rtemself.h: Likewise.
22559 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22562 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22563 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22564 * tree-ssa-structalias.c (alias_get_name): Same.
22566 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22569 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22570 instructions if no du chain is found.
22572 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22574 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22576 2017-07-25 Richard Biener <rguenther@suse.de>
22578 PR middle-end/81505
22579 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22582 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22584 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22585 upper-regs options.
22586 (ISA_2_7_MASKS_SERVER): Likewise.
22587 (ISA_3_0_MASKS_IEEE): Likewise.
22588 (OTHER_P8_VECTOR_MASKS): Likewise.
22589 (OTHER_VSX_VECTOR_MASKS): Likewise.
22590 (POWERPC_MASKS): Likewise.
22591 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22592 duplicate list of options.
22593 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22594 explicit -mupper-regs options.
22595 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22596 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22597 alias for -mupper-regs-df.
22598 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22599 (rs6000_init_hard_regno_mode_ok): Likewise.
22600 (rs6000_option_override_internal): Likewise.
22601 (rs6000_opt_masks): Likewise.
22602 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22603 options in terms of whether -mvsx or -mpower8-vector was used.
22604 (TARGET_UPPER_REGS_DI): Likewise.
22605 (TARGET_UPPER_REGS_SF): Likewise.
22606 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22607 -mupper-regs-* options.
22609 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22611 * passes.c (emergency_dump_function): Print some empty lines and a
22612 header before the RTL dump.
22614 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22616 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22618 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22621 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22622 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22624 2017-07-24 Carl Love <cel@us.ibm.com>
22626 * config/rs6000/rs6000-c.c: Add support for built-in functions
22627 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22628 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22629 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22630 vec_extract_fp_from_shortl): Add defines for the two builtins.
22631 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22632 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22634 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22635 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22636 * doc/extend.texi: Update the built-in documentation file for the
22637 new built-in function.
22639 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22642 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22644 * doc/generic.texi: Likewise.
22645 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22646 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22648 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22650 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22651 (aarch64_mls_elt_merge<mode>): Likewise.
22653 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22655 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22656 having __cxa_atexit.
22658 2017-07-23 Michael Collison <michael.collison@arm.com>
22660 * config/arm/arm.c (arm_option_override): Deprecate
22661 use of -mstructure-size-boundary.
22662 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22663 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22665 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22668 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22669 Reduce cost estimate for direct moves.
22671 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22674 * config/i386/i386.c (ix86_option_override_internal): Disable
22675 BMI, BMI2 and TBM instructions for -m16.
22677 2017-07-21 Carl Love <cel@us.ibm.com>
22679 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22680 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22681 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22682 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22683 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22684 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22685 VMULOSW): New enum "unspec" values.
22686 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22687 altivec_vmulosw): New patterns.
22688 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22689 VMULOSW): Add definitions.
22691 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22693 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22694 (qdf24xx): Likewise.
22695 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22696 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22697 (AARCH64_FL_V8_1): Renumber.
22698 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22699 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22700 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22701 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22702 rdma to feature modifiers list.
22704 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22706 PR middle-end/56727
22707 * ipa-visibility (function_and_variable_visibility): Convert
22708 recursive PLT call to direct call if appropriate.
22710 2017-07-21 Andrew Pinski <apinski@cavium.com>
22712 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22713 operand 1 to see if the types precision matches.
22714 * fold-const.c (operand_equal_p): Likewise.
22716 2017-07-21 Richard Biener <rguenther@suse.de>
22718 PR tree-optimization/81303
22719 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22720 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22721 (vect_peeling_hash_get_lowest_cost): Adjust.
22722 (vect_enhance_data_refs_alignment): Likewise. Use
22723 vect_get_peeling_costs_all_drs to compute the penalty for no
22724 peeling to match up costs.
22726 2017-07-21 Richard Biener <rguenther@suse.de>
22728 PR tree-optimization/81500
22729 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22730 we didn't identify a reduction path.
22732 2017-07-21 Tom de Vries <tom@codesourcery.com>
22733 Cesar Philippidis <cesar@codesourcery.com>
22735 PR gcov-profile/81442
22736 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22739 2017-07-21 Tom de Vries <tom@codesourcery.com>
22742 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22744 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22745 nvptx_override_options_after_change.
22747 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22749 * dwarf2out.c (output_file_names): Avoid double testing for
22750 dwarf_version >= 5.
22752 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22754 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22756 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22758 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22760 (try_crossjump_to_edge): Do not punt on partitioned functions.
22762 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22764 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22765 Put all BBs reachable only via paths crossing cold region to cold
22767 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22769 2016-07-21 Richard Biener <rguenther@suse.de>
22771 PR tree-optimization/81303
22772 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22773 into account prologue and epilogue iterations when raising
22774 min_profitable_iters to sth at least covering one vector iteration.
22776 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22778 * config/arm/arm.c (arm_test_cpu_arch_dat):
22781 2017-07-20 Nathan Sidwell <nathan@acm.org>
22783 Remove TYPE_METHODS.
22784 * tree.h (TYPE_METHODS): Delete.
22785 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22786 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22787 (dbxout_type_methods): Scan TYPE_FIELDS.
22788 (dbxout_type): Don't check TYPE_METHODS here.
22789 * function.c (use_register_for_decl): Always ignore register for
22790 class types when not optimizing.
22791 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22792 * tree.c (free_lang_data_in_type): Stitch out member functions and
22793 templates from TYPE_FIELDS.
22794 (build_distinct_type_copy, verify_type_variant,
22795 verify_type): Member fns are on TYPE_FIELDS.
22796 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22797 * tree-pretty-print.c (dump_generic_node): Likewise.
22799 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22802 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22803 V2TImode and V4TImode.
22804 (ix86_expand_vector_extract): Likewise.
22805 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22806 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22807 (ssescalarmode): Handle V4TImode and V2TImode.
22808 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22809 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22810 (VEXTRACTI128_MODE): New mode iterator.
22811 (splitter for *vec_extractv?ti first element): New.
22812 (VEC_INIT_MODE): New mode iterator.
22813 (vec_init<mode>): Consolidate 3 expanders into one using
22814 VEC_INIT_MODE mode iterator.
22816 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22818 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22819 non_spilled_static_chain_regno_p.
22821 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22823 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22825 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22827 * bb-reorder.c (connect_traces): Allow copying of blocks within
22830 2017-07-20 Richard Biener <rguenther@suse.de>
22832 * gimple.h (gimple_phi_result): Add gphi * overload.
22833 (gimple_phi_result_ptr): Likewise.
22834 (gimple_phi_arg): Likewise. Adjust index assert to only
22835 allow actual argument accesses rather than all slots available
22837 (gimple_phi_arg_def): Add gphi * overload.
22838 * tree-phinodes.c (make_phi_node): Initialize only actual
22840 (resize_phi_node): Clear memory not covered by old node,
22841 do not initialize excess argument slots.
22842 (reserve_phi_args_for_new_edge): Initialize new argument slot
22845 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22847 PR tree-optimization/81388
22849 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22851 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22852 by removing computation of may_be_zero.
22854 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22855 Tom de Vries <tom@codesourcery.com>
22857 PR middle-end/81030
22858 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22859 when gimple level profile disagrees with what RTL expander did.
22861 2017-07-20 Richard Biener <rguenther@suse.de>
22863 PR tree-optimization/61171
22864 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22865 (vect_analyze_stmt): Add slp instance parameter.
22866 (vectorizable_reduction): Likewise.
22867 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22868 (vect_is_simple_reduction): Deal with chains not detected
22869 as SLP reduction chain, specifically not properly associated
22870 chains containing a mix of plus/minus.
22871 (get_reduction_op): Remove.
22872 (get_initial_defs_for_reduction): Simplify, pass in whether
22873 this is a reduction chain, pass in the SLP node for the PHIs.
22874 (vect_create_epilog_for_reduction): Get the SLP instance as
22876 (vectorizable_reduction): Get the SLP instance as arg.
22877 During analysis remember the SLP node with the PHIs in the
22878 instance. Simplify getting at the vectorized reduction PHIs.
22879 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22880 through SLP instance.
22881 (vect_slp_analyze_operations): Likewise.
22882 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22883 (vect_transform_stmt): Likewise.
22885 2017-07-20 Tom de Vries <tom@codesourcery.com>
22887 PR tree-optimization/81489
22888 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22889 read of phi arg location to before loop that modifies phi.
22891 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22893 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22896 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22898 PR middle-end/81331
22899 * except.c (execute): Fix ordering issue.
22901 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22903 PR rtl-optimization/81423
22904 * combine.c (make_compound_operation_int): Don't try to optimize
22905 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22907 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22909 PR rtl-optimization/81423
22910 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22911 with a constant that is -1 in the truncated to mode.
22913 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22915 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22916 (determine_unlikely_bbs): ... here.
22917 * predict.h (propagate_unlikely_bbs_forward): Declare.
22918 * cfgexpand.c (pass_expand::execute): Use it.
22919 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22921 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22922 propagate_unlikely_bbs_forward.
22924 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22926 PR middle-end/81331
22927 * except.c (maybe_add_nop_after_section_switch): New function.
22930 2017-07-19 Tom de Vries <tom@codesourcery.com>
22932 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22934 2017-07-19 Tom de Vries <tom@codesourcery.com>
22936 * gimple.h (gimple_phi_arg): Make assert more strict.
22938 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22940 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22941 * config/rs6000/mmintrin.h: New file.
22942 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22944 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22946 PR tree-optimization/81346
22947 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22949 2017-07-19 Tom de Vries <tom@codesourcery.com>
22951 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22953 2017-07-19 Tom de Vries <tom@codesourcery.com>
22955 * config/nvptx/nvptx-modes.def: Add V2DImode.
22956 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22957 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22958 (nvptx_output_mov_insn): Handle lack of mov.b128.
22959 (nvptx_print_operand): Handle 'H' and 'L' codes.
22960 (nvptx_vector_mode_supported): Allow V2DImode.
22961 (nvptx_preferred_simd_mode): New function.
22962 (nvptx_data_alignment): New function.
22963 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22964 nvptx_preferred_simd_mode.
22965 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22967 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22969 2017-07-19 Tom de Vries <tom@codesourcery.com>
22971 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22972 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22973 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22974 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22975 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22976 (mov<VECIM>_insn): New define_insn.
22977 (define_expand "mov<VECIM>): New define_expand.
22979 2017-07-19 Tom de Vries <tom@codesourcery.com>
22981 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22983 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22985 PR tree-optimization/81346
22986 * fold-const.h (fold_div_compare, range_check_type): Declare.
22987 * fold-const.c (range_check_type): New function.
22988 (build_range_check): Use range_check_type.
22989 (fold_div_compare): No longer static, rewritten into
22990 a match.pd helper function.
22991 (fold_comparison): Don't call fold_div_compare here.
22992 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22993 as helper function.
22995 2017-07-19 Nathan Sidwell <nathan@acm.org>
22997 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22998 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22999 * tree.c (find_decls_types_r, verify_type): Use
23000 TYPE_{MIN,MAX}_VALUE_RAW.
23001 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23002 (hash_tree): Likewise.
23003 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23005 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23008 2017-07-18 Tom de Vries <tom@codesourcery.com>
23010 PR middle-end/81464
23011 * omp-expand.c (expand_omp_for_static_chunk): Handle
23012 equal-argument loop exit phi.
23014 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23017 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23018 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23019 operand 2 predicate.
23020 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23021 operand 2 predicate.
23022 (ror,rol -> rorx splitters): Use const_int_operand as
23023 operand 2 predicate.
23025 2017-06-18 Richard Biener <rguenther@suse.de>
23027 PR tree-optimization/81410
23028 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23029 the gap in the ! slp_perm SLP case after each group.
23031 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23033 PR middle-end/81463
23034 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23037 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23039 PR middle-end/81462
23040 * predict.c (set_even_probabilities): Cleanup; do not affect
23041 probabilities that are already known.
23042 (combine_predictions_for_bb): Call even when count is set.
23044 2017-07-18 Nathan Sidwell <nathan@acm.org>
23046 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23049 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23052 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23053 optimization for loop niter analysis.
23055 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23058 * config/avr/avr.c (avr_optimize_casesi): Don't use
23059 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23061 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23063 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23064 body_cost_vec from _vect_peel_extended_info.
23065 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23066 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23069 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23071 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23073 2017-07-18 Richard Biener <rguenther@suse.de>
23075 PR tree-optimization/80620
23076 PR tree-optimization/81403
23077 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23078 info when re-using a VN table entry.
23080 2017-07-18 Richard Biener <rguenther@suse.de>
23082 PR tree-optimization/81418
23083 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23084 vectype_in. Verify that with lane-reducing reduction operations
23085 we have a single def-use cycle.
23087 2017-07-17 Carl Love <cel@us.ibm.com>
23089 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23091 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23092 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23093 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23094 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23095 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23096 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23097 VMULOSW): New enum "unspec" values.
23098 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23099 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23100 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23101 altivec_vmulosw): New patterns.
23102 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23103 VMULOSW): Add definitions.
23105 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23107 * config/alpha/alpha.c: Include predict.h.
23109 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23111 * tree-vrp.c (compare_assert_loc): Fix comparison function
23112 to return predictable results.
23114 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23116 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23118 (subdi3): Likewise.
23119 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23120 * doc/invoke.texi (mexpand-adddi): Update text.
23122 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23124 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23125 that also clobbers the CC register. The old expand code is moved
23127 (*arc_clzsi2): ... here.
23128 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23129 the CC register. The old expand code is moved to ...
23130 (arc_ctzsi2): ... here.
23132 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23134 * config/arc/arc.opt (mindexed-loads): Use initial value
23135 TARGET_INDEXED_LOADS_DEFAULT.
23136 (mauto-modify-reg): Use initial value
23137 TARGET_AUTO_MODIFY_REG_DEFAULT.
23138 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23139 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23140 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23141 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23143 2017-07-17 Martin Liska <mliska@suse.cz>
23146 * opts.c (finish_options): Do not allow -fgnu-tm
23147 w/ -fsanitize={kernel-,}address. Say sorry.
23149 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23152 * tree-loop-distribution.c (classify_partition): Only assert on
23153 numer of iterations.
23154 (merge_dep_scc_partitions): Delete prameter. Update function call.
23155 (distribute_loop): Remove code handling loop with unknown niters.
23156 (pass_loop_distribution::execute): Skip loop with unknown niters.
23158 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23161 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23162 function sort_partitions_by_post_order.
23164 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23166 PR tree-optimization/81374
23167 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23168 the max index of basic blocks, rather than number of basic blocks.
23170 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23172 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23174 (arc_legitimate_pic_operand_p): Likewise.
23175 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23177 (arc_needs_pcl_p): Likewise.
23178 (arc_legitimate_pc_offset_p): Likewise.
23179 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23180 function is also used in constrains.md.
23181 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23182 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23183 PLUS. Only return true/false in known cases, otherwise assert.
23184 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23185 is already called in arc_legitimate_constant_p.
23186 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23188 (LEGITIMATE_PIC_OPERAND_P): Use
23189 arc_raw_symbolic_reference_mentioned_p function.
23190 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23195 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23196 Andrew Burgess <andrew.burgess@embecosm.com>
23198 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23199 (arc_return_address_register): New function.
23200 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23201 (arc_handle_fndecl_attribute): Add naked attribute.
23202 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23203 (TARGET_WARN_FUNC_RETURN): Likewise.
23204 (arc_allocate_stack_slots_for_args): New function.
23205 (arc_warn_func_return): Likewise.
23206 (machine_function): Change type fn_type.
23207 (arc_compute_function_type): Consider new naked function type,
23208 change function return type.
23209 (arc_must_save_register): Adapt to handle new
23210 arc_compute_function_type's return type.
23211 (arc_expand_prologue): Likewise.
23212 (arc_expand_epilogue): Likewise.
23213 (arc_return_address_regs): Delete.
23214 (arc_return_address_register): New function.
23215 (arc_epilogue_uses): Use above function.
23216 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23217 (arc_function_type): Change encoding, add naked type.
23218 (ARC_INTERRUPT_P): Change to handle the new encoding.
23219 (ARC_FAST_INTERRUPT_P): Likewise.
23220 (ARC_NORMAL_P): Define.
23221 (ARC_NAKED_P): Likewise.
23222 (arc_compute_function_type): Delete prototype.
23223 * config/arc/arc.md (in_ret_delay_slot): Use
23224 arc_return_address_register function.
23225 (simple_return): Likewise.
23226 (p_return_i): Likewise.
23228 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23230 PR tree-optimization/81428
23231 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23232 can't be built for those types.
23234 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23236 Remove stuff dead since r239246.
23238 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23239 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23240 (avr_inform_devices): Remove dead stuff.
23242 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23244 * config/arm/arm_neon.h: Fix softp typo.
23246 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23248 PR tree-optimization/81365
23249 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23250 aggregate moves onto bb predecessor edges, make sure there are no
23251 loads that could alias the lhs in between the start of bb and the
23254 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23257 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23258 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23259 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23261 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23263 PR tree-optimization/81396
23264 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23265 (init_symbolic_number): Initialize it to 1.
23266 (perform_symbolic_merge): Add n_ops from both operands into the new
23268 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23269 without base_addr as useless if they need more than one operation.
23270 (bswap_replace): Handle !bswap case for NULL base_addr.
23272 2017-07-17 Tom de Vries <tom@codesourcery.com>
23275 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23278 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23280 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23281 conditional builtin define __FIX_LEON3FT_B2BST.
23283 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23285 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23286 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23289 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23291 PR rtl-optimization/81424
23292 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23293 to remove potential trapping from operands if -fnon-call-exceptions.
23295 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23297 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23298 profile_proability for scalling.
23299 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23301 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23303 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23305 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23307 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23308 fixpoint arithmetics.
23310 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23312 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23313 fixpoint arithmetics.
23315 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23317 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23318 fixpoint arithmetics.
23320 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23322 * profile-count.h (profile_probability::from_reg_br_prob_note,
23323 profile_probability::to_reg_br_prob_note): New functions.
23324 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23325 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23326 * predict.c (probability_reliable_p): Update.
23327 (edge_probability_reliable_p): Update.
23328 (br_prob_note_reliable_p): Update.
23329 (invert_br_probabilities): Update.
23330 (add_reg_br_prob_note): New function.
23331 (combine_predictions_for_insn): Update.
23332 * asan.c (asan_clear_shadow): Update.
23333 * cfgbuild.c (compute_outgoing_frequencies): Update.
23334 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23335 (update_br_prob_note): Update.
23336 (rtl_verify_edges): Update.
23337 (purge_dead_edges): Update.
23338 (fixup_reorder_chain): Update.
23339 * emit-rtl.c (try_split): Update.
23340 * ifcvt.c (cond_exec_process_insns): Update.
23341 (cond_exec_process_if_block): Update.
23342 (dead_or_predicable): Update.
23343 * internal-fn.c (expand_addsub_overflow): Update.
23344 (expand_neg_overflow): Update.
23345 (expand_mul_overflow): Update.
23346 * loop-doloop.c (doloop_modify): Update.
23347 * loop-unroll.c (compare_and_jump_seq): Update.
23348 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23349 * predict.h: Update.
23350 * reorg.c (mostly_true_jump): Update.
23352 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23353 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23354 * config/arc/arc.c: (emit_unlikely_jump): Update.
23355 * config/arm/arm.c: (emit_unlikely_jump): Update.
23356 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23357 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23358 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23359 (ix86_print_operand): Update.
23360 (ix86_split_fp_branch): Update.
23361 (predict_jump): Update.
23362 * config/ia64/ia64.c (ia64_print_operand): Update.
23363 * config/mmix/mmix.c (mmix_print_operand): Update.
23364 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23365 (rs6000_expand_split_stack_prologue): Update.
23366 * config/rs6000/rs6000.c: Update.
23367 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23368 (s390_expand_vec_movstr): Update.
23369 (s390_expand_cs_tdsi): Update.
23370 (s390_expand_split_stack_prologue): Update.
23371 * config/sh/sh.c (sh_print_operand): Update.
23372 (expand_cbranchsi4): Update.
23373 (expand_cbranchdi4): Update.
23374 * config/sparc/sparc.c (output_v9branch): Update.
23375 * config/spu/spu.c (get_branch_target): Update.
23376 (ea_load_store_inline): Update.
23377 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23378 * config/tilepro/tilepro.c: Update.
23380 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23382 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23383 (gimplify_save_expr): Likewise.
23385 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23387 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23389 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23391 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23393 * timevar.def (TV_IPA_FNSUMMARY): Define.
23395 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23397 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23398 to back store errata sensitive sequence from being generated.
23399 (sqrtdf2_fix): Likewise.
23401 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23403 * tree-ssa-threadupdate.c (compute_path_counts,
23404 update_joiner_offpath_counts): Use profile_probability.
23406 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23409 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23411 * config/arm/arm-c.c (arm_cpu_builtins): Define
23412 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23414 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23416 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23417 array entries to represent __ieee128 versions of the
23418 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23419 scalar_extract_sig, and scalar_insert_exp built-in functions.
23420 (altivec_resolve_overloaded_builtin): Add special case handling
23421 for the __builtin_scalar_insert_exp function, as represented by
23422 the P9V_BUILTIN_VEC_VSIEDP constant.
23423 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23424 exponent support for __ieee128 argument.
23425 (VSESQP): Add scalar extract signature support for __ieee128
23427 (VSTDCNQP): Add scalar test negative support for __ieee128
23429 (VSIEQP): Add scalar insert exponent support for __int128 argument
23430 with __ieee128 result.
23431 (VSIEQPF): Add scalar insert exponent support for __ieee128
23432 argument with __ieee128 result.
23433 (VSTDCQP): Add scalar test data class support for __ieee128
23435 (VSTDCNQP): Add overload support for scalar test negative with
23436 __ieee128 argument.
23437 (VSTDCQP): Add overload support for scalar test data class
23438 __ieee128 argument.
23439 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23440 UNSPEC_VSX_SXSIGDP.
23441 (UNSPEC_VSX_SIEXPQP): New constant.
23442 (xsxexpqp): New insn for VSX scalar extract exponent quad
23444 (xsxsigqp): New insn for VSX scalar extract significand quad
23446 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23447 precision with floating point argument.
23448 (xststdcqp): New expand for VSX scalar test data class quad
23450 (xststdcnegqp): New expand for VSX scalar test negative quad
23452 (xststdcqp): New insn to match expansions for VSX scalar test data
23453 class quad precision and VSX scalar test negative quad precision.
23454 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23455 special case operand checking to enforce that second operand of
23456 VSX scalar test data class with quad precision argument is a 7-bit
23458 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23459 prototypes and descriptions of __ieee128 versions of
23460 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23461 scalar_test_data_class, and scalar_test_neg built-in functions.
23463 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23465 PR tree-optimization/81162
23466 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23467 replace a negate with an add.
23469 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23471 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23473 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23475 * config/arm/arm-c.c (arm_cpu_builtins): Define
23476 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23478 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23480 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23481 (armv8-r): Set ARM Cortex-R52 as default CPU.
23482 * config/arm/arm-tables.opt: Regenerate.
23483 * config/arm/arm-tune.md: Regenerate.
23484 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23486 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23487 extension for -mcpu=cortex-r52.
23489 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23491 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23492 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23493 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23494 (fp-armv8): Define it as FP_ARMv8 only.
23495 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23496 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23498 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23499 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23500 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23501 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23502 than TARGET_FPU_ARMV8.
23503 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23504 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23505 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23507 * config/arm/neon.md (neon_vrint): Likewise.
23508 (neon_vcvt): Likewise.
23509 (neon_<fmaxmin_op><mode>): Likewise.
23510 (<fmaxmin><mode>3): Likewise.
23511 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23512 * config/arm/predicates.md (arm_cond_move_operator): Check against
23513 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23515 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23517 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23518 to top of function.
23520 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23522 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23523 loop in comment with memset.
23525 2017-07-14 Martin Liska <mliska@suse.cz>
23527 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23528 * dwarf2out.c (is_java): Remove the function.
23529 (output_pubname): Remove usage of the function.
23530 (lower_bound_default): Remove usage of DW_LANG_Java.
23531 (gen_compile_unit_die): Likewise.
23532 * gcc.c: Remove compiler defaults for .java and .zip files.
23533 * gimple-expr.c (remove_suffix): Change as there's no longer
23534 extension than 4-letter one.
23535 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23536 (gimplify_save_expr): Likewise.
23537 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23538 as it's possible even for other languages than Java.
23539 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23540 * lto-opts.c (lto_write_options): Remove reference to Java.
23541 * opts.c (strip_off_ending): Update file extension handling.
23542 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23543 * tree-eh.c (lower_resx): Likewise.
23544 * tree.c (free_lang_data_in_type): Remove dead code.
23545 (find_decls_types_r): Likewise.
23546 (build_common_builtin_nodes): Remove Java from a comment.
23547 (verify_type): Remove dead code.
23548 * varasm.c (assemble_external): Remove Java from a comment.
23550 2017-07-14 Martin Liska <mliska@suse.cz>
23552 * opts.c (finish_options): Add quotes.
23553 (common_handle_option): Likewise.
23555 2017-07-14 Martin Liska <mliska@suse.cz>
23557 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23558 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23559 Remove N_SO_PASCAL.
23560 * dwarf2out.c (lower_bound_default): Do not handle
23562 (gen_compile_unit_die): Likewise.
23563 * gcc.c: Remove default extension binding for GNU Pascal.
23564 * stmt.c: Remove Pascal language from a comment.
23565 * xcoffout.c: Likewise.
23567 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23570 * diagnostic-show-locus.c (fixit_cmp): New function.
23571 (layout::layout): Sort m_fixit_hints.
23572 (column_range::column_range): Assert that the values are valid.
23573 (struct char_span): New struct.
23574 (correction::overwrite): New method.
23575 (struct source_line): New struct.
23576 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23577 calls in terms of classes source_line and char_span, and
23578 correction::overwrite.
23579 (selftest::test_overlapped_fixit_printing_2): New function.
23580 (selftest::diagnostic_show_locus_c_tests): Call it.
23582 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23584 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23585 early if there is no lhs.
23587 2017-07-13 Martin Liska <mliska@suse.cz>
23589 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23590 (gen_reference_type_die): Likewise.
23591 * stor-layout.c: Remove Pascal-related comment.
23593 2017-07-13 Martin Liska <mliska@suse.cz>
23595 * opts.c (finish_options): Add quotes to error messages.
23596 (parse_sanitizer_options): Likewise.
23598 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23600 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23602 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23604 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23606 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23608 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23610 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23612 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23615 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23616 provides the hardware capability bits, define the macro
23617 __BUILTIN_CPU_SUPPORTS__.
23618 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23619 if GLIBC does not provide the hardware capability bits. Add a
23620 gcc_unreachable call if the built-in cpu function is neither
23621 __builtin_cpu_is nor __builtin_cpu_supports.
23622 (rs6000_get_function_versions_dispatcher): Change the warning
23623 that an old GLIBC is used which does not export the capability
23624 bits to be an error.
23625 * doc/extend.texi (target_clones attribute): Document the
23626 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23627 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23628 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23629 the macros defined by GCC if the newer GLIBC is available.
23631 2017-07-12 Jeff Law <law@redhat.com>
23633 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23634 remaining includes slightly.
23635 * config/riscv/riscv-builtins.c: Include profile-count.h.
23637 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23640 * config/avr/avr.c (avr_set_current_function): In diagnostic
23641 messages: Quote keywords and (parts of) identifiers.
23642 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23645 2017-07-12 Carl Love <cel@us.ibm.com>
23647 * config/rs6000/rs6000-c.c: Add support for built-in functions
23648 vector bool char vec_revb (vector bool char);
23649 vector bool short vec_revb (vector short char);
23650 vector bool int vec_revb (vector bool int);
23651 vector bool long long vec_revb (vector bool long long);
23652 * doc/extend.texi: Update the built-in documentation file for the
23653 new built-in functions.
23655 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23657 * config/s390/s390.md: Remove movcc splitter.
23659 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23661 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23662 load/store on condition.
23664 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23667 * config/avr/avr.c (avr_encode_section_info)
23668 [progmem && !TREE_READONLY]: Error if progmem object needs
23671 2017-07-11 Michael Collison <michael.collison@arm.com>
23673 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23676 2017-07-11 Carl Love <cel@us.ibm.com>
23678 * config/rs6000/rs6000-c.c: Add support for builtins
23679 vector unsigned int vec_parity_lsbb (vector signed int);
23680 vector unsigned int vec_parity_lsbb (vector unsigned int);
23681 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23682 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23683 vector unsigned long long vec_parity_lsbb (vector signed long long);
23684 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23685 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23686 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23687 * doc/extend.texi: Update the built-in documentation file for the
23688 new built-in functions.
23690 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23692 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23693 (layout::m_primary_loc): New field.
23694 (layout::layout): Initialize new field. Move location filtering
23695 logic from here to...
23696 (layout::maybe_add_location_range): ...this new method. Add
23697 support for filtering to just the lines already specified by other
23699 (layout::will_show_line_p): New method.
23700 (gcc_rich_location::add_location_if_nearby): New method.
23701 (selftest::test_add_location_if_nearby): New test function.
23702 (selftest::diagnostic_show_locus_c_tests): Call it.
23703 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23706 2017-07-11 Tom de Vries <tom@codesourcery.com>
23708 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23709 (bb_first_real_insn): New function.
23710 (nvptx_single): Add extra initialization of broadcasted condition
23713 2017-07-11 Nathan Sidwell <nathan@acm.org>
23715 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23717 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23719 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23720 Binutils doc as TEXI will mess them up.
23721 * doc/invoke.texi (AVR Options): Same here.
23723 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23725 * config/sparc/sparc.opt (mfix-ut700): New option.
23726 (mfix-gr712rc): Likewise.
23727 (sparc_fix_b2bst): New variable.
23728 * doc/invoke.texi (SPARC options): Document them.
23729 (ARM options): Fix warnings.
23730 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23731 instructions to prevent sequences that can trigger the store-store
23732 errata for certain LEON3FT processors.
23733 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23734 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23735 * config/sparc/sparc.md (fix_b2bst): New attribute.
23736 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23738 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23741 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23743 (*rsqrtsf2_sse): Ditto.
23745 (div<mode>3): Macroize insn from divdf3 and divsf3
23746 using MODEF mode iterator.
23748 2017-07-10 Martin Sebor <msebor@redhat.com>
23750 PR tree-optimization/80397
23751 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23752 instead of testing for equality to INTEGER_TYPE.
23754 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23756 * config.gcc: Remove uclibc from arc target spec.
23758 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23760 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23762 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23765 * lto-wrapper.c (remove_option): New function.
23766 (merge_and_complain): Merge PIC/PIE options more realistically.
23768 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23770 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23774 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23775 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23776 * config.in: Regenerate.
23777 * configure: Regenerate.
23778 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23779 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23780 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23781 (TARGET_GASISR_PROLOGUES): ...target mask.
23782 * common/config/avr/avr-common.c
23783 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23784 Set -mgas-isr-prologues.
23785 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23786 INSERT_PASS_BEFORE for it.
23787 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23788 * config/avr/avr.c (avr_option_override)
23789 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23790 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23791 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23792 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23793 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23795 (make_avr_pass_pre_proep): New function.
23796 (emit_push_sfr) <treg>: Add argument to function and use it
23797 instead of TMP_REG.
23798 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23799 and set machine->gasisr.yes.
23800 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23801 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23802 __gcc_isr.n_pushed to .L__stack_usage.
23803 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23804 (avr_asm_final_postscan_insn): ...this new static function.
23805 * config/avr/avr.h (machine_function)
23806 <is_no_gccisr, use_L__stack_usage>: New fields.
23807 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23808 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23809 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23810 (gasisr, *gasisr): New expander and insn.
23811 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23812 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23813 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23815 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23817 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23820 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23822 Move jump-tables out of .text again.
23825 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23826 (ASM_OUTPUT_ADDR_VEC): New function.
23827 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23828 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23829 INSN_ADDRESSes as asm comment.
23830 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23831 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23832 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23833 * config/avr/avr.md (*tablejump): Adjust comment.
23834 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23835 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23837 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23838 (avr_output_addr_vec): New proto.
23839 (avr_log_t) <insn_addresses>: New field.
23841 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23844 * config/i386/i386.c (ix86_function_arg_advance): Set
23845 outgoing_args_on_stack to true if there are outgoing arguments
23847 (ix86_function_arg): Likewise.
23848 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23849 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23850 * config/i386/i386.h (machine_function): Add
23851 outgoing_args_on_stack.
23853 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23855 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23856 supporting pthreds.
23857 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23859 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23861 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23862 (REAL_H): Remove $(MACHMODE_H).
23863 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23865 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23866 $(MACHMODE_H) and double-int.h.
23867 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23869 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23872 2017-07-07 Andrew Pinski <apinski@cavium.com>
23874 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23875 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23877 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23879 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23880 Add warning if GCC was not configured to link against a GLIBC that
23881 exports the hardware capability bits.
23882 (make_resolver_func): Make resolver function private and not a
23883 COMDAT function. Create the name with clone_function_name instead
23884 of make_unique_name.
23887 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23888 correct operand in doing the split.
23890 2017-07-07 Carl Love <cel@us.ibm.com>
23892 * config/rs6000/rs6000-c: Add support for built-in function
23893 vector unsigned short vec_pack_to_short_fp32 (vector float,
23895 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23896 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23897 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23898 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23899 (convert_4f32_8i16): Add define_expand.
23900 * doc/extend.texi: Update the built-in documentation file for the
23901 new built-in function.
23903 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23905 * config/sparc/m8.md: New file.
23906 * config/sparc/sparc.md: Include m8.md.
23908 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23910 * config/sparc/sparc.opt: New option -mvis4b.
23911 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23912 (sparc_option_override): Handle VIS4B.
23913 (enum sparc_builtins): Define
23914 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23915 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23916 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23917 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23918 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23919 (check_constant_argument): New function.
23920 (sparc_vis_init_builtins): Define builtins
23921 __builtin_vis_dictunpack{8,16,32},
23922 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23923 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23924 __builtin_vis_fpcmpde{8,16,32}shl and
23925 __builtin_vis_fpcmpur{8,16,32}shl.
23926 (sparc_expand_builtin): Check that the constant operands to
23927 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23928 constant and in range.
23929 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23931 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23932 (SPARC_IMM5_P): Likewise.
23933 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23934 (enabled): Handle vis4b.
23935 (UNSPEC_DICTUNPACK): New unspec.
23936 (UNSPEC_FPCMPSHL): Likewise.
23937 (UNSPEC_FPUCMPSHL): Likewise.
23938 (UNSPEC_FPCMPDESHL): Likewise.
23939 (UNSPEC_FPCMPURSHL): Likewise.
23940 (cpu_feature): New CPU feature `vis4b'.
23941 (dictunpack{8,16,32}): New insns.
23942 (FPCSMODE): New mode iterator.
23943 (fpcscond): New code iterator.
23944 (fpcsucond): Likewise.
23945 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23946 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23947 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23948 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23949 * config/sparc/constraints.md: Define constraints `q' for unsigned
23950 2-bit integer constants and `t' for unsigned 5-bit integer
23952 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23954 (imm5_operand_dictunpack16): Likewise.
23955 (imm5_operand_dictunpack32): Likewise.
23956 (imm2_operand): Likewise.
23957 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23958 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23959 ditunpack* and fpcmp*shl builtins.
23961 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23963 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23964 * config.in: Add HAVE_AS_SPARC6 define.
23965 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23967 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23969 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23970 (CPP_CPU_SPEC): Handle m8.
23971 (ASM_CPU_SPEC): Likewise.
23972 * config/sparc/sparc-opts.h (enum processor_type): Add
23974 * config/sparc/sparc.c (m8_costs): New struct.
23975 (sparc_option_override): Handle TARGET_CPU_m8.
23976 (sparc32_initialize_trampoline): Likewise.
23977 (sparc64_initialize_trampoline): Likewise.
23978 (sparc_issue_rate): Likewise.
23979 (sparc_register_move_cost): Likewise.
23980 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23981 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23982 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23983 (CPP_CPU_SPEC): Handle M8.
23984 (ASM_CPU_SPEC): Likewise.
23985 (AS_M8_FLAG): Define.
23986 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23987 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23988 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23990 * configure: Regenerate.
23991 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23994 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23996 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23998 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23999 ("*movdi_insn_sp32"): Do not set v3pipe.
24000 ("*movsi_insn"): Likewise.
24001 ("*movdi_insn_sp64"): Likewise.
24002 ("*movsf_insn"): Likewise.
24003 ("*movdf_insn_sp32"): Likewise.
24004 ("*movdf_insn_sp64"): Likewise.
24005 ("*zero_extendsidi2_insn_sp64"): Likewise.
24006 ("*sign_extendsidi2_insn"): Likewise.
24007 ("*mov<VM32:mode>_insn"): Likewise.
24008 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24009 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24010 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24011 ("<vlop:code><VL:mode>3"): Likewise.
24012 ("*not_<vlop:code><VL:mode>3"): Likewise.
24013 ("*nand<VL:mode>_vis"): Likewise.
24014 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24015 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24016 ("one_cmpl<VL:mode>2"): Likewise.
24017 ("faligndata<VM64:mode>_vis"): Likewise.
24018 ("alignaddrsi_vis"): Likewise.
24019 ("alignaddrdi_vis"): Likweise.
24020 ("alignaddrlsi_vis"): Likewise.
24021 ("alignaddrldi_vis"): Likewise.
24022 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24023 ("bmaskdi_vis"): Likewise.
24024 ("bmasksi_vis"): Likewise.
24025 ("bshuffle<VM64:mode>_vis"): Likewise.
24026 ("cmask8<P:mode>_vis"): Likewise.
24027 ("cmask16<P:mode>_vis"): Likewise.
24028 ("cmask32<P:mode>_vis"): Likewise.
24029 ("pdistn<P:mode>_vis"): Likewise.
24030 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24032 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24034 * config/sparc/sparc.md ("subtype"): New insn attribute.
24035 ("*wrgsr_sp64"): Set insn subtype.
24036 ("*rdgsr_sp64"): Likewise.
24037 ("alignaddrsi_vis"): Likewise.
24038 ("alignaddrdi_vis"): Likewise.
24039 ("alignaddrlsi_vis"): Likewise.
24040 ("alignaddrldi_vis"): Likewise.
24041 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24042 ("fexpand_vis"): Likewise.
24043 ("fpmerge_vis"): Likewise.
24044 ("faligndata<VM64:mode>_vis"): Likewise.
24045 ("bshuffle<VM64:mode>_vis"): Likewise.
24046 ("cmask8<P:mode>_vis"): Likewise.
24047 ("cmask16<P:mode>_vis"): Likewise.
24048 ("cmask32<P:mode>_vis"): Likewise.
24049 ("fchksm16_vis"): Likewise.
24050 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24051 ("fmean16_vis"): Likewise.
24052 ("fp<plusminus_insn>64_vis"): Likewise.
24053 ("<plusminus_insn>v8qi3"): Likewise.
24054 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24055 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24056 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24057 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24058 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24059 ("*movqi_insn"): Likewise.
24060 ("*movhi_insn"): Likewise.
24061 ("*movsi_insn"): Likewise.
24062 ("movsi_pic_gotdata_op"): Likewise.
24063 ("*movdi_insn_sp32"): Likewise.
24064 ("*movdi_insn_sp64"): Likewise.
24065 ("movdi_pic_gotdata_op"): Likewise.
24066 ("*movsf_insn"): Likewise.
24067 ("*movdf_insn_sp32"): Likewise.
24068 ("*movdf_insn_sp64"): Likewise.
24069 ("*zero_extendhisi2_insn"): Likewise.
24070 ("*zero_extendqihi2_insn"): Likewise.
24071 ("*zero_extendqisi2_insn"): Likewise.
24072 ("*zero_extendqidi2_insn"): Likewise.
24073 ("*zero_extendhidi2_insn"): Likewise.
24074 ("*zero_extendsidi2_insn_sp64"): Likewise.
24075 ("ldfsr"): Likewise.
24076 ("prefetch_64"): Likewise.
24077 ("prefetch_32"): Likewise.
24078 ("tie_ld32"): Likewise.
24079 ("tie_ld64"): Likewise.
24080 ("*tldo_ldub_sp32"): Likewise.
24081 ("*tldo_ldub1_sp32"): Likewise.
24082 ("*tldo_ldub2_sp32"): Likewise.
24083 ("*tldo_ldub_sp64"): Likewise.
24084 ("*tldo_ldub1_sp64"): Likewise.
24085 ("*tldo_ldub2_sp64"): Likewise.
24086 ("*tldo_ldub3_sp64"): Likewise.
24087 ("*tldo_lduh_sp32"): Likewise.
24088 ("*tldo_lduh1_sp32"): Likewise.
24089 ("*tldo_lduh_sp64"): Likewise.
24090 ("*tldo_lduh1_sp64"): Likewise.
24091 ("*tldo_lduh2_sp64"): Likewise.
24092 ("*tldo_lduw_sp32"): Likewise.
24093 ("*tldo_lduw_sp64"): Likewise.
24094 ("*tldo_lduw1_sp64"): Likewise.
24095 ("*tldo_ldx_sp64"): Likewise.
24096 ("*mov<VM32:mode>_insn"): Likewise.
24097 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24098 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24100 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24102 * config/sparc/sparc.md ("type"): New insn type viscmp.
24103 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24105 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24106 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24107 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24108 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24110 ("n7_vis_logical_11cycle"): Likewise.
24111 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24112 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24113 * config/sparc/niagara.md ("niag_vis"): Likewise.
24114 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24115 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24117 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24119 * config/sparc/sparc.md: New instruction type `bmask'.
24120 (bmaskdi_vis): Use the `bmask' type.
24121 (bmasksi_vis): Likewise.
24122 * config/sparc/ultra3.md (us3_array): Likewise.
24123 * config/sparc/niagara7.md (n7_array): Likewise.
24124 * config/sparc/niagara4.md (n4_array): Likewise.
24125 * config/sparc/niagara2.md (niag2_vis): Likewise.
24126 (niag3_vis): Likewise.
24127 * config/sparc/niagara.md (niag_vis): Likewise.
24129 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24131 * ipa-comdats.c: Remove optimize check from gate.
24132 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24133 for functions not optimized.
24134 (ipa_fn_summary_read): Skip optimize check.
24135 (ipa_fn_summary_write): Likewise.
24136 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24138 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24140 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24141 (check_callers): Give up on caller which is not optimized.
24142 (inline_small_functions): Likewise.
24143 (ipa_inline): Do not give up when not optimizing.
24144 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24145 away unoptimizes cdtors.
24146 (whole_program_function_and_variable_visibility): Do
24147 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24148 * ipa.c (process_references): Do not check optimize.
24149 (symbol_table::remove_unreachable_nodes): Update optimize check.
24150 (set_writeonly_bit): Update optimize check.
24151 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24152 (pass_ipa_single_use::gate): Remove.
24154 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24156 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24157 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24158 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24159 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24160 permute_load, permute_store, adjust_extract, adjust_splat,
24161 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24162 replace_swap_with_copy, dump_swap_insn_table,
24163 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24164 recombine_lvx_pattern, recombine_stvx_pattern,
24165 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24166 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24167 to file rs6000-p8swap.c.
24168 * config/rs6000/rs6000-p8swap.c: New file.
24169 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24170 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24171 and rs6000*-*-* targets.
24173 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24175 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24177 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24179 * lto-wrapper.c (merge_and_complain): Do not merge
24180 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24181 fsigned_zeros, ftrapping_math, fwrapv.
24182 (append_compiler_options): Do not track these options.
24183 (append_linker_options): Likewie
24185 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24187 * cgraphunit.c (cgraph_node::finalize_function): When
24188 !flag_toplevel_reorde set no_reorder flag.
24189 (varpool_node::finalize_decl): Likewise.
24190 (symbol_table::compile): Drop no toplevel reorder path.
24192 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24194 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24195 edges; zero probability is not better than uninitialized.
24197 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24199 * asan.h (asan_sanitize_allocas_p): Declare.
24200 * asan.c (asan_sanitize_allocas_p): New function.
24201 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24202 (handle_builtin_alloca): Likewise.
24203 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24204 if !asan_sanitize_allocas_p.
24205 * params.def (asan-instrument-allocas): Add new option.
24206 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24207 * opts.c (common_handle_option): Disable allocas sanitization for
24210 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24212 * asan.c: Include gimple-fold.h.
24213 (get_last_alloca_addr): New function.
24214 (handle_builtin_stackrestore): Likewise.
24215 (handle_builtin_alloca): Likewise.
24216 (asan_emit_allocas_unpoison): Likewise.
24217 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24218 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24219 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24220 (instrument_builtin_call): Pass gimple iterator to
24221 get_mem_refs_of_builtin_call.
24222 (last_alloca_addr): New global.
24223 * asan.h (asan_emit_allocas_unpoison): Declare.
24224 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24225 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24226 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24227 if function calls alloca.
24228 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24229 * gimple-fold.h (replace_call_with_value): Declare.
24230 * internal-fn.c: Include asan.h.
24231 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24232 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24234 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24236 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24237 (C_SELFTEST_FLAGS): New.
24238 (CPP_SELFTEST_FLAGS): New.
24239 (SELFTEST_DEPS): New, from deps of s-selftest.
24240 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24241 (CPP_SELFTEST_DEPS): New.
24242 (selftest): Add dependency on s-selftest-c++.
24243 (s-selftest): Rename to...
24244 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24245 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24246 than SELFTEST_FLAGS.
24247 (selftest-gdb): Rename to...
24248 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24250 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24251 (selftest-valgrind): Rename to...
24252 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24254 (selftest-valgrind): Reintroduce as an alias for
24255 selftest-c-valgrind.
24256 (s-selftest-c++): New.
24257 (selftest-c++-gdb): New.
24258 (selftest-c++-valgrind): New.
24260 2017-07-06 Olivier Hainque <hainque@adacore.com>
24262 * gcc.c (process_command): When deciding if undefined variables
24263 should be ignored when processing specs, accept "gcc -v" as well.
24265 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24267 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24268 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24270 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24272 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24273 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24274 * config/arm/arm-tables.opt: Regenerate.
24275 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24277 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24279 2017-07-06 Carl Love <cel@us.ibm.com>
24281 * ChangeLog: Clean up from mid air collision
24283 2017-07-06 Carl Love <cel@us.ibm.com>
24285 * config/rs6000/rs6000-c.c: Add support for built-in functions
24286 vector signed int vec_subc (vector signed int, vector signed int);
24287 vector signed __int128 vec_subc (vector signed __int128,
24288 vector signed __int128);
24289 vector unsigned __int128 vec_subc (vector unsigned __int128,
24290 vector unsigned __int128);
24291 vector signed int vec_sube (vector signed int, vector signed int,
24292 vector signed int);
24293 vector unsigned int vec_sube (vector unsigned int,
24294 vector unsigned int,
24295 vector unsigned int);
24296 vector signed __int128 vec_sube (vector signed __int128,
24297 vector signed __int128,
24298 vector signed__int128);
24299 vector unsigned __int128 vec_sube (vector unsigned __int128,
24300 vector unsigned __int128,
24301 vector unsigned __int128);
24302 vector signed int vec_subec (vector signed int, vector signed int,
24303 vector signed int);
24304 vector unsigned int vec_subec (vector unsigned int,
24305 vector unsigned int,
24306 vector unsigned int);
24307 vector signed __int128 vec_subec (vector signed __int128,
24308 vector signed __int128,
24309 vector signed__int128);
24310 vector unsigned __int128 vec_subec (vector unsigned __int128,
24311 vector unsigned __int128,
24312 vector unsigned __int128);
24313 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24314 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24315 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24316 BU_ALTIVEC_OVERLOAD_X definitions.
24317 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24318 * doc/extend.texi: Update the built-in documentation file for the new
24319 built-in functions.
24321 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24324 * diagnostic-show-locus.c (layout::layout): Use start and finish
24325 spelling location for the start and finish of each range.
24326 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24327 Add unused aspect param.
24328 * input.c (expand_location_1): Add "aspect" param, and use it
24329 to access the correct part of the location.
24330 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24332 (expand_location_to_spelling_point): Likewise.
24333 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24334 param, and pass it to expand_location_1.
24336 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24338 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24339 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24340 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24341 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24342 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24343 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24344 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24345 _mm_maskz_getmant_ss): New intrinsics.
24346 (__builtin_ia32_getexpss128_mask): Changed to ...
24347 __builtin_ia32_getexpss128_round ... this.
24348 (__builtin_ia32_getexpsd128_mask): Changed to ...
24349 __builtin_ia32_getexpsd128_round ... this.
24350 * config/i386/i386-builtin-types.def
24351 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24352 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24353 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24354 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24355 __builtin_ia32_getmantss_mask_round): New builtins.
24356 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24357 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24358 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24359 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24360 * config/i386/sse.md
24361 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24362 avx512f_sgetexp<mode><mask_scalar_name>
24363 <round_saeonly_scalar_name> ... this.
24364 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24365 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24366 vgetexp<ssescalarmodesuffix>
24367 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24368 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24369 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24370 avx512f_vgetmant<mode><mask_scalar_name>
24371 <round_saeonly_scalar_name> ... this.
24372 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24373 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24374 vgetmant<ssescalarmodesuffix>
24375 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24376 %0<mask_scalar_operand4>, %1, %2
24377 <round_saeonly_scalar_mask_op4>, %3} ... this.
24378 * config/i386/subst.md (mask_scalar_operand4,
24379 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24380 round_saeonly_scalar_nimm_predicate): New subst attributes.
24382 2017-07-06 Julia Koval <julia.koval@intel.com>
24384 * config/i386/i386.c (ix86_erase_embedded_rounding):
24385 Remove code for old rounding pattern.
24387 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24389 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24391 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24393 * doc/sourcebuild.texi (Test Directives, Variants of
24394 dg-require-support): Add documentation for dg-require-stack-check.
24396 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24398 * config/i386/subst.md (mask_scalar, round_scalar,
24399 round_saeonly_scalar): New meta-templates.
24400 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24401 round_scalar_mask_operand3, round_scalar_mask_op3,
24402 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24403 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24404 round_saeonly_scalar_constraint,
24405 round_saeonly_scalar_prefix): New subst attribute.
24406 * config/i386/sse.md
24407 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24408 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24409 <round_scalar_name> ... this.
24410 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24411 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24412 <round_scalar_name> ... this.
24413 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24414 <sse>_vm<code><mode>3<mask_scalar_name>
24415 <round_saeonly_scalar_name> ... this.
24416 (v<plusminus_mnemonic><ssescalarmodesuffix>
24417 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24418 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24419 v<plusminus_mnemonic><ssescalarmodesuffix>
24420 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24421 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24422 (v<multdiv_mnemonic><ssescalarmodesuffix>
24423 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24424 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24425 v<multdiv_mnemonic><ssescalarmodesuffix>
24426 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24427 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24428 (v<maxmin_float><ssescalarmodesuffix>
24429 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24430 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24431 v<maxmin_float><ssescalarmodesuffix>
24432 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24433 %0<mask_scalar_operand3>, %1, %<iptr>2
24434 <round_saeonly_scalar_mask_op3>} ... this.
24436 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24438 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24439 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24441 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24442 Alan Hayward <alan.hayward@arm.com>
24443 David Sherwood <david.sherwood@arm.com>
24445 * combine.c (simplify_if_then_else): Remove "enum" before
24447 * compare-elim.c (can_eliminate_compare): Likewise.
24448 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24450 (aarch64_lookup_simd_builtin_type): Likewise.
24451 (aarch64_simd_builtin_type): Likewise.
24452 (aarch64_init_simd_builtin_types): Likewise.
24453 (aarch64_simd_expand_args): Likewise.
24454 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24456 (aarch64_reverse_mask): Likewise.
24457 (aarch64_simd_emit_reg_reg_move): Likewise.
24458 (aarch64_gen_adjusted_ldpstp): Likewise.
24459 (aarch64_ccmp_mode_to_code): Likewise.
24460 (aarch64_operands_ok_for_ldpstp): Likewise.
24461 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24462 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24464 (aarch64_min_divisions_for_recip_mul): Likewise.
24465 (aarch64_reassociation_width): Likewise.
24466 (aarch64_get_condition_code_1): Likewise.
24467 (aarch64_simd_emit_reg_reg_move): Likewise.
24468 (aarch64_simd_attr_length_rglist): Likewise.
24469 (aarch64_reverse_mask): Likewise.
24470 (aarch64_operands_ok_for_ldpstp): Likewise.
24471 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24472 (aarch64_gen_adjusted_ldpstp): Likewise.
24473 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24475 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24476 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24477 (arm_lookup_simd_builtin_type): Likewise.
24478 (arm_simd_builtin_type): Likewise.
24479 (arm_init_simd_builtin_types): Likewise.
24480 (arm_expand_builtin_args): Likewise.
24481 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24482 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24483 (ft32_setup_incoming_varargs): Likewise.
24484 (ft32_function_arg): Likewise.
24485 (ft32_function_arg_advance): Likewise.
24486 (ft32_pass_by_reference): Likewise.
24487 (ft32_arg_partial_bytes): Likewise.
24488 (ft32_valid_pointer_mode): Likewise.
24489 (ft32_addr_space_pointer_mode): Likewise.
24490 (ft32_addr_space_legitimate_address_p): Likewise.
24491 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24493 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24494 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24495 (iamcu_alignment): Likewise.
24496 (canonicalize_vector_int_perm): Likewise.
24497 (ix86_noce_conversion_profitable_p): Likewise.
24498 (ix86_mpx_bound_mode): Likewise.
24499 (ix86_operands_ok_for_move_multiple): Likewise.
24500 * config/microblaze/microblaze-protos.h
24501 (microblaze_expand_conditional_branch_reg): Likewise.
24502 * config/microblaze/microblaze.c
24503 (microblaze_expand_conditional_branch_reg): Likewise.
24504 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24506 (rs6000_reassociation_width): Likewise.
24507 (rs6000_invalid_binary_op): Likewise.
24508 (fusion_p9_p): Likewise.
24509 (emit_fusion_p9_load): Likewise.
24510 (emit_fusion_p9_store): Likewise.
24511 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24513 (riscv_hard_regno_mode_ok_p): Likewise.
24514 (riscv_address_insns): Likewise.
24515 (riscv_split_symbol): Likewise.
24516 (riscv_legitimize_move): Likewise.
24517 (riscv_function_value): Likewise.
24518 (riscv_hard_regno_nregs): Likewise.
24519 (riscv_expand_builtin): Likewise.
24520 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24521 (riscv_build_integer): Likewise.
24522 (riscv_split_integer): Likewise.
24523 (riscv_legitimate_constant_p): Likewise.
24524 (riscv_cannot_force_const_mem): Likewise.
24525 (riscv_regno_mode_ok_for_base_p): Likewise.
24526 (riscv_valid_base_register_p): Likewise.
24527 (riscv_valid_offset_p): Likewise.
24528 (riscv_valid_lo_sum_p): Likewise.
24529 (riscv_classify_address): Likewise.
24530 (riscv_legitimate_address_p): Likewise.
24531 (riscv_address_insns): Likewise.
24532 (riscv_load_store_insns): Likewise.
24533 (riscv_force_binary): Likewise.
24534 (riscv_split_symbol): Likewise.
24535 (riscv_force_address): Likewise.
24536 (riscv_legitimize_address): Likewise.
24537 (riscv_move_integer): Likewise.
24538 (riscv_legitimize_const_move): Likewise.
24539 (riscv_legitimize_move): Likewise.
24540 (riscv_address_cost): Likewise.
24541 (riscv_subword): Likewise.
24542 (riscv_output_move): Likewise.
24543 (riscv_canonicalize_int_order_test): Likewise.
24544 (riscv_emit_int_order_test): Likewise.
24545 (riscv_function_arg_boundary): Likewise.
24546 (riscv_pass_mode_in_fpr_p): Likewise.
24547 (riscv_pass_fpr_single): Likewise.
24548 (riscv_pass_fpr_pair): Likewise.
24549 (riscv_get_arg_info): Likewise.
24550 (riscv_function_arg): Likewise.
24551 (riscv_function_arg_advance): Likewise.
24552 (riscv_arg_partial_bytes): Likewise.
24553 (riscv_function_value): Likewise.
24554 (riscv_pass_by_reference): Likewise.
24555 (riscv_setup_incoming_varargs): Likewise.
24556 (riscv_print_operand): Likewise.
24557 (riscv_elf_select_rtx_section): Likewise.
24558 (riscv_save_restore_reg): Likewise.
24559 (riscv_for_each_saved_reg): Likewise.
24560 (riscv_register_move_cost): Likewise.
24561 (riscv_hard_regno_mode_ok_p): Likewise.
24562 (riscv_hard_regno_nregs): Likewise.
24563 (riscv_class_max_nregs): Likewise.
24564 (riscv_memory_move_cost): Likewise.
24565 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24566 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24567 (rl78_addr_space_address_mode): Likewise.
24568 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24570 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24571 (rs6000_reassociation_width): Likewise.
24572 (rs6000_invalid_binary_op): Likewise.
24573 (fusion_p9_p): Likewise.
24574 (emit_fusion_p9_load): Likewise.
24575 (emit_fusion_p9_store): Likewise.
24576 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24577 (ok_for_simple_move_operands): Likewise.
24578 (ok_for_simple_move_strict_operands): Likewise.
24579 (ok_for_simple_arith_logic_operands): Likewise.
24580 (visium_legitimize_reload_address): Likewise.
24581 (visium_select_cc_mode): Likewise.
24582 (output_cbranch): Likewise.
24583 (visium_split_double_move): Likewise.
24584 (visium_expand_copysign): Likewise.
24585 (visium_expand_int_cstore): Likewise.
24586 (visium_expand_fp_cstore): Likewise.
24587 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24588 (visium_function_arg): Likewise.
24589 (visium_function_arg_advance): Likewise.
24590 (visium_libcall_value): Likewise.
24591 (visium_setup_incoming_varargs): Likewise.
24592 (visium_legitimate_constant_p): Likewise.
24593 (visium_legitimate_address_p): Likewise.
24594 (visium_legitimize_address): Likewise.
24595 (visium_secondary_reload): Likewise.
24596 (visium_register_move_cost): Likewise.
24597 (visium_memory_move_cost): Likewise.
24598 (prepare_move_operands): Likewise.
24599 (ok_for_simple_move_operands): Likewise.
24600 (ok_for_simple_move_strict_operands): Likewise.
24601 (ok_for_simple_arith_logic_operands): Likewise.
24602 (visium_function_value_1): Likewise.
24603 (rtx_ok_for_offset_p): Likewise.
24604 (visium_legitimize_reload_address): Likewise.
24605 (visium_split_double_move): Likewise.
24606 (visium_expand_copysign): Likewise.
24607 (visium_expand_int_cstore): Likewise.
24608 (visium_expand_fp_cstore): Likewise.
24609 (visium_split_cstore): Likewise.
24610 (visium_select_cc_mode): Likewise.
24611 (visium_split_cbranch): Likewise.
24612 (output_cbranch): Likewise.
24613 (visium_print_operand_address): Likewise.
24614 * expmed.c (flip_storage_order): Likewise.
24615 * expmed.h (emit_cstore): Likewise.
24616 (flip_storage_order): Likewise.
24617 * genrecog.c (validate_pattern): Likewise.
24618 * hsa-gen.c (gen_hsa_addr): Likewise.
24619 * internal-fn.c (expand_arith_overflow): Likewise.
24620 * ira-color.c (allocno_copy_cost_saving): Likewise.
24621 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24622 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24623 (process_invariant_for_inheritance): Likewise.
24624 * lra-eliminations.c (move_plus_up): Likewise.
24625 * omp-low.c (lower_oacc_reductions): Likewise.
24626 * simplify-rtx.c (simplify_subreg): Likewise.
24627 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24628 (TARGET_CHKP_BOUND_MODE): Likewise..
24629 * targhooks.c (default_chkp_bound_mode): Likewise.
24630 (default_setup_incoming_vararg_bounds): Likewise.
24631 * targhooks.h (default_chkp_bound_mode): Likewise.
24632 (default_setup_incoming_vararg_bounds): Likewise.
24633 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24634 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24635 (have_whole_vector_shift): Likewise.
24636 * tree-vect-stmts.c (vectorizable_load): Likewise.
24637 * doc/tm.texi: Regenerate.
24639 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24641 Graceful degrade if Binutils PR21472 is not available.
24644 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24645 .rodata in flash test fails.
24646 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24647 * confgure: Regenerate.
24648 * config.in: Regenerate.
24649 * config/avr/avr.c (avr_asm_named_section)
24650 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24651 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24652 (avr_asm_init_sections): Same.
24654 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24656 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24657 (fma<VH:mode>4_intrinsic): Likewise.
24658 (*fmsub<VCVTF:mode>4): Likewise.
24659 (*fmsub<VH:mode>4_intrinsic): Likewise.
24661 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24664 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24665 Don't depend on "optimize > 0".
24666 (out_movhi_r_mr, out_movqi_mr_r): Same.
24667 (out_movhi_mr_r, out_movqi_r_mr): Same.
24668 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24669 io_address_operand on "optimize > 0".
24671 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24673 * tree-loop-distribution.c: Add general explanantion on the pass.
24674 (generate_loops_for_partition): Mark distributed loop.
24675 (pg_add_dependence_edges): New parameter. Handle alias data
24676 dependence specially and record it in the parameter if asked.
24677 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24678 (init_partition_graph_vertices, add_partition_graph_edge): New.
24679 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24680 (free_partition_graph_vdata, build_partition_graph): New.
24681 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24682 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24683 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24684 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24685 (version_for_distribution_p, finalize_partitions): New.
24686 (distribute_loop): Handle alias data dependence specially. Factor
24687 out loop fusion code as functions and call these functions.
24689 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24691 * tree-loop-distribution.c (classify_partition): New parameter and
24692 better handle reduction statement.
24693 (rdg_build_partitions): Revise comment.
24694 (distribute_loop): Compute statements in all partitions and pass it
24695 to classify_partition.
24697 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24699 * tree-loop-distribution.c (enum partition_type): New.
24700 (struct partition): New field type.
24701 (partition_merge_into): Add parameter. Update partition type.
24702 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24703 (build_rdg_partition_for_vertex): Compute partition type.
24704 (rdg_build_partitions): Dump partition type.
24705 (distribute_loop): Update calls to partition_merge_into.
24707 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24709 * tree-loop-distribution.c (struct ddr_hasher): New.
24710 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24712 (classify_partition): Call get_data_dependence.
24713 (pg_add_dependence_edges): Ditto.
24714 (distribute_loop): Release data dependence hash table.
24716 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24718 * tree-loop-distribution.c (ref_base_address): Delete.
24719 (similar_memory_accesses): Rename ...
24720 (share_memory_accesses): ... to this. Check if partitions access
24721 the same memory reference.
24722 (distribute_loop): Call share_memory_accesses.
24724 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24726 * tree-loop-distribution.c (struct partition): New field recording
24727 its data reference.
24728 (partition_alloc, partition_free): Init and release data refs.
24729 (partition_merge_into): Merge data refs.
24730 (build_rdg_partition_for_vertex): Collect data refs for partition.
24731 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24733 (distribute_loop): Remve data refs from vertice data of partition
24736 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24738 * tree-loop-distribution.c (params.h): Include header file.
24739 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24740 (datarefs_vec): New global var.
24741 (create_rdg_vertices): Use datarefs_vec directly.
24742 (free_rdg): Don't free data references.
24743 (build_rdg): Update use. Don't free data references.
24744 (distribute_loop): Compute global variable for data references.
24745 Bail out if there are too many data references.
24747 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24749 * tree-loop-distribution.c (loop_nest): New global var.
24750 (build_rdg): Use loop directly, rather than loop nest.
24751 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24753 (distribute_loop): Compute global variable loop nest. Update use.
24755 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24757 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24758 (partition_merge_into): New parameter. Dump reason for fusion.
24759 (distribute_loop): Update use of partition_merge_into.
24761 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24763 * tree-loop-distribution.c (bb_top_order_index): New.
24764 (bb_top_order_index_size, bb_top_order_cmp): New.
24765 (stmts_from_loop): Use topological order.
24766 (pass_loop_distribution::execute): Compute and release topological
24767 order for basic blocks.
24769 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24771 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24774 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24776 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24777 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24778 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24779 * internal-fn.def (LOOP_DIST_ALIAS): New.
24780 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24781 (fold_loop_internal_call): ... this.
24782 (vect_loop_dist_alias_call): New function.
24783 (set_uid_loop_bbs): Call fold_loop_internal_call.
24784 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24787 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24790 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24791 Require dead FLAGS_REG at the beginning of a peephole.
24793 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24796 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24797 arguments in the call to __builtin_ia32_sbb_u32.
24798 (_subborrow_u64): Swap _X and _Y arguments in the call to
24799 __builtin_ia32_sbb_u64.
24801 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24804 * tree-vrp.c (compare_assert_loc): Turn into a function template
24805 with stable template parameter. Only test if a->e is NULL,
24806 !a->e == !b->e has been verified already. Use e == NULL or
24807 e != NULL instead of e or ! e tests. If stable is true, don't use
24808 iterative_hash_expr, on the other side allow a or b or both NULL
24809 and sort the NULLs last.
24810 (process_assert_insertions): Sort using compare_assert_loc<false>
24811 instead of compare_assert_loc, later sort using
24812 compare_assert_loc<true> before calling process_assert_insertions_for
24813 in a loop. Use break instead of continue once seen NULL pointer.
24815 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24817 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24818 Cortex-R7 and Cortex-R8 processors.
24820 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24822 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24823 uninitialized while src is not.
24825 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24827 * common/config/arm/arm-common.c: Adjust include path for
24829 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24830 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24831 (arm-cpu-data.h): Likewise.
24832 (arm-cpu-cdata.h): Likewise.
24833 * config/arm/arm-cpu.h: Delete.
24834 * config/arm/arm-cpu-cdata.h: Delete.
24835 * config/arm/arm-cpu-data.h: Delete.
24837 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24839 * config/arm/arm-cpus.in (cortex-a55): New.
24840 (cortex-a75): Likewise.
24841 (cortex-a75.cortex-a55): Likewise.
24842 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24844 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24845 * config/arm/arm-cpu-cdata.h: Regenerate.
24846 * config/arm/arm-cpu-data.h: Regenerate.
24847 * config/arm/arm-cpu.h: Regenerate.
24848 * config/arm/arm-tables.opt: Regenerate.
24849 * config/arm/arm-tune.md: Regenerate.
24851 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24853 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24855 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24857 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24860 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24862 PR tree-optimization/81292
24863 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24864 full_string_p, also call adjust_related_strinfos if the adjustment
24865 is simple, otherwise invalidate related strinfos.
24867 2017-07-04 Martin Liska <mliska@suse.cz>
24870 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24871 newly created variable as DECL_IGNORED_P.
24873 2017-07-04 Martin Liska <mliska@suse.cz>
24876 * ipa-inline.c (inline_small_functions):
24877 Use xstrdup_for_dump.
24879 2017-07-04 Tom de Vries <tom@codesourcery.com>
24881 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24883 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24886 * config/darwin.c (darwin_function_switched_text_sections):
24889 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24891 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24893 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24895 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24897 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24899 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24900 min_profitable_iters, and th as inclusive lower bounds.
24901 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24902 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24903 for min_profitable_iters and min_profitable_estimate.
24904 (vect_transform_loop): Treat th as an inclusive lower bound.
24905 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24907 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24910 * config/darwin.c (darwin_function_switched_text_sections):
24911 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24912 in two pieces, and suppress the use of buf.
24914 2017-07-03 Nathan Sidwell <nathan@acm.org>
24916 * hash-table.h (hash_table_mod1): Fix indentation.
24918 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24920 PR middle-end/81290
24921 * predict.c (force_edge_cold): Be more careful about propagation
24923 * profile-count.h (profile_probability::guessed,
24924 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24926 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24928 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24930 * doc/invoke.texi (rcpc architecture extension): Document it.
24932 2017-07-03 Richard Biener <rguenther@suse.de>
24934 PR tree-optimization/60510
24935 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24936 the scalar reduction PHI and use it.
24937 (vectorizable_reduction): Properly guard the single_defuse_cycle
24938 path for non-SLP reduction chains where we cannot use it.
24939 Rework reduc_def/index and vector type deduction. Rework
24940 vector operand gathering during reduction op code-gen.
24941 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24942 chains dissolve the chain and leave it to non-SLP reduction
24945 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24947 * tree-data-ref.h (dr_alignment): Declare.
24948 * tree-data-ref.c (dr_alignment): New function.
24949 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24950 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24952 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24954 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24956 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24957 and base_misalignment fields.
24958 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24959 * tree-data-ref.c: Include builtins.h.
24960 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24961 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24962 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24963 * tree-vect-data-refs.c: Include tree-cfg.h.
24964 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24965 fields instead of calculating an alignment here.
24966 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24967 innermost_loop_behavior fields.
24969 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24971 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24973 (DR_STEP_ALIGNMENT): New macro.
24974 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24975 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24976 (create_data_ref): Print it.
24977 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24978 to tell whether the step preserves vector (mis)alignment.
24979 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24980 Move the check for an integer step and generalise to all INTEGER_CST.
24981 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24982 Print the outer step alignment.
24984 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24986 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24987 with offset_alignment.
24988 (DR_ALIGNED_TO): Delete.
24989 (DR_OFFSET_ALIGNMENT): New macro.
24990 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24991 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24992 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24993 (create_data_ref): Likewise.
24994 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24995 (vect_analyze_data_refs): Likewise.
24996 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24997 creating dummy innermost behavior.
24999 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25001 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25002 with a "innermost_loop_behavior *" and refeence tree.
25003 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25004 (create_data_ref): Update call accordingly.
25005 * tree-predcom.c (find_looparound_phi): Likewise.
25007 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25009 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25010 fields with dr_wrt_vec_loop.
25011 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25012 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25013 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25014 (vect_dr_behavior): New function.
25015 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25016 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25017 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25018 track whether the step preserves the misalignment.
25019 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25020 Use vect_dr_behavior.
25021 (vect_setup_realignment): Update call accordingly.
25022 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25023 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25024 call to vect_create_addr_base_for_vector_ref.
25025 (vect_create_cond_for_align_checks): Likewise.
25026 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25027 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25028 (vect_recog_mask_conversion_pattern): Likewise.
25029 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25030 (new_stmt_vec_info): Remove redundant zeroing.
25032 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25034 * common/config/arm/arm-common.c (arm_be8_option): New function.
25035 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25036 (ISA_ARMv6): Add isa_bit_be8.
25037 * config/arm/arm.h (arm_be8_option): Add prototype.
25038 (BE8_SPEC_FUNCTION): New define.
25039 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25040 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25041 (mlittle-endian): Similarly.
25042 (mbe8, mbe32): New options.
25043 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25044 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25046 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25048 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25050 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25052 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25053 (cleanup_tree_cfg_bb): Use it.
25054 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25056 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25058 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25061 * loop-doloop.c (add_test): Update profile.
25063 2017-07-03 Martin Liska <mliska@suse.cz>
25066 * sanopt.c (rewrite_usage_of_param): New function.
25067 (sanitize_rewrite_addressable_params): Likewise.
25068 (pass_sanopt::execute): Call rewrite_usage_of_param.
25070 2017-07-03 Richard Biener <rguenther@suse.de>
25072 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25073 back to using VIEW_CONVERT_EXPR.
25075 2017-07-03 Martin Liska <mliska@suse.cz>
25078 * doc/extend.texi: Document when a resolver function is
25079 generated for target_clones.
25081 2017-07-03 Martin Liska <mliska@suse.cz>
25083 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25084 and shadow memory of auto variables which are subject of
25085 use-after-scope sanitization.
25086 (asan_expand_mark_ifn): Add do set only when is_poison.
25088 2016-07-03 Richard Biener <rguenther@suse.de>
25090 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25092 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25093 (vectorizable_reduction): Perform reduction PHI creation when
25094 visiting a reduction PHI and adjust and simplify code generation
25095 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25096 (vect_transform_loop): Visit reduction PHIs.
25097 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25098 defs into the SLP tree.
25099 (vect_build_slp_tree): Reduction defs terminate the recursion.
25100 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25102 (vect_get_vec_defs_for_stmt_copy): Export.
25103 (vect_get_vec_defs): Likewise.
25104 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25106 (vect_get_vec_defs_for_stmt_copy): Declare.
25107 (vect_get_vec_defs): Likewise.
25109 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25111 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25112 parameter with a "loop" parameter and use it instead of the
25113 loop containing DR_STMT. Don't check simple_iv when doing
25114 BB analysis. Describe the two analysis modes in the comment.
25116 2017-07-03 Tom de Vries <tom@codesourcery.com>
25118 PR tree-optimization/69468
25119 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25120 (find_same_succ_bb): Handle ignore_edge_flags.
25122 2017-07-03 Tom de Vries <tom@codesourcery.com>
25124 PR tree-optimization/81192
25125 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25127 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25129 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25131 2017-07-03 Tom de Vries <tom@codesourcery.com>
25133 PR tree-optimization/81192
25134 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25135 BB_SAME_SUCC (bb) == NULL.
25137 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25139 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25142 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25144 * dumpfile.c: Include profile-count.h
25145 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25147 (insert_cond_bb): Update profile.
25148 * tree-cfg.h (insert_cond_bb): Update prototype.
25149 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25150 * tree-dump.c: Do not include tree-cfg.
25152 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25154 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25156 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25158 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25161 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25163 * tree-complex.c (expand_complex_div_wide): update profile.
25165 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25166 Alan Hayward <alan.hayward@arm.com>
25167 David Sherwood <david.sherwood@arm.com>
25169 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25170 (CORETYPES_H): New define.
25171 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25172 (insn-modes-inline.h, s-modes-inline-h): New rules.
25173 (generated_files): Add insn-modes-inline.h.
25174 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25175 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25176 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25177 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25178 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25179 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25180 (build/gencodes.o, build/genconditions.o): Likewise.
25181 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25182 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25183 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25184 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25185 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25186 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25187 * coretypes.h: Include everything up to real.h for generators.
25188 Include insn-modes.h first. Include wide-int-print.h after
25189 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25190 * machmode.h: Don't include insn-modes.h here.
25191 * function-tests.c: Remove includes of signop.h, machmode.h,
25192 double-int.h and wide-int.h.
25194 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25196 * optc-save-gen.awk: Likewise.
25197 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25198 * godump.c: Remove include of wide-int-print.h.
25199 * pretty-print.h: Likewise.
25200 * wide-int-print.cc: Likewise.
25201 * wide-int.cc: Likewise.
25202 * hash-map-tests.c: Remove include of signop.h.
25203 * hash-set-tests.c: Likewise.
25204 * rtl-tests.c: Likewise.
25205 * mkconfig.sh: Remove include of machmode.h.
25206 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25208 (emit_insn_modes_inline_h): ...this new function. Emit the code
25209 into an insn-modes-inline.h header file, adding appropriate
25210 include guards and end comments.
25211 (emit_insn_modes_c_header): Remove include of machmode.h.
25212 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25214 (main): Handle -i flag and call emit_insn_modes_inline_h when
25217 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25219 * tree-ssa-strlen.c (strinfo): Rename the length field to
25220 nonzero_chars. Add a full_string_p field.
25221 (compare_nonzero_chars, zero_length_string_p): New functions.
25222 (get_addr_stridx): Add an offset_out parameter.
25223 Use compare_nonzero_chars.
25224 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25225 (new_strinfo): Update after above changes to strinfo.
25226 (set_endptr_and_length): Set full_string_p.
25227 (get_string_length): Update after above changes to strinfo.
25228 (unshare_strinfo): Update call to new_strinfo.
25229 (maybe_invalidate): Likewise.
25230 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25231 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25232 as a uhwi instead of an shwi. Update after above changes to
25233 strinfo and new_strinfo.
25234 (zero_length_string): Assert that chainsi contains full strings.
25235 Use zero_length_string_p. Update call to new_strinfo.
25236 (adjust_related_strinfos): Update after above changes to strinfo.
25237 Copy full_string_p from origsi.
25238 (adjust_last_stmt): Use zero_length_string_p.
25239 (handle_builtin_strlen): Update after above changes to strinfo and
25240 new_strinfo. Install the lhs as the string length if the previous
25241 entry didn't describe a full string.
25242 (handle_builtin_strchr): Update after above changes to strinfo
25244 (handle_builtin_strcpy): Likewise.
25245 (handle_builtin_strcat): Likewise.
25246 (handle_builtin_malloc): Likewise.
25247 (handle_pointer_plus): Likewise.
25248 (handle_builtin_memcpy): Likewise. Track nonzero characters
25249 that aren't necessarily followed by a nul terminator.
25250 (handle_char_store): Likewise.
25252 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25254 PR tree-optimization/80769
25255 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25256 for malloc and calloc. Document the new invariant that all related
25257 strinfos have delayed lengths or none do.
25258 (verify_related_strinfos): Move earlier in file.
25259 (set_endptr_and_length): New function, split out from...
25260 (get_string_length): ...here. Also set the lengths of related
25262 (zero_length_string): Assert that chainsi has known (rather than
25264 (adjust_related_strinfos): Likewise.
25266 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25268 PR tree-optimization/81136
25269 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25270 assert that two references with the same misalignment have the same
25271 compile-time misalignment if those compile-time misalignments
25274 2017-07-01 Andi Kleen <ak@linux.intel.com>
25276 * print-tree.c (print_node): Print all attributes.
25278 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25280 * cfg.c (scale_bbs_frequencies): New function.
25281 * cfg.h (scale_bbs_frequencies): Declare it.
25282 * cfgloopanal.c (single_likely_exit): Cleanup.
25283 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25285 (scale_loop_profile): Likewise.
25286 (loop_version): Likewise.
25287 (create_empty_loop_on_edge): Update.
25288 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25289 scale_loop_frequencies, scale_loop_profile, loopify,
25290 loop_version): Update prototypes.
25291 * modulo-sched.c (sms_schedule): Update.
25292 * predict.c (unlikely_executed_edge_p): Also check probability.
25293 (probably_never_executed_edge_p): Fix typo.
25294 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25295 * tree-parloops.c (gen_parallel_loop): Update.
25296 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25297 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25298 * tree-ssa-loop-split.c (split_loop): Update.
25299 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25300 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25301 (vect_loop_versioning): Update.
25302 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25304 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25306 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25308 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25310 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25311 to keep profile consistent.
25313 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25315 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25316 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25317 * profile-count.h (max_safe_multiplier): Make unsigned.
25318 (profile_count::guessed_zero): New.
25320 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25322 * bb-reorder.c (fix_up_crossing_landing_pad,
25323 fix_crossing_conditional_branches): Use make_single_succ_edge
25324 to keep profile consistent.
25326 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25328 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25331 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25334 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25335 the right scopes, make sure cond_jump isn't preserved between multiple
25336 iterations. Search for fallthru edge whenever there are 3+ edges and
25337 use find_fallthru_edge for it.
25339 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25341 Patch by Alexander Monakov <amonakov@ispras.ru>
25342 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25343 probabilities consistently.
25345 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25347 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25348 emit_cmp_and_jump_insns.
25350 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25353 * tree-inline.c (expand_call_inline): Combine profile statuses.
25355 2017-06-30 Andrew Pinski <apinski@cavium.com>
25357 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25358 fold_stmt returned true.
25360 2017-06-30 Nathan Sidwell <nathan@acm.org>
25362 * ggc.h (empty_string): Delete.
25363 * cfgexpand.c (expand_asm_stmt): Use plain "".
25364 * optabs.c (expand_asm_memory_barrier): Likewise.
25365 * stringpool.c (empty_string): Delete.
25366 (digit_vector, digit_string): Delete.
25367 (ggc_alloc_string): Use plain "", don't optimize single digit
25368 strings. Use ggc_alloc_atomic.
25370 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25372 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25373 comparison set and one other set, use the cost of the non-comparison
25376 2017-06-30 Nathan Sidwell <nathan@acm.org>
25378 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25381 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25383 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25384 loops. Remove now unneeded calls to gimple_switch_set_label() that
25385 just set removed labels to NULL_TREE.
25387 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25389 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25390 (set_range_info): ...here. Only call set_range_info_raw if domain
25392 (set_nonzero_bits): Call set_range_info_raw.
25393 * tree-ssanames.h (set_range_info_raw): New.
25395 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25398 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25399 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25400 of nonimmediate_operand and <store_mask_constraint> instead of m
25401 for the input operand. For V8FI iterator, always split if input
25402 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25403 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25404 <store_mask_predicate> instead of register_operand and
25405 <store_mask_constraint> instead of v for the input operand. Make
25406 sure both operands aren't MEMs for if not <mask_applied>.
25408 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25410 * lto-wrapper.c (copy_file) Close both file descriptors before
25413 2017-06-30 Martin Liska <mliska@suse.cz>
25416 * multiple_target.c (create_dispatcher_calls): Make ifunc
25417 also for function that don't have calls or are not referenced.
25419 2017-06-30 Richard Biener <rguenther@suse.de>
25421 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25422 analyze the first scalar stmt. Move vector type computation
25423 for the BB case here from ...
25424 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25425 live operation processing in the SLP case properly.
25427 2017-06-30 Richard Biener <rguenther@suse.de>
25429 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25431 2017-06-30 Martin Liska <mliska@suse.cz>
25434 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25435 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25437 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25439 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25442 2017-06-30 Richard Biener <rguenther@suse.de>
25444 PR tree-optimization/81249
25445 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25446 condition reduction result to original scalar type.
25448 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25450 * profile-count.h (enum profile_quality): Fix typos and whitespace
25453 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25455 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25456 type for branch probabilities.
25458 2017-06-29 Julian Brown <julian@codesourcery.com>
25459 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25461 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25462 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25463 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25464 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25466 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25468 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25469 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25470 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25471 CC usage from generic code to here.
25472 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25473 CC usage into the target macros.
25475 2017-06-29 Maya Rashish <coypu@sdf.org>
25477 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25480 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25482 * arm/arm-builtins.c: Include profile-count.h
25483 * except.c (sjlj_emit_function_enter): Use
25484 profile_probability::unlikely.
25486 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25488 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25489 and tocrel_offset be pointer args rather than implicitly using
25491 (legitimate_constant_pool_address_p, rs6000_emit_move,
25492 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25493 tocrel_offset and use in toc_relative_expr_p call.
25494 (print_operand, print_operand_address): Use static tocrel_base_oac
25495 and tocrel_offset_oac.
25496 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25499 2017-06-29 Maya Rashish <coypu@sdf.org>
25501 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25503 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25505 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25506 objects, take into account only the alignment of 'op0' and 'mode1' if
25509 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25511 * ccmp.c (ccmp_tree_comparison_p): New function.
25512 (ccmp_candidate_p): Update to use above function.
25513 (get_compare_parts): New function.
25514 (expand_ccmp_next): Update to use new functions.
25515 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25517 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25518 take mode as argument.
25519 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25520 * expr.c (expand_expr_real_1): Pass mode as argument.
25522 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25524 * combine.c (combine_instructions): Print insns to dump_file, together
25527 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25529 * asan.c (asan_emit_stack_protection): Update.
25530 (create_cond_insert_point): Update.
25531 * auto-profile.c (afdo_propagate_circuit): Update.
25532 * basic-block.h (struct edge_def): Turn probability to
25533 profile_probability.
25534 (EDGE_FREQUENCY): Update.
25535 * bb-reorder.c (find_traces_1_round): Update.
25536 (better_edge_p): Update.
25537 (sanitize_hot_paths): Update.
25538 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25539 (make_single_succ_edge): Update.
25540 (check_bb_profile): Update.
25541 (dump_edge_info): Update.
25542 (update_bb_profile_for_threading): Update.
25543 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25544 probabilitycount to 0.
25545 * cfgbuild.c (compute_outgoing_frequencies): Update.
25546 * cfgcleanup.c (try_forward_edges): Update.
25547 (outgoing_edges_match): Update.
25548 (try_crossjump_to_edge): Update.
25549 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25550 (expand_gimple_tailcall): Update.
25551 (construct_init_block): Use make_single_succ_edge.
25552 (construct_exit_block): Use make_single_succ_edge.
25553 * cfghooks.c (verify_flow_info): Update.
25554 (redirect_edge_succ_nodup): Update.
25555 (split_edge): Update.
25556 (account_profile_record): Update.
25557 * cfgloopanal.c (single_likely_exit): Update.
25558 * cfgloopmanip.c (scale_loop_profile): Update.
25559 (set_zero_probability): Remove.
25560 (duplicate_loop_to_header_edge): Update.
25561 * cfgloopmanip.h (loop_version): Update prototype.
25562 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25563 (force_nonfallthru_and_redirect): Update.
25564 (update_br_prob_note): Update.
25565 (rtl_verify_edges): Update.
25566 (purge_dead_edges): Update.
25567 (rtl_lv_add_condition_to_bb): Update.
25568 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25569 * cgraphunit.c (init_lowered_empty_function): Update.
25570 (cgraph_node::expand_thunk): Update.
25571 * cilk-common.c: Include profile-count.h
25572 * dojump.c (inv): Remove.
25573 (jumpifnot): Update.
25574 (jumpifnot_1): Update.
25575 (do_jump_1): Update.
25577 (do_jump_by_parts_greater_rtx): Update.
25578 (do_compare_rtx_and_jump): Update.
25579 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25580 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25581 * dwarf2cfi.c: Include profile-count.h
25582 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25583 (sjlj_emit_dispatch_table): Likewise.
25584 * explow.c: Include profile-count.h
25585 * expmed.c (emit_store_flag_force): Update.
25586 (do_cmp_and_jump): Update.
25587 * expr.c (compare_by_pieces_d::generate): Update.
25588 (compare_by_pieces_d::finish_mode): Update.
25589 (emit_block_move_via_loop): Update.
25590 (store_expr_with_bounds): Update.
25591 (store_constructor): Update.
25592 (expand_expr_real_2): Update.
25593 (expand_expr_real_1): Update.
25594 * expr.h (try_casesi, try_tablejump): Update prototypes.
25595 * gimple-pretty-print.c (dump_probability): Update.
25596 (dump_profile): New.
25597 (dump_gimple_label): Update.
25598 (dump_gimple_bb_header): Update.
25599 * graph.c (draw_cfg_node_succ_edges): Update.
25600 * hsa-gen.c (convert_switch_statements): Update.
25601 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25602 (find_if_case_1): Update.
25603 (find_if_case_2): Update.
25604 * internal-fn.c (expand_arith_overflow_result_store): Update.
25605 (expand_addsub_overflow): Update.
25606 (expand_neg_overflow): Update.
25607 (expand_mul_overflow): Update.
25608 (expand_vector_ubsan_overflow): Update.
25609 * ipa-cp.c (good_cloning_opportunity_p): Update.
25610 * ipa-split.c (split_function): Use make_single_succ_edge.
25611 * ipa-utils.c (ipa_merge_profiles): Update.
25612 * loop-doloop.c (add_test): Update.
25613 (doloop_modify): Update.
25614 * loop-unroll.c (compare_and_jump_seq): Update.
25615 (unroll_loop_runtime_iterations): Update.
25616 * lra-constraints.c (lra_inheritance): Update.
25617 * lto-streamer-in.c (input_cfg): Update.
25618 * lto-streamer-out.c (output_cfg): Update.
25619 * mcf.c (adjust_cfg_counts): Update.
25620 * modulo-sched.c (sms_schedule): Update.
25621 * omp-expand.c (expand_omp_for_init_counts): Update.
25622 (extract_omp_for_update_vars): Update.
25623 (expand_omp_ordered_sink): Update.
25624 (expand_omp_for_ordered_loops): Update.
25625 (expand_omp_for_generic): Update.
25626 (expand_omp_for_static_nochunk): Update.
25627 (expand_omp_for_static_chunk): Update.
25628 (expand_cilk_for): Update.
25629 (expand_omp_simd): Update.
25630 (expand_omp_taskloop_for_outer): Update.
25631 (expand_omp_taskloop_for_inner): Update.
25632 * omp-simd-clone.c (simd_clone_adjust): Update.
25633 * optabs.c (expand_doubleword_shift): Update.
25634 (expand_abs): Update.
25635 (emit_cmp_and_jump_insn_1): Update.
25636 (expand_compare_and_swap_loop): Update.
25637 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25638 * predict.c (predictable_edge_p): Update.
25639 (edge_probability_reliable_p): Update.
25640 (set_even_probabilities): Update.
25641 (combine_predictions_for_insn): Update.
25642 (combine_predictions_for_bb): Update.
25643 (propagate_freq): Update.
25644 (estimate_bb_frequencies): Update.
25645 (force_edge_cold): Update.
25646 * profile-count.c (profile_count::dump): Add missing space into dump.
25647 (profile_count::debug): Add newline.
25648 (profile_count::differs_from_p): Explicitly convert to unsigned.
25649 (profile_count::stream_in): Update.
25650 (profile_probability::dump): New member function.
25651 (profile_probability::debug): New member function.
25652 (profile_probability::differs_from_p): New member function.
25653 (profile_probability::differs_lot_from_p): New member function.
25654 (profile_probability::stream_in): New member function.
25655 (profile_probability::stream_out): New member function.
25656 * profile-count.h (profile_count_quality): Rename to ...
25657 (profile_quality): ... this one.
25658 (profile_probability): New.
25659 (profile_count): Update.
25660 * profile.c (compute_branch_probabilities): Update.
25661 * recog.c (peep2_attempt): Update.
25662 * sched-ebb.c (schedule_ebbs): Update.
25663 * sched-rgn.c (find_single_block_region): Update.
25664 (compute_dom_prob_ps): Update.
25665 (schedule_region): Update.
25666 * sel-sched-ir.c (compute_succs_info): Update.
25667 * stmt.c (struct case_node): Update.
25668 (do_jump_if_equal): Update.
25669 (get_outgoing_edge_probs): Update.
25670 (conditional_probability): Update.
25671 (emit_case_dispatch_table): Update.
25672 (expand_case): Update.
25673 (expand_sjlj_dispatch_table): Update.
25674 (emit_case_nodes): Update.
25675 * targhooks.c: Update.
25676 * tracer.c (better_p): Update.
25677 (find_best_successor): Update.
25678 * trans-mem.c (expand_transaction): Update.
25679 * tree-call-cdce.c: Update.
25680 * tree-cfg.c (gimple_split_edge): Upate.
25681 (move_sese_region_to_fn): Upate.
25682 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25683 * tree-eh.c (lower_resx): Upate.
25684 (cleanup_empty_eh_move_lp): Upate.
25685 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25686 * tree-inline.c (copy_edges_for_bb): Update.
25687 (copy_cfg_body): Update.
25688 * tree-parloops.c (gen_parallel_loop): Update.
25689 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25690 (gimple_gen_time_profiler): Update.
25691 * tree-ssa-dce.c (remove_dead_stmt): Update.
25692 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25693 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25694 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25695 (unloop_loops): Update.
25696 (try_peel_loop): Update.
25697 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25698 * tree-ssa-loop-split.c (connect_loops): Update.
25699 (split_loop): Update.
25700 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25701 (hoist_guard): Update.
25702 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25703 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25704 (value_replacement): Update.
25705 * tree-ssa-reassoc.c (branch_fixup): Update.
25706 * tree-ssa-tail-merge.c (replace_block_by): Update.
25707 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25708 (create_edge_and_update_destination_phis): Update.
25709 (compute_path_counts): Update.
25710 (recompute_probabilities): Update.
25711 (update_joiner_offpath_counts): Update.
25712 (freqs_to_counts_path): Update.
25713 (duplicate_thread_path): Update.
25714 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25715 (struct switch_conv_info): Update.
25716 (gen_inbound_check): Update.
25717 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25718 (vect_do_peeling): Update.
25719 (vect_loop_versioning): Update.
25720 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25721 (optimize_mask_stores): Update.
25722 * ubsan.c (ubsan_expand_null_ifn): Update.
25723 * value-prof.c (gimple_divmod_fixed_value): Update.
25724 (gimple_divmod_fixed_value_transform): Update.
25725 (gimple_mod_pow2): Update.
25726 (gimple_mod_pow2_value_transform): Update.
25727 (gimple_mod_subtract): Update.
25728 (gimple_mod_subtract_transform): Update.
25729 (gimple_ic): Update.
25730 (gimple_stringop_fixed_value): Update.
25731 (gimple_stringops_transform): Update.
25732 * value-prof.h: Update.
25734 2017-06-29 Carl Love <cel@us.ibm.com>
25736 * config/rs6000/rs6000-c.c: Add support for built-in functions
25737 vector signed int vec_signed (vector float);
25738 vector signed long long vec_signed (vector double);
25739 vector signed int vec_signed2 (vector double, vector double);
25740 vector signed int vec_signede (vector double);
25741 vector signed int vec_signedo (vector double);
25742 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25743 instruction generator.
25744 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25745 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25746 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25748 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25749 vunsignede_v2df): Add define_expands.
25750 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25751 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25752 VEC_UNSIGNEDO): Add definitions.
25753 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25754 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25755 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25756 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25757 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25758 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25759 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25760 vec_unsignede, vec_unsignedo): Add builtin defines.
25761 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25763 * doc/extend.texi: Update the built-in documentation file for the
25764 new built-in functions.
25766 2017-06-29 Richard Biener <rguenther@suse.de>
25768 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25769 reduction chains to LOOP_VINFO_REDUCTIONS.
25770 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25771 SLP reductions after processing reduction chains.
25773 2017-06-29 Nathan Sidwell <nathan@acm.org>
25775 * builtins.c (fold_builtin_FUNCTION): Use
25776 lang_hooks.decl_printable_name.
25778 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25780 PR middle-end/81194
25781 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25782 with only one label.
25783 * stmt.c (expand_case): Assert NCASES is greater than one.
25785 2017-06-29 Richard Biener <rguenther@suse.de>
25787 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25789 (group_case_labels): Likewise.
25790 (find_taken_edge): Push sanity checking on val to workers...
25791 (find_taken_edge_cond_expr): ... here
25792 (find_taken_edge_switch_expr): ... and here, handle cases
25793 with just a default label.
25794 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25795 (group_case_labels): Likewise.
25796 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25797 group_case_labels does anything cleanup the CFG again.
25799 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25801 PR tree-optimization/81196
25802 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25803 exit condition comparing two IVs.
25805 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25807 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25808 profile to the dummy entry at the end of the list of architectures.
25809 * config/arm/arm-cpu-cdata.h: Regenerated.
25811 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25812 Michael Collison <michael.collison@arm.com>
25815 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25817 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25818 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25819 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25820 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25821 of shift when the shift amount is masked with constant equal to
25822 the size of the mode.
25823 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25826 2017-06-29 Martin Liska <mliska@suse.cz>
25828 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25830 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25833 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25836 (edge_growth_cache_entry::edge_growth_cache_entry): New
25838 (reset_edge_growth_cache): Update to use constructor.
25840 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25842 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25843 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25844 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25846 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25848 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25849 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25851 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25853 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25854 (*-linux-uclibc*): Add t-uclibc tmake_file.
25855 * config/t-musl: New.
25856 * config/t-uclibc: New.
25858 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25860 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25862 (gen_comm_data): Emit architectural setting of arch_prof.
25863 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25865 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25866 (armv8-m.base, armv8-m.main): Likewise.
25867 * arm-protos.h (arm_build_target): Add profile field.
25868 (arch_option): Likewise.
25869 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25871 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25872 arm_active_target.profile.
25874 2017-06-28 Richard Biener <rguenther@suse.de>
25876 PR middle-end/81227
25877 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25878 TYPE_OVERFLOW_WRAPS.
25879 * match.pd (negate_expr_p): Likewise.
25880 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25881 fold_build2, not fold_binary.
25883 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25885 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25886 Convert memory address to Pmode.
25887 (aarch64_print_operand): Assert MEM operands are always Pmode.
25889 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25892 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25893 Remove redundant if.
25894 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25895 * config/arm/aarch-common-protos.h
25896 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25897 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25899 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25902 * multiple_target.c (create_dispatcher_calls): Set the default
25903 clone to be static, not public.
25905 2017-06-28 Richard Biener <rguenther@suse.de>
25907 * tree-vect-loop.c (vectorizable_reduction): Move special
25908 cond reduction IV var creation ...
25909 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25910 parameter. Use STMT_VINFO_VECTYPE.
25911 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25914 2017-06-28 Martin Liska <mliska@suse.cz>
25917 * ipa-visibility.c (non_local_p): Handle visibility.
25919 2017-06-28 Martin Liska <mliska@suse.cz>
25922 * common.opt: Add IntegerRange to various options.
25923 * opt-functions.awk (integer_range_info): New function.
25924 * optc-gen.awk: Add integer_range_info to cl_options struct.
25925 * opts-common.c (decode_cmdline_option): Handle
25926 CL_ERR_INT_RANGE_ARG.
25927 (cmdline_handle_error): Likewise.
25928 * opts.c (print_filtered_help): Show valid interval in
25929 when --help is provided.
25930 * opts.h (struct cl_option): Add range_min and range_max fields.
25931 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25933 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25935 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25936 (x * C EQ/NE y * C): New transformation.
25938 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25940 * genmultilib (combination_space): Accept '+' in option names.
25942 2017-06-28 Martin Liska <mliska@suse.cz>
25945 * asan.c (instrument_derefs): Bail out inner references
25946 that are hard register variables.
25948 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25951 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25952 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25954 2017-06-28 Richard Biener <rguenther@suse.de>
25956 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25957 (vect_get_slp_defs): Adjust.
25958 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25960 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25962 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25963 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25964 (vectorizable_reduction): Adjust.
25965 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25967 (vect_get_slp_defs): Likewise.
25968 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25969 (vectorizable_bswap): Adjust.
25970 (vectorizable_call): Likewise.
25971 (vectorizable_conversion): Likewise.
25972 (vectorizable_assignment): Likewise.
25973 (vectorizable_shift): Likewise.
25974 (vectorizable_operation): Likewise.
25975 (vectorizable_store): Likewise.
25976 (vectorizable_condition): Likewise.
25977 (vectorizable_comparison): Likewise.
25979 2017-06-28 Michael Collison <michael.collison@arm.com>
25982 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25984 (arm_gen_movmemqi): Removed unused variable 'i'.
25985 Convert 'for' loop into 'while' loop.
25986 (arm_expand_prologue): Remove last unnecessary set of insn.
25987 (thumb_pop): Remove unused variable 'pushed_words'.
25988 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25990 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25992 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25993 * config/s390/s390-protos.h: Add prototype of
25994 s390_rel_address_ok_p.
25995 * config/s390/s390.c (s390_got_symbol): New function.
25996 (s390_rel_address_ok_p): New function.
25997 (legitimize_pic_address): Use s390_rel_address_ok_p.
25998 (s390_load_got): Use s390_got_symbol.
25999 (s390_option_override): Issue error if
26000 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26001 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26003 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26005 2017-06-27 Andrew Pinski <apinski@cavium.com>
26007 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26008 (X * copysign (1.0, X)): New pattern.
26009 (X * copysign (1.0, -X)): New pattern.
26010 (copysign (-1.0, CST)): New pattern.
26012 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26014 * genmultilib (combination_space): Remove variable.
26015 Validate reuse rules against regular expression for any sequence
26016 of multilib options in any order.
26018 2017-06-27 Michael Collison <michael.collison@arm.com>
26020 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26021 call aarch64_split_simd_combine.
26022 * (aarch64_combine_internal<mode>): Delete pattern.
26023 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26024 Allow register and subreg operands.
26026 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26028 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26029 specific need, just fallback on defaults.
26030 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26032 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26033 Olivier Hainque <hainque@adacore.com>
26035 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26037 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26038 targets. Pick a default if no particular attempt applied.
26039 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26042 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26044 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26045 (x86_64-wrs-vxworks7): Likewise.
26047 2017-06-27 Marek Polacek <polacek@redhat.com>
26050 * ubsan.c (instrument_null): Check get_base_address's result for null.
26052 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26054 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26056 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26058 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26059 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26060 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26061 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26062 New function types.
26063 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26064 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26065 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26066 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26067 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26068 BUILT_IN_FEUPDATEENV): New builtins.
26069 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26070 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26071 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26072 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26074 (builtin_structptr_types): Adjust size.
26075 * tree.c (builtin_structptr_types): Add four entries.
26077 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26078 Olivier Hainque <hainque@adacore.com>
26080 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26081 (TLS_SYM): New local macro, forcing reference to __tls__ on
26082 link command lines for VxWorks 7 RTPs, triggering initialization
26084 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26085 OS features TLS support, true for RTPs on VxWorks 7.
26086 * config/vxworks.c (vxworks_override_options): Setup emutls
26089 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26091 * predict.c (test_prediction_value_range): Use -1U instead of -1
26092 to avoid narrowing conversion warning.
26093 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26094 to avoid narrowing conversion warning.
26095 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26097 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26099 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26101 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26102 64bit configurations.
26103 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26104 (SIZE_TYPE): Likewise.
26105 * config/vxworks.c (vxworks_emutls_var_fields): Use
26106 long_unsigned_type_node instead of unsigned_type_node as the offset
26107 field type, which is "pointer" mode in emutls.c.
26109 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26112 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26114 PR middle-end/81207
26115 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26116 gimple_vuse copying separately from gimple_vdef copying.
26118 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26120 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26123 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26124 Olivier Hainque <hainque@adacore.com>
26126 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26127 for all vxworks7 targets.
26128 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26129 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26130 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26131 variations for VX6/VX7 and 32/64bits later on in ...
26132 (VXWORKS_LIB_SPEC): Leverage new macros.
26133 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26134 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26136 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26138 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26139 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26141 2017-06-26 Carl Love <cel@us.ibm.com>
26143 * config/rs6000/rs6000-c.c: Add support for built-in functions
26144 vector bool char vec_reve (vector bool char);
26145 vector signed char vec_reve (vector signed char);
26146 vector unsigned char vec_reve (vector unsigned char);
26147 vector bool int vec_reve (vector bool int);
26148 vector signed int vec_reve (vector signed int);
26149 vector unsigned int vec_reve (vector unsigned int);
26150 vector bool long long vec_reve (vector bool long long);
26151 vector signed long long vec_reve (vector signed long long);
26152 vector unsigned long long vec_reve (vector unsigned long long);
26153 vector bool short vec_reve (vector bool short);
26154 vector signed short vec_reve (vector signed short);
26155 vector double vec_reve (vector double);
26156 vector float vec_reve (vector float);
26157 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26158 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26159 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26160 (altivec_vreve): New pattern.
26161 * config/rs6000/altivec.h (vec_reve): New define.
26162 * doc/extend.texi (vec_rev): Update the built-in documentation file
26163 for the new built-in functions.
26165 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26167 PR tree-optimization/71815
26168 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26170 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26172 (slsr_process_phi): Likewise.
26173 (replace_uncond_cands_and_profitable_phis): Don't replace a
26174 multiply candidate with a stride of 1 (copy or cast).
26175 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26177 (lowest_cost_path): Likewise.
26178 (total_savings): Likewise.
26180 2017-06-26 Richard Biener <rguenther@suse.de>
26183 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26184 Use def_builtin_pure for all gather builtins.
26186 2017-06-26 Richard Biener <rguenther@suse.de>
26188 PR tree-optimization/81203
26189 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26190 non-dominating BBs.
26192 2017-06-26 Marek Polacek <polacek@redhat.com>
26195 * doc/invoke.texi: Document -Wmultistatement-macros.
26197 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26199 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26200 arm_neon_ok_no_float_abi effective target.
26202 2017-06-26 Richard Biener <rguenther@suse.de>
26204 PR tree-optimization/80928
26205 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26206 (copy_bbs): Set BB_DUPLICATED flag early.
26207 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26209 (execute_on_shrinking_pred): Likewise.
26210 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26211 BB_DUPLICATED blocks.
26212 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26213 iterate over all PHIs considering removal of *gsi.
26215 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26217 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26220 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26222 * config/rs6000/rs6000-string.c: (expand_block_clear,
26223 do_load_for_compare, select_block_compare_mode,
26224 compute_current_alignment, expand_block_compare,
26225 expand_strncmp_align_check, expand_strn_compare,
26226 expand_block_move, rs6000_output_load_multiple)
26227 Move functions related to string/block move/compare
26228 to a separate file.
26229 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26230 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26231 for this function which is now used in two files.
26232 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26233 * config.gcc: Add rs6000-string.o to extra_objs for
26234 targets powerpc*-*-* and rs6000*-*-*.
26236 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26239 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26240 32-bit, since indexed is not valid for DImode.
26241 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26242 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26243 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26244 (define_peephole2 for Altivec d-form store): Likewise.
26247 * multiple_target.c (create_dispatcher_calls): Only create the
26248 dispatcher call if the function is the default clone of a
26249 versioned function.
26251 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26253 PR middle-end/80902
26254 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26255 a call, force the call to not be a tail call.
26257 2017-06-23 Jeff Law <law@redhat.com>
26259 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26262 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26264 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26265 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26266 handling for early expansion of vector shifts (sl,sr,sra,rl).
26267 (builtin_function_type): Add vector shift right instructions
26268 to the unsigned argument list.
26270 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26272 rtl-optimizatoin/79286
26273 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26274 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26275 trap. PIC register plus a const unspec without offset can never trap.
26277 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26279 * tree.h (builtin_structptr_type): New type.
26280 (builtin_structptr_types): Declare new array.
26281 * tree.c (builtin_structptr_types): New array.
26282 (free_lang_data, build_common_tree_nodes): Use it.
26284 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26287 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26290 2017-06-22 Matt Turner <mattst88@gmail.com>
26292 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26293 Lake models to skylake case. Assume skylake for unknown
26294 models with clflushopt.
26296 2017-06-22 Jeff Law <law@redhat.com>
26298 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26299 frame sizes that do not satisfy aarch64_uimm12_shift.
26301 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26303 * profile-count.h (apply_probability,
26304 apply_scale, probability_in): Fix checks for zero.
26306 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26308 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26309 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26311 2016-06-22 Richard Biener <rguenther@suse.de>
26313 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26314 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26315 REDUC_MAX_EXPR support.
26316 (vectorizable_reduction): Likewise.
26317 (vect_create_epilog_for_reduction): Likewise.
26319 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26321 * match.pd (A / (1 << B) -> A >> B): New.
26322 * generic-match-head.c: Include optabs-tree.h.
26323 * gimple-match-head.c: Likewise.
26324 * optabs-tree.h (target_supports_op_p): New.
26325 * optabs-tree.c (target_supports_op_p): New.
26327 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26329 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26330 $gcc_cv_ld --help output.
26331 (gcc_cv_ld_demangle): Likewise.
26332 (gcc_cv_ld_eh_frame_hdr): Likewise.
26333 (gcc_cv_ld_pie): Likewise.
26334 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26335 (gcc_cv_ld_buildid): Likewise.
26336 (gcc_cv_ld_sysroot): Likewise.
26337 (ld_bndplt_support): Likewise.
26338 (ld_pushpopstate_support): Likewise.
26339 * configure: Regenerate.
26340 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26342 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26345 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26346 operands indexes to avoid gap between operands and match_dups.
26348 2017-06-21 Andrew Pinski <apinski@cavium.com>
26350 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26351 Increment Arith_shift and Arith_shift_reg by 1.
26352 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26354 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26355 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26356 (aarch64_strip_extend): Add new argument and test for it.
26357 (aarch64_cheap_mult_shift_p): New function.
26358 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26359 add a cost if it is true.
26360 Update calls to aarch64_strip_extend.
26361 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26363 2017-06-21 Andrew Pinski <apinski@cavium.com>
26365 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26367 (thunderxt88): Likewise.
26368 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26369 (thunderx_prefetch_tune): New variable.
26370 (thunderx2t99_prefetch_tune): Update for the correct values.
26371 (thunderxt88_tunings): New variable.
26372 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26373 generic_prefetch_tune.
26374 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26376 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26378 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26379 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26380 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26381 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26382 (aarch64_atomic_cas<mode>, GPI): Likewise.
26384 2017-06-21 Martin Liska <mliska@suse.cz>
26386 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26387 statements on cold and hot labels.
26388 * predict.c (tree_estimate_probability_bb): Remove the
26389 prediction from this place.
26391 2017-06-21 Martin Liska <mliska@suse.cz>
26393 PR tree-optimization/79489
26394 * gimplify.c (maybe_add_early_return_predict_stmt): New
26396 (gimplify_return_expr): Call the function.
26397 * predict.c (tree_estimate_probability_bb): Remove handling
26399 * predict.def: Update comment about early return predictor.
26400 * gimple-predict.h (is_gimple_predict): New function.
26401 * predict.def: Change default value of early return to 66.
26402 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26404 * passes.def: Put pass_strip_predict_hints to the beginning of
26407 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26409 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26410 FUNCTION_DECL declarations.
26411 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26413 (dwaf2out_decl): Likewise.
26414 * godump.c (go_early_global_decl): Skip call to the real debug hook
26415 for FUNCTION_DECL declarations.
26416 * passes.c (rest_of_decl_compilation): Skip call to the
26417 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26418 -fdump-go-spec is passed.
26420 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26422 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26423 Reorder for compactness.
26424 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26425 (def_builtin_pure, def_builtin_pure2): New functions.
26426 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26428 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26430 * match.pd (nop_convert): New predicate.
26431 ((A +- CST1) +- CST2): Allow some NOP conversions.
26433 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26436 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26437 with ctors/dtors if GOVD_SHARED is set.
26439 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26441 * config/aarch64/aarch64.md (movti_aarch64):
26442 Emit mov rather than orr.
26443 (movtf_aarch64): Likewise.
26444 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26445 Emit mov rather than orr.
26447 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26449 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26450 Swap alternatives, make integer dup more expensive.
26452 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26454 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26455 Return true for non-tls symbols.
26457 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26459 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26460 (cortex-a75): Likewise.
26461 (cortex-a75.cortex-a55): Likewise.
26462 * config/aarch64/aarch64-tune.md: Regenerate.
26463 * doc/invoke.texi (-mtune): Document new values for -mtune.
26465 2017-06-21 Tom de Vries <tom@codesourcery.com>
26467 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26468 stack_size feature.
26469 (Effective-Target Keywords, Other attributes): Suggest using
26470 dg-add-options stack_size feature to get stack limit in stack_size
26471 effective target documentation.
26473 2017-06-21 Julian Brown <julian@codesourcery.com>
26474 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26476 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26477 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26478 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26480 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26481 attribute type list for neon_multiply.
26482 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26483 attribute type list for neon_multiply.
26484 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26485 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26486 attribute type list for neon_multiply.
26487 * config/arm/types.md (crypto_pmull): Add.
26488 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26489 attribute type list.
26491 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26493 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26496 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26498 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26499 to make sure not to dereference a NULL cost_classes_ptr pointer.
26501 2017-06-20 Carl Love <cel@us.ibm.com>
26503 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26504 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26505 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26506 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26507 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26508 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26509 VMULOSW): New enum "unspec" values.
26510 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26511 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26512 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26513 altivec_vmulosw): New patterns.
26514 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26515 VMULOSW): Add definitions.
26517 2017-06-20 Julia Koval <julia.koval@intel.com>
26519 * config/i386/i386.c: Fix rounding expand for new pattern.
26520 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26522 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26524 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26525 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26527 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26529 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26532 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26534 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26535 architecture, then by implementer ID.
26536 * config/aarch64/aarch64-tune.md: Regenerate.
26538 2017-06-20 Richard Biener <rguenther@suse.de>
26540 PR middle-end/81097
26541 * fold-const.c (split_tree): Fold to type before negating.
26543 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26545 * diagnostic-show-locus.c
26546 (selftest::test_fixit_deletion_affecting_newline): New function.
26547 (selftest::diagnostic_show_locus_c_tests): Call it.
26549 2017-06-20 Andreas Schwab <schwab@suse.de>
26552 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26555 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26557 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26558 __ARM_FEATURE_COPROC according to support.
26560 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26562 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26563 Rewritten to avoid overflow for > 32-bit pointers.
26566 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26567 by removing enum keyword.
26568 (ubsan_type_descriptor): Likewise. Formatting fix.
26571 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26572 splitter): Require TARGET_SSE2 in the condition.
26574 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26577 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26578 for doing vector set of SFmode on ISA 3.0.
26579 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26580 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26582 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26583 SFmode value into a V4SF variable that was extracted from another
26584 V4SF variable without converting the element to double precision
26585 and back to single precision vector format.
26586 (vsx_insert_extract_v4sf_p9_2): Likewise.
26588 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26590 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26591 in UWHI to avoid undefined overflow.
26594 * ubsan.h (enum ubsan_encode_value_phase): New.
26595 (ubsan_encode_value): Change second argument to
26596 enum ubsan_encode_value_phase with default value of
26597 UBSAN_ENCODE_VALUE_GENERIC.
26598 * ubsan.c (ubsan_encode_value): Change second argument to
26599 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26600 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26601 create_tmp_var_raw instead of create_tmp_var and use a
26603 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26604 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26605 ubsan_encode_value callers.
26608 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26609 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26610 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26612 2017-06-19 Richard Biener <rguenther@suse.de>
26614 PR middle-end/81118
26615 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26616 estimates if we changed anything.
26618 2017-06-19 Richard Biener <rguenther@suse.de>
26620 PR tree-optimization/80887
26621 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26622 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26623 simplified lookups, then reset mprts_hook.
26624 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26626 (try_to_simplify): Likewise.
26628 2017-06-19 Martin Liska <mliska@suse.cz>
26631 * gimplify.c (gimplify_switch_expr):
26632 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26634 2017-06-19 Martin Liska <mliska@suse.cz>
26636 * doc/install.texi: Document that PGO runs in 4 stages.
26638 2017-06-19 Martin Liska <mliska@suse.cz>
26641 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26642 to dispatcher function name.
26643 * multiple_target.c (replace_function_decl): New function.
26644 (create_dispatcher_calls): Redirect both edges and references.
26646 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26648 * profile-count.c (profile_count::dump): Dump quality.
26649 (profile_count::differs_from_p): Update for unsigned val.
26650 * profile-count.h (profile_count_quality): New enum.
26651 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26653 2017-06-19 Richard Biener <rguenther@suse.de>
26655 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26656 struct function as arg.
26657 (estimate_numbers_of_iterations): Export overload with loop arg.
26658 (free_numbers_of_iterations_estimates_loop): Use an overload of
26659 free_numbers_of_iterations_estimates instead.
26660 * tree-cfg.c (remove_bb): Adjust.
26661 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26662 * tree-parloops.c (gen_parallel_loop): Likewise.
26663 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26665 (tree_unroll_loops_completely): Likewise.
26666 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26667 Use an overload instead and export.
26668 (estimated_loop_iterations): Adjust.
26669 (max_loop_iterations): Likewise.
26670 (likely_max_loop_iterations): Likewise.
26671 (estimate_numbers_of_iterations): Take struct function as arg
26673 (loop_exits_before_overflow): Adjust.
26674 (free_numbers_of_iterations_estimates_loop): Use an overload.
26675 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26676 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26678 2017-06-19 Richard Biener <rguenther@suse.de>
26681 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26682 RANGE_EXPR conservatively.
26684 2017-06-16 Carl Love <cel@us.ibm.com>
26686 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26687 definitions for vec_float, vec_float2, vec_floato,
26688 vec_floate built-ins.
26689 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26690 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26692 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26693 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26694 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26695 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26696 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26697 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26698 vec_floato): Add builtin defines.
26699 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26700 Update the built-in documentation file for the new built-in
26703 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26705 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26706 (mthumb): Mark as the negative of -marm.
26708 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26710 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26712 (ARM Options, -mtune): Document that this accepts the same
26713 extension options as -mcpu.
26714 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26716 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26718 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26719 permitted extensions.
26721 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26723 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26724 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26725 (armv8-m.main): Add option +nodsp.
26726 * config/arm/arm-cpu-cdata.h: Regenerated.
26728 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26730 * config/arm/t-fuchsia: New file.
26731 * config.gcc (arm*-*-fuchsia*): Use it.
26733 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26735 * config/arm/t-symbian: Rewrite for new option infrastructure.
26737 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26739 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26740 (MULTILIB_REQUIRED): Likewise.
26742 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26744 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26745 (MULTILIB_RESUE): Likewise.
26746 (MULTILIB_MATCHES): Likewise.
26747 (MULTLIB_REQUIRED): Likewise.
26749 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26751 * config/arm/t-rtems: Rewrite for new option framework.
26753 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26755 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26756 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26757 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26758 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26759 * config/arm/t-multilib: ... here.
26760 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26761 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26762 armv7-a and armv8*-a when A-profile libraries have not been built.
26763 * config/arm/t-rmprofile: Rewrite.
26765 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26767 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26768 with a backslash. Remove the backslash after substituting unescaped
26770 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26772 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26774 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26775 the driver to pass through the default setting of -mfloat-abi.
26776 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26778 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26779 (all_feat_combs): New rule.
26780 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26782 * config/arm/t-aprofile: Rewrite.
26784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26786 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26787 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26788 fpu is not specified by the user/command-line.
26789 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26790 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26791 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26792 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26793 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26794 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26798 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26799 * config/arm/t-arm-elf: Rewritten.
26801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26803 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26804 have some floating-point instructions.
26805 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26806 (TARGET_MAYBE_HARD_FLOAT): New macro.
26807 * config/arm/arm-builtins.c (arm_init_builtins): Use
26808 TARGET_MAYBE_HARD_FLOAT.
26809 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26811 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26813 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26814 (configargs.h): Include it.
26815 (arm_print_hint_for_fpu_option): New function.
26816 (arm_parse_fpu_option): New function.
26817 (candidate_extension): New class.
26818 (arm_canon_for_multilib): New function.
26819 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26820 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26821 (ARCH_CANONICAL_SPECS): New macro.
26822 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26824 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26826 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26827 are set after handling multilib fragments. Set target_cpu_default2
26830 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26832 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26834 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26835 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26837 (check_arch): Likewise.
26838 * config/arm/arm.c (arm_configure_build_target): Handle
26839 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26840 options in the default.
26842 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26844 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26845 when an option is an alias of another.
26846 * config/arm/parsecpu.awk (optalias): New parser token.
26847 (gen_comm_data): Mark non-alias options as such. Emit entries
26848 for extension aliases.
26849 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26850 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26851 (armv6kz, armv6zk, armv6t2): Likewise.
26852 (armv7): Make vfpv3-d16 an alias.
26853 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26855 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26856 Sort in canonical order.
26857 (armv8-a): Sort in canonical order.
26858 (armv8.1-a, armv8.2-a): Likewise.
26859 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26861 (cortex-a9): Sort in canonical order.
26862 * config/arm/arm.c (selftests.h): Include it.
26863 (arm_test_cpu_arch_data): New function.
26864 (arm_run_self_tests): New function.
26865 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26866 (targetm): Move declaration to the end of the file.
26867 * arm-cpu-cdata.h: Regenerated.
26869 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26871 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26872 call to target_mode_check describing the type of option passed.
26873 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26874 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26875 arm_parse_cpu_option_name to match parameters against list of
26877 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26878 arm_arch_core_flags data structure.
26879 * config/arm/arm-cpu_cdata.h: Regenerated.
26881 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26883 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26885 (arm_print_hint_for_cpu_option): Likewise.
26886 (arm_print_hint_for_arch_option): Likewise.
26887 (arm_parse_cpu_option_name): Likewise.
26888 (arm_parse_arch_option_name): Likewise.
26889 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26890 of entries in the all_fpus list.
26891 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26892 (arm_parse_cpu_option_name): Declare.
26893 (arm_parse_arch_option_name): Declare.
26894 (arm_parse_option_features): Declare.
26895 (arm_intialize_isa): Declare.
26896 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26898 (gen_comm_data): ... here. Make definitions non-static.
26899 * config/arm/arm-cpu-data.h: Regenerated.
26900 * config/arm/arm-cpu-cdata.h: Regenerated.
26902 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26904 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26905 (cpu_arch_extension): New structure.
26906 (cpu_arch_option, arch_option, cpu_option): New structures.
26907 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26908 architecture types.
26909 (gen_data): Generate new format data tables.
26910 * config/arm/arm.c (cpu_tune): New structure.
26911 (cpu_option, processors): Delete.
26912 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26913 (arm_print_hint_for_cpu_option): ... this and ...
26914 (arm_print_hint_for_arch_option): ... this.
26915 (arm_parse_arch_cpu_name): Delete. Replace with ...
26916 (arm_parse_cpu_option_name): ... this and ...
26917 (arm_parse_arch_option_name): ... this.
26918 (arm_unrecognized_feature): Change type of target parameter to
26920 (arm_parse_arch_cpu_features): Delete. Replace with ...
26921 (arm_parse_option_features): ... this.
26922 (arm_configure_build_target): Rework to use new configuration data
26924 (arm_print_tune_info): Rework for new configuration data tables.
26925 * config/arm/arm-cpu-data.h: Regenerated.
26926 * config/arm/arm-cpu.h: Regenerated.
26928 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26930 * Makefile.in (OBJS): Move sbitmap.o from here ...
26931 (OBJS-libcommon): ... to here.
26933 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26935 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26936 (ISA_ALL_CRYPTO): New macro.
26937 (ISA_ALL_SIMD): New macro
26938 (ISA_ALL_FP): New macro.
26939 * config/arm/arm.c (fpu_bitlist): Update initializer.
26940 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26942 (arm9e): Add fpu. Add option for nofp
26943 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26944 (arm926ej-s, arm1026ej-s): Likewise.
26945 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26946 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26947 neon-fp16, neon-vfpv4, nofp and nosimd.
26948 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26949 (cortex-a8): Add fpu. Add option for nofp.
26950 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26951 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26952 (cortex-r4f): Add fpu.
26953 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26954 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26956 (cortex-r8): Likewise.
26957 (cortex-m4): Add fpu. Add option for nofp.
26958 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26959 (cortex-a17.cortex-a7): Likewise.
26960 (cortex-a32): Add fpu. Add options for crypto and nofp.
26961 (cortex-a35, cortex-a53): Likewise.
26962 (cortex-a57): Add fpu. Add option for crypto.
26963 (cortex-a72, cortex-a73): Likewise.
26964 (exynos-m1): Likewise.
26965 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26966 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26967 (cortex-m33): Add fpu. Add option for nofp.
26968 * config/arm/arm-cpu-cdata.h: Regenerated
26969 * config/arm/arm-cpu-data.h: Regenerated.
26971 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26973 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26974 (armv5te, armv5tej): Likewise.
26975 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26976 (armv7): Add options fp and vfpv3-d16.
26977 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26978 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26980 (armv7ve): Likewise.
26981 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26982 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26983 (armv8-a): Add nocrypto option.
26984 (armv8.1-a, armv8.2-a): Likewise.
26985 (armv8-m.main): add options fp, fp.dp and nofp.
26987 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26989 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26991 (armv8-a+crc): Delete.
26992 (armv8.1-a): Add options simd, crypto and nofp.
26993 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26994 (armv8.2-a+fp16): Delete.
26995 (armv8-m.main): Add option dsp.
26996 (armv8-m.main+dsp): Delete.
26997 (cortex-a8): Add fpu. Add nofp option.
26998 (cortex-a9): Add fpu. Add nofp and nosimd options.
26999 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27000 link to main cpu and architecture data structures.
27001 (gen_comm_data): Only put isa attributes from the main architecture
27003 (option): New statement for architecture and CPU entries.
27004 * arm.c (struct cpu_option): New structure.
27005 (struct processors): Add entry for options.
27006 (arm_unrecognized_feature): New function.
27007 (arm_parse_arch_cpu_name): Ignore any characters after the first
27009 (arm_parse_arch_cpu_feature): New function.
27010 (arm_configure_build_target): Separate out any CPU and architecture
27011 features and parse separately. Don't error out if -mfpu=auto is
27012 used with only an architecture string.
27013 (arm_print_asm_arch_directives): New function.
27014 (arm_file_start): Call it.
27015 * config/arm/arm-cpu-cdata.h: Regenerated.
27016 * config/arm/arm-cpu-data.h: Likewise.
27017 * config/arm/arm-tables.opt: Likewise.
27019 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27021 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27022 assembler when it is not -mfpu=auto.
27024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27026 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27027 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27028 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27029 (ASM_CPU_SPEC): Rewrite.
27030 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27031 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27032 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27033 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27034 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27035 copied string is NUL-terminated. Also strip any characters prefixed
27037 (arm_rewrite_selected_arch): New function.
27038 (arm_rewrite_march): New function.
27040 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27042 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27043 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27044 (march, mcpu, mtune): Convert to string-based options.
27045 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27046 (arm_parse_arch_cpu_name): New function.
27047 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27048 identify selected architecture or CPU.
27049 (arm_option_save): New function.
27050 (TARGET_OPTION_SAVE): Redefine.
27051 (arm_option_restore): Restore string options.
27052 (arm_option_print): Print string options.
27054 2017-06-16 Martin Sebor <msebor@redhat.com>
27056 PR tree-optimization/80933
27057 PR tree-optimization/80934
27058 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27059 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27060 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27061 (gimple_fold_builtin): Call them.
27063 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27065 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27066 as unlikely; update profile.
27068 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27070 * predict.c (force_edge_cold): Handle declaring edges impossible
27073 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27075 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27077 (try_unroll_loop_completely): Fix reporting.
27079 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27081 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27083 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27086 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27087 if given a non-constant argument for an intrinsic which requires a
27090 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27092 * profile.c (compare_freqs): New function.
27093 (branch_prob): Sort edge list.
27094 (find_spanning_tree): Assume that the list is priority sorted.
27096 2017-06-16 Richard Biener <rguenther@suse.de>
27098 PR tree-optimization/81090
27099 * passes.def (pass_record_bounds): Remove.
27100 * tree-pass.h (make_pass_record_bounds): Likewise.
27101 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27102 make_pass_record_bounds): Likewise.
27103 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27104 not free niter estimates at the beginning but at the end.
27105 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27107 2017-06-16 Richard Biener <rguenther@suse.de>
27109 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27110 initializer to workaround ICE in host GCC 4.8.
27112 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27114 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27116 (clone_inlined_nodes): Update.
27118 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27120 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27121 prefetch settings, and enable prefetching by default at -O3.
27123 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27125 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27126 Set flag_prefetch_loop_arrays according to tuning data.
27128 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27130 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27131 New tune structure.
27132 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27133 [Unrelated to main purpose of the patch] Place the pointer field last
27134 to enable type checking errors when tune structure are wrongly merged.
27135 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27136 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27137 (thunderx2t99_prefetch_tune): New tune constants.
27138 (tune_params *_tunings): Update all tunings (no functional change).
27139 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27140 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27141 from tunings structures.
27143 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27146 * ubsan.c (instrument_null): Add T argument, use it instead
27147 of computing it based on IS_LHS.
27148 (instrument_object_size): Likewise.
27149 (pass_ubsan::execute): Adjust instrument_null and
27150 instrument_object_size callers to pass gimple_get_lhs or
27151 gimple_assign_rhs1 result to it. Use instrument_null instead of
27152 calling get_base_address and instrument_mem_ref. Handle
27153 aggregate call arguments for object-size sanitization.
27155 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27157 PR tree-optimization/81089
27158 * tree-vrp.c (is_masked_range_test): Validate operands of
27161 2017-06-15 Martin Sebor <msebor@redhat.com>
27164 * dumpfile.c (dump_register): Avoid calling memset to initialize
27165 a class with a default ctor.
27166 * gcc.c (struct compiler): Remove const qualification.
27167 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27168 * hash-table.h: Ditto.
27169 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27171 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27172 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27174 * params.h (struct param_info): Make struct members non-const.
27175 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27176 with default initialization.
27177 * vec.h (vec_copy_construct, vec_default_construct): New helper
27179 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27180 with vec_copy_construct.
27181 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27182 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27183 * doc/invoke.texi (-Wclass-memaccess): Document.
27185 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27187 * emit-rtl.h (is_leaf): Update comment about local
27188 register allocator.
27190 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27193 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27194 for a variable to have a section before checking if the section has a
27196 Set section to.persistent if persistent attribute is set.
27197 Warn if .persistent attribute is used on an automatic variable.
27199 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27201 PR rtl-optimization/80474
27202 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27204 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27206 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27209 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27211 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27212 (casesi): Emit bounds checking as RTL.
27213 (casesi_internal_mips16_<mode>): Remove bounds checking.
27215 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27217 * config/xtensa/xtensa.c (xtensa_option_override): Append
27218 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27219 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27220 xtensa_doloop_hooks): Define unconditionally.
27221 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27223 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27224 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27225 for it in xtensa_option_override.
27226 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27227 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27229 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27231 * doc/cppopts.texi: Document '-' special value to -MF.
27233 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27235 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27236 (cortex_a53_fconst): Likewise.
27237 (cortex_a53_fpmul): Likewise.
27238 (cortex_a53_f_load_64): Likewise.
27239 (cortex_a53_f_load_many): Likewise.
27240 (cortex_a53_advsimd_alu): Likewise.
27241 (cortex_a53_advsimd_alu_q): Likewise.
27242 (cortex_a53_advsimd_mul): Likewise.
27243 (cortex_a53_advsimd_mul_q): Likewise.
27244 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27245 Add missing fmul, r2f_cvt and fconst cases.
27247 2017-06-14 Richard Biener <rguenther@suse.de>
27249 PR middle-end/81088
27250 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27252 (fold_binary_loc): When associating do not treat pre-existing
27253 TREE_OVERFLOW on literal constants as a reason to allow
27254 TREE_OVERFLOW on associated literal constants.
27256 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27258 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27259 (MASK_FEATURES): New macro.
27260 * config/sparc/sparc.c (sparc_option_override): Remove the special
27261 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27263 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27265 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27266 a division of 0 if non-call exceptions are enabled.
27268 2017-06-14 Andrew Pinski <apinski@cavium.com>
27269 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27272 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27273 Improve vector initialization code gen for only variable case.
27275 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27277 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27279 2017-06-14 Richard Biener <rguenther@suse.de>
27281 PR tree-optimization/81083
27282 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27285 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27287 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27288 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27289 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27290 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27291 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27292 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27294 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27296 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27297 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27299 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27301 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27303 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27305 * config/rs6000/t-rtems: Don't handle SPE.
27307 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27309 * config/rs6000/t-linux: Don't handle SPE.
27311 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27313 * config/rs6000/eabispe.h: Delete file.
27315 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27317 * config/rs6000/t-spe: Delete file.
27319 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27321 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27322 (rs6000_legitimate_offset_address_p): Return false for anything in
27323 V2SImode or V2SFmode.
27325 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27327 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27328 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27329 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27331 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27332 (rs6000_legitimate_offset_address_p): Ditto.
27333 (rs6000_emit_move): Ditto.
27334 (rs6000_init_builtins): Remove V4HI_type_node.
27336 2017-06-13 Martin Liska <mliska@suse.cz>
27339 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27340 (gate_asan): Likewise.
27341 * asan.h (asan_no_sanitize_address_p): Remove the function.
27342 (sanitize_flags_p): New function.
27343 * builtins.def: Fix coding style.
27344 * common.opt: Use renamed enum value.
27345 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27346 * doc/extend.texi: Document no_sanitize attribute.
27347 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27348 to SANITIZE_UNDEFINED_NONDEFAULT.
27349 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27350 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27351 Use sanitize_flags_p.
27352 * gimplify.c (gimplify_function_tree): Likewise.
27353 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27354 * opts.c (parse_no_sanitize_attribute): New function.
27355 (common_handle_option): Use renamed enum value.
27356 * opts.h (parse_no_sanitize_attribute): Declare.
27357 * tree.c (sanitize_flags_p): New function.
27358 * tree.h: Declared here.
27359 * tsan.c: Use sanitize_flags_p.
27360 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27361 (instrument_mem_ref): Likewise.
27362 (instrument_bool_enum_load): Likewise.
27363 (do_ubsan_in_current_function): Remove the function.
27364 (pass_ubsan::execute): Use sanitize_flags_p.
27365 * ubsan.h: Remove do_ubsan_in_current_function
27366 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27367 (dump_function_to_file): Use it here.
27369 2017-06-13 Martin Jambor <mjambor@suse.cz>
27371 PR tree-optimization/80803
27372 PR tree-optimization/81063
27373 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27374 (propagate_subaccesses_across_link): Enqueue subtree whenever
27375 necessary instead of relying on the caller.
27377 2017-06-13 Martin Jambor <mjambor@suse.cz>
27379 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27380 that have a first_link.
27381 (sort_and_splice_var_accesses): Do not check first_link before
27383 (subtree_mark_written_and_enqueue): Likewise.
27384 (propagate_all_subaccesses): Likewise and do not stop at first
27385 parent with a first_link.
27387 2017-06-13 Martin Jambor <mjambor@suse.cz>
27389 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27392 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27394 * match.pd: New pattern.
27396 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27398 * tree-vrp.c (is_masked_range_test): New function.
27399 (register_edge_assert_for): Determine ranges for
27402 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27404 PR tree-optimization/67328
27405 * fold-const.c (maskable_range_p): New function.
27406 (build_range_check): Generate bittests if possible.
27408 2017-06-13 Martin Liska <mliska@suse.cz>
27410 * gimple-pretty-print.c (dump_probability): Add new argument.
27411 (dump_edge_probability): Dump both probability and count.
27412 (dump_gimple_label): Likewise.
27413 (dump_gimple_bb_header): Likewise.
27415 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27418 * config/avr/avr-devices.c: Fix indentation.
27419 * config/avr/gen-avr-mmcu-specs.c: Dito.
27421 2017-06-13 Richard Biener <rguenther@suse.de>
27423 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27424 instead get vector type from stmt_info.
27425 (vectorizable_reduction): Adjust. Remove dead code.
27427 2017-06-13 Richard Biener <rguenther@suse.de>
27429 PR middle-end/81065
27430 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27431 case of C * (x * C2 + C3).
27432 (fold_addr_of_array_ref_difference): Properly fold index difference.
27434 2017-06-12 David S. Miller <davem@davemloft.net>
27437 * config/sparc/sparc.md (return expander): Emit frame blockage if
27438 function uses alloca.
27440 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27442 * combine.c (make_field_assignment): Check len rather than the mode
27443 precision when calling force_to_mode.
27445 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27447 Support multilibs and devices that see flash in RAM address range.
27450 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27451 (avr_mcu_t) <flash_pm_offset>: New field.
27452 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27453 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27454 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27455 (AVR_TINY_PM_OFFSET): Remove macro.
27456 * config/avr/avr.opt (-mshort-calls): New option.
27457 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27458 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27459 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27460 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27461 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27462 instead of avr_arch->have_jmp_call.
27463 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27464 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27465 avr_arch->flash_pm_offset to define.
27466 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27467 new field flash_pm_offset. Add entry for avrxmega3.
27468 (avr_texinfo): Add entry for avrxmega3.
27469 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27470 attiny212, attiny214,
27471 attiny412, attiny414, attiny416, attiny417,
27472 attiny814, attiny816, attiny817,
27473 attiny1614, attiny1616, attiny1617,
27474 attiny3214, attiny3216, attiny3217.
27475 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27476 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27477 (avr_print_operand_address) [AVR_TINY]: Same.
27478 (avr_asm_init_sections) <readonly_data_section>: Only patch
27479 callback if avr_arch->flash_pm_offset = 0.
27480 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27481 for rodata if avr_arch->flash_pm_offset != 0.
27482 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27483 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27484 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27485 (m_options): Append opt_rcall.
27486 (m_dirnames): Append dir_rcall.
27487 * config/avr/t-multilib: Regenerate.
27489 * configure.ac [target=avr]: Check whether avrxmega3 default
27490 linker description file works as needed.
27491 * configure: Regenerate.
27492 * doc/avr-mmcu.texi: Regenerate.
27493 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27494 <__AVR_ARCH__>: Document avrxmega3 and 103.
27495 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27496 <__AVR_SHORT_CALLS__>: Document it.
27497 <__AVR_PM_BASE_ADDRESS__>: Document it.
27498 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27499 (AVR Variable Attributes) <progmem>: Document this is
27500 not needed for avrxmega3.
27501 (AVR Named Address Spaces) <__flash>: Dito.
27503 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27505 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27507 2017-06-12 Doug Rupp <rupp@adacore.com>
27509 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27510 Append vxworks-stdint.h to the tm_file list.
27511 * config/vxworks-stdint.h: New file.
27513 2017-06-12 Martin Liska <mliska@suse.cz>
27515 PR tree-optimization/81041
27516 * tree-profile.c (gimple_gen_ic_func_profiler):
27517 Create an extra BB in profile-generate
27518 (gimple_gen_time_profiler): Likewise.
27520 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27522 PR tree-optimization/81003
27523 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27524 (update_range_test): Use it instead of force_gimple_operand_gsi.
27526 2017-06-12 Richard Biener <rguenther@suse.de>
27528 PR tree-optimization/81053
27529 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27530 with backedge value not defined in loop. Simplify def stmt
27533 2017-06-11 Tom de Vries <tom@codesourcery.com>
27536 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27538 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27539 nvptx_cannot_force_const_mem.
27541 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27543 * opts.c (finish_options): Move test for flag_split_stack after
27544 it has been initialized.
27546 2017-06-11 Jason Merrill <jason@redhat.com>
27548 * tree.h (id_equal): New.
27549 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27550 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27551 instead of strcmp of IDENTIFIER_POINTER.
27553 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27555 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27556 (mark_all_inlined_calls_cdtor): Fix formating.
27557 (inline_transform): Rescale profile before inlining.
27559 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27561 * cgraph.h (cgraph_edge::clone): Update prototype.
27562 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27563 (cgraph_node::create_clone): Update.
27564 (cgraph_node::create_version_clone): Update.
27565 * tree-inline.c (copy_bb): Update.
27566 (expand_call_inline): Update.
27568 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27570 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27571 factored out from ...
27572 (rs6000_emit_prologue): ... here.
27574 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27576 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27577 factored out from ...
27578 (rs6000_emit_prologue): ... here.
27580 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27582 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27584 (handle_missing_profiles): Fix computation of tp_first_run.
27585 (counts_to_freqs): Do not touch freqs when count is 0.
27587 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27589 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27592 2017-06-10 Tom de Vries <tom@codesourcery.com>
27594 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27595 attributes): Document signal effective target.
27597 2017-06-10 Tom de Vries <tom@codesourcery.com>
27599 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27600 Document effective target stack_size.
27602 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27604 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27605 to the edit_context if they can be auto-applied.
27607 2017-06-9 Ian Lance Taylor <iant@golang.org>
27609 * opts.c (finish_options): If -fsplit-stack, disable implicit
27610 -forder-blocks-and-partition.
27611 * doc/invoke.texi (Optimize Options): Document that when using
27612 -fsplit-stack -forder-blocks-and-partition is not implicitly
27615 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27617 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27618 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27619 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27620 * builtins.def (abort, trap, unreachable): Declare cold.
27621 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27622 * tree-core.h (ECF_COLD): New.
27623 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27624 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27626 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27628 * predict.c (unlikely_executed_stmt_p): Cleanup.
27630 2017-06-09 Richard Biener <rguenther@suse.de>
27632 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27633 model if the ref is always written to.
27635 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27637 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27639 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27641 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27644 2017-06-09 Tom de Vries <tom@codesourcery.com>
27647 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27648 "target cannot support label values" when encountering LABEL_REF.
27650 2017-06-09 Martin Liska <mliska@suse.cz>
27652 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27653 (gimple_gen_ic_func_profiler): Emit direct comparison
27654 of __gcov_indirect_call_callee with NULL.
27655 (gimple_gen_time_profiler): Change probability from
27656 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27658 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27660 * profile.c (edge_gcov_counts): Turn to pointer.
27661 (compute_branch_probabilities, compute_branch_probabilities): Update.
27662 (branch_prob): Do not clear edge_gcov_count.
27663 * profile.h (edge_gcov_counts): Turn to pointer.
27664 (edge_gcov_count): Update.
27666 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27668 * gimple.h (gimple_check_failed): Mark cold.
27670 2017-06-09 Richard Biener <rguenther@suse.de>
27672 PR tree-optimization/66623
27673 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27674 refactor check_reduction into two parts, properly computing
27675 whether we have to check reduction validity for outer loop
27678 2017-06-09 Richard Biener <rguenther@suse.de>
27680 PR tree-optimization/79483
27681 * graphite-scop-detection.c (order): New global.
27682 (get_order): Compute bb to order mapping that satisfies code
27683 generation constraints.
27684 (cmp_pbbs): New helper.
27685 (build_scops): Start domwalk at entry block, sort generated
27688 2017-06-09 Richard Biener <rguenther@suse.de>
27690 PR middle-end/81007
27691 * ipa-polymorphic-call.c
27692 (ipa_polymorphic_call_context::restrict_to_inner_class):
27693 Skip FIELD_DECLs with error_mark_node type.
27694 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27697 2017-06-09 Martin Liska <mliska@suse.cz>
27699 * predict.c (struct branch_predictor): New struct.
27700 (test_prediction_value_range): New test.
27701 (predict_c_tests): New function.
27702 * selftest-run-tests.c (selftest::run_tests): Run the function.
27703 * selftest.h: Declare new tests.
27705 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27708 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27709 gen_add3_insn did not fail.
27710 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27711 r0, construct that number in a temporary reg and add that reg to r0.
27712 If asked to put the result in r0 as well, fail.
27714 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27716 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27717 for early expansion of vec_eqv.
27719 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27721 PR middle-end/81005
27722 * ubsan.c (instrument_null): Avoid pointless code temporary.
27723 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27725 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27729 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27732 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27735 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27737 * predict.c (unlikely_executed_edge_p): Move ahead.
27738 (probably_never_executed_edge_p): Use it.
27740 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27742 PR middle-end/79988
27743 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27744 gimple_call_builtin_p call.
27746 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27748 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27749 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27750 rtl_check_failed_type2, rtl_check_failed_code1,
27751 rtl_check_failed_code2, rtl_check_failed_code_mode,
27752 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27753 rtvec_check_failed_bounds, rtl_check_failed_flag,
27754 _fatal_insn_not_found, _fatal_insn): Likewise.
27755 * tree.h (tree_contains_struct_check_failed,
27756 tree_check_failed, tree_not_check_failed,
27757 tree_class_check_failed, tree_range_check_failed,
27758 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27759 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27760 tree_operand_check_failed, omp_clause_check_failed,
27761 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27764 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27766 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27767 flag_branch_probabilities.
27768 * ipa-inline.c (edge_badness): Likewise.
27769 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27770 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27771 * predict.c (maybe_hot_frequency_p): Likewise.
27772 (probably_never_executed): Likewise.
27773 * sched-ebb.c (schedule_ebbs): Likewise.
27774 * sched-rgn.c (find_single_block_region): Likewise.
27775 * tracer.c (tail_duplicate): Likewise.
27777 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27779 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27780 longer requires x_flag_profile_use.
27782 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27784 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27785 instead of flag_reorder_blocks_and_partition.
27786 * dbxout.c (dbxout_function_end): Likewise.
27787 * dwarf2out.c (gen_subprogram_die): Likewise.
27788 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27789 * hw-doloop.c (reorg_loops): Likewise.
27790 * varasm.c (assemble_start_function,
27791 assemble_end_function): Likewise.
27792 (decide_function_section): Do not check for
27793 flag_reorder_blocks_and_partition.
27795 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27797 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27799 (chkp_get_hard_register_fake_addr_expr): Ditto.
27800 (chkp_build_addr_expr): Add check for hard reg case.
27801 (chkp_parse_array_and_component_ref): Ditto.
27802 (chkp_find_bounds_1): Ditto.
27803 (chkp_process_stmt): Don't generate bounds store for
27806 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27808 * predict.c (maybe_hot_bb_p): Do not check profile status.
27809 (maybe_hot_edge_p): Likewise.
27810 (probably_never_executed): Check for zero counts even if profile
27812 (unlikely_executed_edge_p): New function.
27813 (unlikely_executed_stmt_p): New function.
27814 (unlikely_executed_bb_p): New function.
27815 (set_even_probabilities): Use unlikely predicates.
27816 (combine_predictions_for_bb): Likewise.
27817 (predict_paths_for_bb): Likewise.
27818 (predict_paths_leading_to_edge): Likewise.
27819 (determine_unlikely_bbs): New function.
27820 (estimate_bb_frequencies): Use it.
27821 (compute_function_frequency): Use zero counts even if profile is
27823 * profile-count.h: Fix typo.
27825 2017-08-08 Julia Koval <julia.koval@intel.com>
27827 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27828 _mm512_mask_cvtsepi16_storeu_epi8,
27829 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27830 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27831 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27832 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27833 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27834 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27835 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27836 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27837 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27838 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27839 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27840 __builtin_ia32_pmovuswb256mem_mask,
27841 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27842 __builtin_ia32_pmovwb512mem_mask): New builtins.
27844 2017-08-08 Julia Koval <julia.koval@intel.com>
27846 PR target/73350,80862
27847 * config/i386/subst.md (round): Fix round pattern.
27848 * config/i386/i386.c (ix86_erase_embedded_rounding):
27849 Fix erasing rounding for the fixed pattern.
27851 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27853 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27855 2017-06-08 Martin Liska <mliska@suse.cz>
27857 PR gcov-profile/80911
27858 * gcov.c (block_info::block_info): New constructor.
27860 2017-06-07 Carl Love <cel@us.ibm.com>
27862 * config/rs6000/rs6000-c: The return type of the following
27863 built-in functions was implemented as int not long long. Fix sign
27864 of return value for the unsigned version of vec_mulo and vec_mule.
27865 vector unsigned long long vec_bperm (vector unsigned long long,
27866 vector unsigned char)
27867 vector signed long long vec_mule (vector signed int,
27869 vector unsigned long long vec_mule (vector unsigned int,
27870 vector unsigned int)
27871 vector signed long long vec_mulo (vector signed int,
27873 vector unsigned long long vec_mulo (vector unsigned int,
27874 vector unsigned int)
27875 * doc/extend.texi: Fix the documentation for the built-in
27878 2017-06-07 Carl Love <cel@us.ibm.com>
27881 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27884 2017-06-07 Carl Love <cel@us.ibm.com>
27886 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27887 support, Generate doublehv for signed int/float for BE case only.
27889 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27891 * doc/invoke.texi (mcx16): Rewrite.
27893 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27895 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27896 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27897 *mov<mode>_softfloat, and an anonymous splitter): Use
27898 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27900 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27902 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27904 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27905 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27906 (rs6000_debug_reg_global): Adjust.
27907 (rs6000_init_hard_regno_mode_ok): Adjust.
27908 (rs6000_dbx_register_number): Adjust.
27909 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27910 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27911 Remove SPE_ACC and SPEFSCR.
27912 (REG_ALLOC_ORDER): Ditto.
27913 (FRAME_POINTER_REGNUM): Change to 111.
27914 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27915 (REG_CLASS_NAMES): Ditto.
27916 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27917 (REGISTER_NAMES): Ditto.
27918 (ADDITIONAL_REG_NAMES): Ditto.
27919 (rs6000_reg_names): Ditto.
27920 * config/rs6000/rs6000.md: Renumber some register number
27923 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27925 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27927 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27928 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27930 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27931 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27932 Delete the SPE high registers.
27933 (REG_ALLOC_ORDER): Ditto.
27934 (enum reg_class): Remove SPE_HIGH_REGS.
27935 (REG_CLASS_NAMES): Ditto.
27936 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27937 (REGISTER_NAMES): Ditto.
27938 (rs6000_reg_names): Ditto.
27939 * doc/tm.texi.in: Remove SPE as example.
27940 * doc/tm.texi: Regenerate.
27942 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27944 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27945 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27946 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27947 * config/rs6000/rs6000.md (type): Remove "brinc".
27949 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27951 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27952 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27953 * config/rs6000/linuxspe.h: Delete file.
27954 * config/rs6000/rs6000.md: Don't include spe.md.
27955 * config/rs6000/spe.h: Delete file.
27956 * config/rs6000/spe.md: Delete file.
27957 * config/rs6000/t-rs6000: Remove spe.md.
27959 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27961 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27962 (reg_or_none500mem_operand): Delete.
27963 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27964 instead of reg_or_none500mem_operand.
27966 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27968 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27969 handling of SPE flags.
27970 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27972 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27974 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27976 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27977 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27978 paired_divv2sf3): Similar.
27979 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27980 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27981 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27983 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27984 Rename the paired_* instruction patterns.
27985 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27987 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27988 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27989 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27990 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27991 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27992 PAIRED_VECTOR_MODE.
27993 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27994 (spe_func_has_64bit_regs_p): Delete.
27995 (spe_expand_predicate_builtin): Delete.
27996 (spe_expand_evsel_builtin): Delete.
27997 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27998 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27999 (invalid_e500_subreg): Delete.
28000 (rs6000_legitimize_address): Always force_reg op2 as well, for
28001 paired single memory accesses.
28002 (rs6000_member_type_forces_blk): Delete.
28003 (rs6000_spe_function_arg): Delete.
28004 (rs6000_expand_unop_builtin): Delete SPE handling.
28005 (rs6000_expand_binop_builtin): Ditto.
28006 (spe_expand_stv_builtin): Delete.
28007 (bdesc_2arg_spe): Delete.
28008 (spe_expand_builtin): Delete.
28009 (spe_expand_predicate_builtin): Delete.
28010 (spe_expand_evsel_builtin): Delete.
28011 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28012 (spe_init_builtins): Delete.
28013 (spe_func_has_64bit_regs_p): Delete.
28014 (savres_routine_name): Delete "info" parameter. Adjust callers.
28015 (rs6000_emit_stack_reset): Ditto.
28016 (rs6000_dwarf_register_span): Delete.
28017 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28018 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28019 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28021 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28023 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28024 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28025 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28026 mulv2sf3, divv2sf3): Delete expanders.
28028 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28030 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28032 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28034 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28035 * config/rs6000/rs6000.c: Ditto.
28037 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28039 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28040 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28041 comparison_operator.
28043 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28045 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28046 * config/rs6000/rs6000.opt: Ditto.
28047 * config/rs6000/t-rtems: Ditto.
28049 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28051 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28052 TARGET_E500_SINGLE by 0, simplify.
28053 * config/rs6000/rs6000.c: Ditto.
28054 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28055 (spe_build_register_parallel): Delete.
28056 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28057 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28058 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28059 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28060 (E500_CONVERT): Delete.
28061 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28063 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28065 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28066 * config/rs6000/dfp.md: Ditto.
28067 (negdd2, *negdd2_fpr): Merge.
28068 (absdd2, *absdd2_fpr): Merge.
28069 (negtd2, *negtd2_fpr): Merge.
28070 (abstd2, *abstd2_fpr): Merge.
28071 * config/rs6000/e500.h: Delete file.
28072 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28073 TARGET_FPRS by 1 and simplify.
28074 * config/rs6000/rs6000-c.c: Ditto.
28075 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28076 TARGET_DF_SPE by 0.
28077 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28079 * config/rs6000/rs6000.md: Ditto.
28080 (floatdidf2, *floatdidf2_fpr): Merge.
28081 (move_from_CR_gt_bit): Delete.
28082 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28083 (E500_CR_IOR_COMPARE): Delete.
28084 (All patterns that require !TARGET_FPRS): Delete.
28085 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28087 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28089 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28091 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28093 * graphds.c (add_edge): Intitialize edge's attached data.
28094 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28095 pointer parameter. Call pointed function on each edge during
28096 graph traversing. Skip traversing the edge when the function
28098 (graphds_dfs, graphds_scc): Ditto.
28099 (for_each_edge): New parameter. Pass the new parameter to callback
28101 * graphds.h (skip_edge_callback): New function pointer type.
28102 (graphds_dfs, graphds_scc): New function pointer parameter.
28103 (graphds_edge_callback, for_each_edge): New parameter.
28105 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28107 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28108 out code checking if runtime alias check is possible to below ...
28109 Call the new function.
28110 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28111 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28113 2017-06-07 Marek Polacek <polacek@redhat.com>
28116 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28117 TYPE_OVERFLOW_WRAPS check.
28119 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28121 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28122 if versioning is required.
28123 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28124 peeling with the check for versioning.
28126 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28128 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28129 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28130 Set true to new parameter if new ssa variable is defined.
28131 (vect_gen_vector_loop_niters): Refactor. Set range information
28132 for the new vector loop bound variable.
28133 (vect_do_peeling): Ditto.
28135 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28137 * tree-affine.c (ssa.h): Include header file.
28138 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28139 has wrapping overflow behavior.
28141 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28143 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28145 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28147 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28148 (tree_to_aff_combination): ... here.
28150 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28152 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28153 reg_pressure model function.
28154 (ivopts_global_cost_for_size): Delete.
28155 (determine_set_costs, iv_ca_recount_cost): Call new model function
28156 ivopts_estimate_reg_pressure.
28158 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28160 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28161 expensive than udiv. Remove floating point cases from mod.
28163 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28165 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28166 Increase idiv cost.
28168 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28170 * config/aarch64/aarch64.md
28171 (copysignsf3): Fix mask generation.
28173 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28175 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28177 (class dump_manager): Add register_dumps method.
28178 * dumpfile.c: Include langhooks.h.
28179 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28180 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28181 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28182 (dump_manager::dump_register): Start with 512 entries instead of 32.
28183 (dump_manager::register_dumps): New method.
28184 * toplev.c (general_init): Instead of invoking register_dumps
28185 langhook, invoke register_dumps method on the dump manager.
28186 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28189 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28191 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28192 Say that # requires an associated define_split to exist, and that
28193 the define_split must be suitable for use after register allocation.
28195 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28197 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28198 (compute_outgoing_frequencies): Also initialize zero counts.
28199 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28200 around loops; preserve more of profile when nothing changes.
28202 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28204 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28206 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28207 * config/arm/arm-cpu-cdata.h: Regenerate.
28208 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28209 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28210 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28211 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28212 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28214 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28216 * config/arm/t-rmprofile: Likewise.
28217 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28219 2017-06-06 David S. Miller <davem@davemloft.net>
28222 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28223 blockage if function uses alloca.
28225 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28227 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28228 New "uid" fields to hold pretty-print IDs of group and ref.
28229 Memory references are now identified as <group_id>:<ref_id>
28230 instead of using [random] addresses.
28231 (dump_mem_details): Simplify, no functional change.
28232 (dump_mem_ref): Simplify and make output more concise.
28233 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28234 (find_or_create_group): Initialize group uid.
28235 (record_ref): Initialize ref uid. Improve debug output.
28236 (prune_group_by_reuse, should_issue_prefetch_p,)
28237 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28238 (mark_nontemporal_store, determine_loop_nest_reuse):
28239 Improve debug output.
28241 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28243 * dbgcnt.def (prefetch): New debug counter.
28244 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28245 (schedule_prefetches): Stop issueing prefetches if debug counter
28248 2017-06-06 Tom de Vries <tom@codesourcery.com>
28250 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28251 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28253 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28255 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28256 Use aarch64_reg_or_zero predicate for operand 4.
28257 (aarch64_compare_and_swap<mode> define_insn_and_split):
28258 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28259 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28261 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28263 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28264 (arm_compute_save_core_reg_mask): This.
28265 (thumb1_compute_save_reg_mask): Rename into ...
28266 (thumb1_compute_save_core_reg_mask): This.
28267 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28268 (arm_compute_frame_layout): Likewise.
28270 2017-06-06 Richard Biener <rguenther@suse.de>
28272 PR tree-optimization/80974
28273 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28274 keep or clear leaders SSA info.
28276 2017-06-06 Tom de Vries <tom@codesourcery.com>
28278 * config/nvptx/nvptx.c (split_mode_p): New function.
28279 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28281 2017-06-06 Tom de Vries <tom@codesourcery.com>
28283 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28285 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28288 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28291 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28293 * shrink-wrap.c (handle_simple_exit): Update profile.
28294 (try_shrink_wrapping): Upate profile.
28296 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28298 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28299 (tree_guess_outgoing_edge_probabilities): New.
28300 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28301 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28303 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28305 * ipa-split.c (split_function): Initialize return bb profile.
28307 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28309 * profile.c (compute_branch_probabilities): Also initialize
28310 EXIT_BLOCK profile.
28312 2017-06-06 Richard Biener <rguenther@suse.de>
28314 PR tree-optimization/80928
28315 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28316 (vect_analyze_loop_operations): Properly guard analysis for
28318 (vect_transform_loop): Likewise.
28319 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28320 (vect_model_induction_cost): Do not cost for pure SLP.
28321 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28322 of induction in inner loop vectorization.
28323 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28324 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28325 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28327 (vect_analyze_slp_cost_1): Cost induction.
28328 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28329 (vect_get_slp_vect_defs): Likewise.
28330 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28331 (vect_transform_stmt): Handle SLP reductions.
28332 * tree-vectorizer.h (vectorizable_induction): Adjust.
28334 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28336 * config/rs6000/rs6000.c (make_resolver_func): Update
28337 init_lowered_empty_function call.
28339 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28341 * doc/invoke.texi: Document the -fprofile-abs-path option.
28342 * common.opt (fprofile-abs-path): New option.
28343 * gcov-io.h (gcov_write_filename): Declare.
28344 * gcov-io.c (gcov_write_filename): New function.
28345 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28346 * profile.c (output_location): Likewise.
28348 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28350 * shring-wrap.c: Revert accidental commit.
28352 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28354 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28356 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28358 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28360 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28361 profile in callgraph edge.
28362 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28363 (apply_scale): Likewise.
28364 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28365 Also scale profile when inlining function with zero profile.
28366 (initialize_cfun): Update exit block profile even when it is zero.
28367 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28368 when profile is read.
28370 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28372 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28373 (CLONE_*): New constants to define the processors we can generate
28374 code for with the target_clone attribute.
28375 (rs6000_clone_map): New array to identify which clone processors
28376 the current program is running on.
28377 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28378 target_clone attribute.
28379 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28380 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28381 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28382 (cpu_expand_builtin): Add support for target_clone attribute.
28383 (rs6000_valid_attribute_p): Allow "default" attribute.
28384 (get_decl_name): New debug function to simplify printing the
28385 current function name in debugging statements.
28386 (rs6000_clone_priority): New functions to support the target_clone
28387 attribute, and be able to generate code to switch between ISA 2.05
28388 through ISA 3.0 (power6 through power9).
28389 (rs6000_compare_version_priority): Likewise.
28390 (rs6000_get_function_versions_dispatcher): Likewise.
28391 (make_resolver_func): Likewise.
28392 (add_condition_to_bb): Likewise.
28393 (dispatch_function_versions): Likewise.
28394 (rs6000_generate_version_dispatcher_body): Likewise.
28395 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28396 (fusion_gpr_load_p): Fix a spacing issue.
28397 * doc/extend.texi (Common Function Attributes): Document that the
28398 PowerPC supports the target_clone attribute.
28400 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28402 * config/arm/arm.h: explain F symbol found in description of ARM
28403 register allocation in its legend.
28405 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28407 * config/mips/frame-header-opt.c: Include profile-count.h.
28408 * config/riscv/riscv.c: Include profile-count.h
28410 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28412 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28414 (sm_set_flag_if_changed): Add bbs field.
28415 (execute_sm_if_changed_flag_set): Pass BBS.
28416 (execute_sm): Update.
28418 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28420 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28423 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28425 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28426 (peephole2): New peephole2 to emit the above.
28427 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28429 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28431 * config/aarch64/aarch64.c (define_peephole2 above
28432 *sub_<shift>_<mode>): New peephole.
28434 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28436 * config/i386/i386.c (make_resolver_func): Update.
28437 * Makefile.in: Add profile-count.h and profile-count.o
28438 * auto-profile.c (afdo_indirect_call): Update to new API.
28439 (afdo_set_bb_count): Update.
28440 (afdo_propagate_edge): Update.
28441 (afdo_propagate_circuit): Update.
28442 (afdo_calculate_branch_prob): Update.
28443 (afdo_annotate_cfg): Update.
28444 * basic-block.h: Include profile-count.h
28445 (struct edge_def): Turn count to profile_count.
28446 (struct basic_block_def): Likewie.
28447 (REG_BR_PROB_BASE): Move to profile-count.h
28448 (RDIV): Move to profile-count.h
28449 * bb-reorder.c (max_entry_count): Turn to profile_count.
28450 (find_traces): Update.
28451 (rotate_loop):Update.
28452 (connect_traces):Update.
28453 (sanitize_hot_paths):Update.
28454 * bt-load.c (migrate_btr_defs): Update.
28455 * cfg.c (RDIV): Remove.
28456 (init_flow): Use alloc_block.
28457 (alloc_block): Uninitialize count.
28458 (unchecked_make_edge): Uninitialize count.
28459 (check_bb_profile): Update.
28460 (dump_edge_info): Update.
28461 (dump_bb_info): Update.
28462 (update_bb_profile_for_threading): Update.
28463 (scale_bbs_frequencies_int): Update.
28464 (scale_bbs_frequencies_gcov_type): Update.
28465 (scale_bbs_frequencies_profile_count): New.
28466 * cfg.h (update_bb_profile_for_threading): Update.
28467 (scale_bbs_frequencies_profile_count): Declare.
28468 * cfgbuild.c (compute_outgoing_frequencies): Update.
28469 (find_many_sub_basic_blocks): Update.
28470 * cfgcleanup.c (try_forward_edges): Update.
28471 (try_crossjump_to_edge): Update.
28472 * cfgexpand.c (expand_gimple_tailcall): Update.
28473 (construct_exit_block): Update.
28474 * cfghooks.c (verify_flow_info): Update.
28475 (dump_bb_for_graph): Update.
28476 (split_edge): Update.
28477 (make_forwarder_block): Update.
28478 (duplicate_block): Update.
28479 (account_profile_record): Update.
28480 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28481 (get_estimated_loop_iterations): Update.
28482 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28483 (single_likely_exit): Update.
28484 * cfgloopmanip.c (scale_loop_profile): Update.
28486 (set_zero_probability): Update.
28487 (lv_adjust_loop_entry_edge): Update.
28488 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28489 (purge_dead_edges): Update.
28490 (rtl_account_profile_record): Update.
28491 * cgraph.c (cgraph_node::create): Uninitialize count.
28492 (symbol_table::create_edge): Uninitialize count.
28493 (cgraph_update_edges_for_call_stmt_node): Update.
28494 (cgraph_edge::dump_edge_flags): Update.
28495 (cgraph_node::dump): Update.
28496 (cgraph_edge::maybe_hot_p): Update.
28497 * cgraph.h: Include profile-count.h
28498 (create_clone), create_edge, create_indirect_edge): Update.
28499 (cgraph_node): Turn count to profile_count.
28500 (cgraph_edge0: Likewise.
28501 (make_speculative, clone): Update.
28502 (create_edge): Update.
28503 (init_lowered_empty_function): Update.
28504 * cgraphclones.c (cgraph_edge::clone): Update.
28505 (duplicate_thunk_for_node): Update.
28506 (cgraph_node::create_clone): Update.
28507 * cgraphunit.c (cgraph_node::analyze): Update.
28508 (cgraph_node::expand_thunk): Update.
28509 * final.c (dump_basic_block_info): Update.
28510 * gimple-streamer-in.c (input_bb): Update.
28511 * gimple-streamer-out.c (output_bb): Update.
28512 * graphite.c (print_global_statistics): Update.
28513 (print_graphite_scop_statistics): Update.
28514 * hsa-brig.c: Include basic-block.h.
28515 * hsa-dump.c: Include basic-block.h.
28516 * hsa-gen.c (T sum_slice): Update.
28517 (convert_switch_statements):Update.
28518 * hsa-regalloc.c: Include basic-block.h.
28519 * ipa-chkp.c (chkp_produce_thunks): Update.
28520 * ipa-cp.c (struct caller_statistics): Update.
28521 (init_caller_stats): Update.
28522 (gather_caller_stats): Update.
28523 (ipcp_cloning_candidate_p): Update.
28524 (good_cloning_opportunity_p): Update.
28525 (get_info_about_necessary_edges): Update.
28526 (dump_profile_updates): Update.
28527 (update_profiling_info): Update.
28528 (update_specialized_profile): Update.
28529 (perhaps_add_new_callers): Update.
28530 (decide_about_value): Update.
28531 (ipa_cp_c_finalize): Update.
28532 * ipa-devirt.c (struct odr_type_warn_count): Update.
28533 (struct decl_warn_count): Update.
28534 (struct final_warning_record): Update.
28535 (possible_polymorphic_call_targets): Update.
28536 (ipa_devirt): Update.
28537 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28538 * ipa-icf.c (sem_function::merge): Update.
28539 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28540 * ipa-inline.c (compute_uninlined_call_time): Update.
28541 (compute_inlined_call_time): Update.
28542 (want_inline_small_function_p): Update.
28543 (want_inline_self_recursive_call_p): Update.
28544 (edge_badness): Update.
28545 (lookup_recursive_calls): Update.
28546 (recursive_inlining): Update.
28547 (inline_small_functions): Update.
28548 (dump_overall_stats): Update.
28549 (dump_inline_stats): Update.
28550 * ipa-profile.c (ipa_profile_generate_summary): Update.
28551 (ipa_propagate_frequency): Update.
28552 (ipa_profile): Update.
28553 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28554 * ipa-utils.c (ipa_merge_profiles): Update.
28555 * loop-doloop.c (doloop_modify): Update.
28556 * loop-unroll.c (report_unroll): Update.
28557 (unroll_loop_runtime_iterations): Update.
28558 * lto-cgraph.c (lto_output_edge): Update.
28559 (lto_output_node): Update.
28560 (input_node): Update.
28561 (input_edge): Update.
28562 (merge_profile_summaries): Update.
28563 * lto-streamer-in.c (input_cfg): Update.
28564 * lto-streamer-out.c (output_cfg): Update.
28565 * mcf.c (create_fixup_graph): Update.
28566 (adjust_cfg_counts): Update.
28567 (sum_edge_counts): Update.
28568 * modulo-sched.c (sms_schedule): Update.
28569 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28570 * predict.c (maybe_hot_count_p): Update.
28571 (probably_never_executed): Update.
28572 (dump_prediction): Update.
28573 (combine_predictions_for_bb): Update.
28574 (propagate_freq): Update.
28575 (handle_missing_profiles): Update.
28576 (counts_to_freqs): Update.
28577 (rebuild_frequencies): Update.
28578 (force_edge_cold): Update.
28579 * predict.h: Include profile-count.h
28580 (maybe_hot_count_p, counts_to_freqs): UPdate.
28581 * print-rtl-function.c: Do not include cfg.h
28582 * print-rtl.c: Include basic-block.h
28583 * profile-count.c: New file.
28584 * profile-count.h: New file.
28585 * profile.c (is_edge_inconsistent): Update.
28586 (correct_negative_edge_counts): Update.
28587 (is_inconsistent): Update.
28588 (set_bb_counts): Update.
28589 (read_profile_edge_counts): Update.
28590 (compute_frequency_overlap): Update.
28591 (compute_branch_probabilities): Update; Initialize and deinitialize
28593 (branch_prob): Update.
28594 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28595 (edge_gcov_count): New.
28596 (bb_gcov_count): New.
28597 * shrink-wrap.c (try_shrink_wrapping): Update.
28598 * tracer.c (better_p): Update.
28599 * trans-mem.c (expand_transaction): Update.
28600 (ipa_tm_insert_irr_call): Update.
28601 (ipa_tm_insert_gettmclone_call): Update.
28602 * tree-call-cdce.c: Update.
28603 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28604 (gimple_duplicate_sese_tail): Update.
28605 (gimple_account_profile_record): Update.
28606 (execute_fixup_cfg): Update.
28607 * tree-inline.c (copy_bb): Update.
28608 (copy_edges_for_bb): Update.
28609 (initialize_cfun): Update.
28610 (freqs_to_counts): Update.
28611 (copy_cfg_body): Update.
28612 (expand_call_inline): Update.
28613 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28614 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28615 (try_unroll_loop_completely): Update.
28616 (try_peel_loop): Update.
28617 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28618 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28619 * tree-ssa-loop-split.c (connect_loops): Update.
28620 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28621 * tree-ssa-reassoc.c (branch_fixup): Update.
28622 * tree-ssa-tail-merge.c (replace_block_by): Update.
28623 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28624 (compute_path_counts): Update.
28625 (update_profile): Update.
28626 (recompute_probabilities): Update.
28627 (update_joiner_offpath_counts): Update.
28628 (estimated_freqs_path): Update.
28629 (freqs_to_counts_path): Update.
28630 (clear_counts_path): Update.
28631 (ssa_fix_duplicate_block_edges): Update.
28632 (duplicate_thread_path): Update.
28633 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28634 (struct switch_conv_info): Update.
28635 * tree-tailcall.c (decrease_profile): Update.
28636 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28637 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28638 * value-prof.c (check_counter): Update.
28639 (gimple_divmod_fixed_value): Update.
28640 (gimple_mod_pow2): Update.
28641 (gimple_mod_subtract): Update.
28642 (gimple_ic_transform): Update.
28643 (gimple_stringop_fixed_value): Update.
28644 * value-prof.h (gimple_ic): Update.
28646 2017-06-02 Carl Love <cel@us.ibm.com>
28648 * config/rs6000/rs6000-c: Add support for built-in functions
28649 vector double vec_doublee (vector signed int);
28650 vector double vec_doublee (vector unsigned int);
28651 vector double vec_doublee (vector float);
28652 vector double vec_doubleh (vector signed int);
28653 vector double vec_doubleh (vector unsigned int);
28654 vector double vec_doubleh (vector float);
28655 vector double vec_doublel (vector signed int);
28656 vector double vec_doublel (vector unsigned int);
28657 vector double vec_doublel (vector float);
28658 vector double vec_doubleo (vector signed int);
28659 vector double vec_doubleo (vector unsigned int);
28660 vector double vec_doubleo (vector float);.
28661 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28662 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28664 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28665 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28666 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28668 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28669 vec_doublel, vec_doubleh.
28670 * doc/extend.texi: Update the built-in documentation file for the
28671 new built-in functions.
28673 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28676 * ipa-inline-analysis.c (free_growth_caches): Set
28677 edge_removal_hook_holder to NULL after removing it.
28679 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28681 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28682 comparision with zero.
28684 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28685 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28686 for early expansion of vec_min and vec_max builtins.
28687 (builtin_function_type): Add min/max unsigned variants to those
28688 identified as having unsigned arguments.
28690 2017-06-02 Olivier Hainque <hainque@adacore.com>
28692 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28694 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28696 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28697 Use VALL_F16 iterator rather than VALL.
28699 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28701 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28702 Emit CBNZ inside loop when doing a strong exchange and comparing
28703 against zero. Generate the CC flags after the loop.
28705 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28707 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28708 (dl_section_ref): New.
28709 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28710 On AIX, append an expression to subtract the size of the
28711 section length to dl_section_ref.
28713 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28715 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28716 for early expansion of vector absolute builtins.
28718 2017-06-02 Richard Biener <rguenther@suse.de>
28720 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28721 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28723 2017-06-02 Richard Biener <rguenther@suse.de>
28725 PR tree-optimization/80948
28726 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28727 stmt order as well.
28729 2017-06-02 Richard Biener <rguenther@suse.de>
28731 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28733 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28734 for inductions as relevant.
28736 2017-06-02 Richard Biener <rguenther@suse.de>
28738 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28739 (vectorizable_induction): ... this. Remove dead code.
28741 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28743 * builtins. (expand_builtin_alloca): Remove second parameter and
28744 infer its value from the first parameter instead.
28745 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28747 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28749 PR rtl-optimization/80903
28750 * loop-doloop.c (add_test): Unshare sequence.
28752 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28754 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28756 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28758 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28760 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28761 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28762 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28763 (xlogue_layout::get_stub_rtx): Make static.
28764 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28765 (xlogue_layout::compute_stub_managed_regs): Rename to...
28766 (xlogue_layout::count_stub_managed_regs): ...this.
28767 (xlogue_layout::is_stub_managed_reg): New function.
28768 (xlogue_layout::m_stub_names): Rename to...
28769 (xlogue_layout::s_stub_names): ...this, make static.
28770 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28771 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28772 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28773 xlogue_layout::s_stub_names): Instantiate statics.
28774 (stub_managed_regs): Remove.
28775 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28776 (disable_call_ms2sysv_xlogues): Rename to...
28777 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28778 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28780 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28781 change after reload_completed.
28782 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28784 (ix86_expand_prologue): Likewise.
28785 (ix86_expand_epilogue): Likewise.
28786 (ix86_expand_split_stack_prologue): Likewise.
28787 (ix86_compute_frame_layout): Remove frame parameter ...
28788 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28789 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28791 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28792 (ix86_frame): Move from here ...
28793 * config/i386/i386.h (ix86_frame): ... to here.
28794 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28795 complete ix86_frame data structure instead. Remove some_ld_name.
28797 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28799 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28800 symbols that hold a DECL_VALUE_EXPR.
28802 2017-06-01 Martin Jambor <mjambor@suse.cz>
28804 PR tree-optimization/80898
28805 * tree-sra.c (process_subtree_disqualification): Removed.
28806 (disqualify_candidate): Do not acll
28807 process_subtree_disqualification.
28808 (subtree_mark_written_and_enqueue): New function.
28809 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28810 RHS has been disqualified and re-queue LHS if necessary. Apart
28811 from that, ignore disqualified RHS.
28813 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28815 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28816 address fetch for z10 or later.
28818 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28820 * config/arc/arc.md (tst_movb): Add guard when splitting.
28822 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28824 * config/arc/arc.c (arc_can_eliminate): Test against
28825 arc_frame_pointer_needed.
28827 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28829 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28830 to prevent store reordering.
28831 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28832 (type): Add block type.
28833 (stack_tie): Define special instruction to be used in
28836 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28838 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28839 constraint. It is not valid for the pattern.
28840 (noncommutative_binary_comparison): Likewise.
28842 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28844 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28847 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28849 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28850 be used by the reg-alloc.
28852 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28854 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28855 reg-alloc when having mul64 or mul32x16 instructions.
28856 (mulsidi3): Likewise.
28857 (umulsidi3): Likewise.
28858 (mulsi32x16): New pattern.
28859 (mulsi64): Likewise.
28860 (mulsidi64): Likewise.
28861 (umulsidi64): Likewise.
28862 (MUL32x16_REG): Define.
28863 (mul64_600): Use MUL32x16_REG.
28864 (mac64_600): Likewise.
28865 (umul64_600): Likewise.
28866 (umac64_600): Likewise.
28868 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28870 * config/arc/arc.md (mulsi3_700): Make it commutative.
28872 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28874 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28876 (*sign_extendsidi2_insn): Likewise for movstosw.
28878 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28880 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28881 the type of the input discriminant value. Convert the
28882 discriminant value of signedness vary.
28884 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28886 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28887 Add to -Wall section.
28889 2017-06-01 Richard Biener <rguenther@suse.de>
28891 PR middle-end/66313
28892 * fold-const.c (fold_plusminus_mult_expr): If the factored
28893 factor may be zero use a wrapping type for the inner operation.
28894 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28895 and handle moved defs.
28896 (process_assignment): Properly guard the unary op case. Return a
28897 tri-state indicating that moving the stmt before the call may allow
28898 to continue. Pass through to_move.
28899 (find_tail_calls): Handle moving unrelated defs before
28902 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28905 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28906 splitter result in the canonical way.
28908 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28910 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28911 also for 32bit target. Update insn attributes.
28912 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28914 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28916 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28917 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28918 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28919 (_mm_maskz_min_ss): New intrinsics.
28921 2017-05-31 Martin Liska <mliska@suse.cz>
28923 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28924 Change comment style to one we normally use.
28925 (vectorizable_reduction): Likewise.
28926 (vectorizable_induction): Likewise.
28927 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28928 (vectorizable_call): Likewise.
28929 (vectorizable_simd_clone_call): Likewise.
28930 (vectorizable_conversion): Likewise.
28931 (vectorizable_assignment): Likewise.
28932 (vectorizable_shift): Likewise.
28933 (vectorizable_operation): Likewise.
28934 (vectorizable_store): Likewise.
28935 (vectorizable_load): Likewise.
28936 * tree-vectorizer.h: Likewise.
28938 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28940 * passes.c (emergency_dump_function): New.
28941 * tree-pass.h (emergency_dump_function): Declare.
28942 * plugin.c (plugins_internal_error_function): Remove.
28943 * plugin.h (plugins_internal_error_function): Remove declaration.
28944 * toplev.c (internal_error_function): New static function. Use it...
28945 (general_init): ...here.
28947 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28949 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28950 (arc_rtx_costs): Add costs for new patterns.
28951 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28952 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28954 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28956 * tree-ssa-strlen.c (get_next_strinfo): New function.
28957 (get_stridx_plus_constant): Use it.
28958 (zero_length_string): Likewise.
28959 (adjust_related_strinfos): Likewise.
28960 (adjust_last_stmt): Likewise.
28962 2017-05-31 Richard Biener <rguenther@suse.de>
28965 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28966 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28968 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28970 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28971 loop_vinfo argument and use of dependence distance vectors.
28972 Check instead whether the two references differ only in their
28973 initial value and assume that they have the same alignment if the
28974 difference is a multiple of the vector alignment.
28975 (vect_analyze_data_refs_alignment): Update call accordingly.
28977 2017-05-31 Martin Liska <mliska@suse.cz>
28980 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28982 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28984 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28985 (create_intersect_range_checks): Move from ...
28986 * tree-data-ref.c (create_intersect_range_checks_index)
28987 (create_intersect_range_checks): ... to here.
28988 (create_runtime_alias_checks): New function factored from ...
28989 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28990 here. Call above function.
28991 * tree-data-ref.h (create_runtime_alias_checks): New function.
28993 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28995 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28996 segment length for dr_b and compute it in wide_int.
28998 2017-05-31 Richard Biener <rguenther@suse.de>
29000 PR tree-optimization/80906
29001 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29002 and pass through iv_map.
29003 (copy_bb_and_scalar_dependences): Adjust.
29004 (translate_pending_phi_nodes): Likewise.
29005 (copy_loop_close_phi_args): Handle code-generating IVs instead
29008 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29010 * diagnostic-color.c (color_dict): Add "type-diff".
29011 (parse_gcc_colors): Update comment.
29012 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29013 -fdiagnostics-show-template-tree and -fno-elide-type.
29014 (GCC_COLORS): Add type-diff to example.
29016 (-fdiagnostics-show-template-tree): New.
29017 (-fno-elide-type): New.
29018 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29019 the pp_format_decoder callback. Call any m_format_postprocessor's
29021 (pretty_printer::pretty_printer): Initialize
29022 m_format_postprocessor.
29023 (pretty_printer::~pretty_printer): Delete any
29024 m_format_postprocessor.
29025 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29026 (class format_postprocessor): New class.
29027 (struct pretty_printer::format_decoder): Document the new parameters.
29028 (struct pretty_printer::m_format_postprocessor): New field.
29029 * tree-diagnostic.c (default_tree_printer): Update for new
29030 bool and const char ** params.
29031 * tree-diagnostic.h (default_tree_printer): Likewise.
29033 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29035 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29036 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29037 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29038 rs6000_gen_cell_microcode code.
29039 (rs6000_final_prescan_insn): Delete.
29040 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29041 "warn-cell-microcode" entries.
29042 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29043 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29044 throughout. Change cc_reg_not_micro_cr0_operand to
29045 cc_reg_not_cr0_operand throughout.
29046 (*extendhi<mode>2_noload): Delete.
29047 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29048 (mwarn-cell-microcode): Delete.
29049 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29050 -mgen-cell-microcode and -mwarn-cell-microcode.
29052 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29055 * config/i386/constraints.md (Yd): New constraint.
29057 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29058 and (?Yd, r) alternatives. Update insn attributes.
29059 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29060 and (?*Yd, r) alternatives. Update insn attributes.
29061 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29063 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29065 * gimplify.c (gimplify_modify_expr): Don't create a
29066 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29069 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29071 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29073 2017-05-30 Richard Biener <rguenther@suse.de>
29075 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29076 and reduc_def fields.
29077 (STMT_VINFO_REDUC_TYPE): New define.
29078 (STMT_VINFO_REDUC_DEF): Likewise.
29079 (vect_force_simple_reduction): Adjust prototype.
29080 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29081 (vect_is_simple_reduction): Remove check_reduction argument.
29082 (vect_force_simple_reduction): Adjust and set
29083 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29084 (vectorizable_reduction): Do not re-do reduction analysis
29085 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29086 * tree-parloops.c (gather_scalar_reductions): Adjust.
29088 2017-05-30 Richard Biener <rguenther@suse.de>
29090 PR middle-end/80901
29091 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29094 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29096 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29097 Introduce unknown_misalignment parameter and remove vf.
29098 (vect_peeling_hash_get_lowest_cost):
29099 Pass unknown_misalignment parameter.
29100 (vect_enhance_data_refs_alignment):
29101 Fix unsupportable data ref treatment.
29103 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29105 * tree-vect-data-refs.c (vect_get_data_access_cost):
29106 Workaround for SLP handling.
29107 (vect_enhance_data_refs_alignment):
29108 Compute costs for doing no peeling at all, compare to the best
29109 peeling costs so far and avoid peeling if cheaper.
29111 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29113 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29114 Return peeling info and set costs to zero for unlimited cost
29116 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29117 with unknown misalignment. Compute and costs for unknown
29118 misalignment, compare them to the costs for known misalignment
29119 and choose the cheapest for peeling.
29121 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29123 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29124 (vect_get_peeling_costs_all_drs): Create function.
29125 (vect_peeling_hash_get_lowest_cost):
29126 Use vect_get_peeling_costs_all_drs.
29127 (vect_peeling_supportable): Create function.
29128 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29130 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29132 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29133 DR_HAS_NEGATIVE_STEP.
29134 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29135 (vect_enhance_data_refs_alignment): Use.
29136 (vect_duplicate_ssa_name_ptr_info): Use.
29137 * tree-vectorizer.h (dr_misalignment): Use.
29138 (known_alignment_for_access_p): Use.
29140 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29143 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29145 (has_section_name): New function.
29147 2017-05-30 Martin Liska <mliska@suse.cz>
29150 * auto-profile.c (get_function_decl_from_block): Fix
29153 2017-05-30 Richard Biener <rguenther@suse.de>
29155 PR middle-end/80876
29156 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29158 2017-05-30 Martin Liska <mliska@suse.cz>
29160 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29161 * dumpfile.h (struct dump_file_info): Remove ctors.
29163 2017-05-30 Martin Liska <mliska@suse.cz>
29165 * predict.def: Fix GNU coding style.
29167 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29169 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29170 Mark 'to' argument with ATTRIBUTE_UNUSED.
29172 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29174 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29175 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29176 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29179 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29181 * doc/install.texi (Options specification): Restore entry of
29182 --enable-sjlj-exceptions.
29184 2017-05-27 Michael Eager <eager@eagercon.com>
29187 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29189 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29191 * config/microblaze/microblaze.h
29192 (FIXED_REGISTERS): Update in macro.
29193 (CALL_USED_REGISTERS): Update in macro.
29195 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29197 * doc/install.texi: Add links to macOS binary distributions.
29199 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29203 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29205 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29207 2017-05-26 Martin Liska <mliska@suse.cz>
29209 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29211 2017-05-26 Martin Liska <mliska@suse.cz>
29213 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29214 always leading ';; '.
29215 (dump_bb_info): Likewise.
29216 (brief_dump_cfg): Likewise.
29217 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29218 * dumpfile.c: Remove usage of TDF_VERBOSE.
29219 * dumpfile.h (enum dump_kind): Likewise.
29220 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29221 * print-tree.c (debug_verbose): Remove.
29222 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29223 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29224 * tree-diagnostic.c (default_tree_printer): Replace
29225 TDF_DIAGNOSTIC with TDF_SLIM.
29227 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29229 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29230 in parameter loop, rather than loop_vinfo.
29231 (create_intersect_range_checks): Ditto.
29232 (vect_create_cond_for_alias_checks): Update call to above functions.
29234 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29236 PR tree-optimization/80815
29237 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29238 for merging runtime alias checks. Handle negative DR_STEPs.
29240 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29242 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29244 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29245 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29246 out code pruning runtime alias checks.
29247 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29248 factored out from above.
29249 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29251 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29253 (prune_runtime_alias_test_list): New decalaration.
29255 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29257 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29258 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29259 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29260 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29261 (operator==, comp_dr_with_seg_len_pair): Ditto.
29262 (vect_prune_runtime_alias_test_list): Ditto.
29264 2017-05-26 Martin Liska <mliska@suse.cz>
29267 * params.def: Bound partial-inlining-entry-probability param.
29269 2017-05-26 Marek Polacek <polacek@redhat.com>
29272 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29275 2017-05-26 Richard Biener <rguenther@suse.de>
29277 PR tree-optimization/80842
29278 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29281 2017-05-26 Richard Biener <rguenther@suse.de>
29283 PR tree-optimization/80844
29284 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29286 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29288 * doc/md.texi (Machine Constraints): Update x86 family
29289 machine constraints section to match 'config/i386/constraints.md'.
29291 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29293 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29295 2017-05-25 Nathan Sidwell <nathan@acm.org>
29297 * doc/invoke.texi (--enable-languages): Update documentation.
29299 2017-05-25 Martin Liska <mliska@suse.cz>
29301 * dumpfile.c: Add TDF_FOLDING.
29302 * dumpfile.h (enum dump_kind): Likewise.
29303 * genmatch.c (dt_simplify::gen_1): Use it.
29305 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29307 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29309 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29311 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29312 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29314 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29316 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29317 * match.pd (X == C): Rewrite it here.
29318 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29319 with_certain_nonzero_bits2): New predicates.
29320 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29322 2017-05-24 Nathan Sidwell <nathan@acm.org>
29324 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29327 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29330 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29332 * config/powerpcspe: New port. Files are copied from the rs6000
29333 port, with "rs6000" in filenames replaced by "powerpcspe".
29335 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29337 PR rtl-optimization/80754
29338 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29340 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29342 * config/sparc/sparc.md (length): Return the correct value for -mflat
29343 sibcalls to match output_sibcall.
29345 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29349 * config/rs6000/rs6000.c (struct machine_function): Add new field
29351 (rs6000_get_separate_components): Init that field, use it.
29352 (rs6000_components_for_bb): Use the field.
29354 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29356 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29358 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29360 PR middle-end/80823
29361 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29363 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29366 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29367 address against address_operand predicate.
29368 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29370 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29372 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29373 parameters passed indirectly.
29375 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29377 * config/i386/i386.md (*movdi_internal): Remove SSE4
29378 alternative 18 (?r, *v). Update insn attributes.
29379 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29380 Update insn attributes.
29381 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29382 Update insn attributes.
29383 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29384 alternative 1 (r, v). Remove isa attribute.
29385 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29386 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29387 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29389 2017-05-23 Tom de Vries <tom@codesourcery.com>
29391 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29394 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29396 * cgraphunit.c (symbol_table::process_new_functions): Update.
29397 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29398 (inline_generate_summary): Rename to ...
29399 (ipa_fn_summary_generate): ... this one.
29400 (inline_read_summary): Rename to ...
29401 (ipa_fn_summary_read): ... this one.
29402 (inline_write_summary): Rename to ...
29403 (ipa_fn_summary_write): ... this one.
29404 (inline_free_summary): Rename to ...
29405 (ipa_free_fn_summary): ... this one.
29406 (pass_data_local_fn_summary, pass_local_fn_summary,
29407 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29408 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29409 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29410 make_pass_ipa_fn_summary): New.
29411 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29412 inline_write_summary, inline_free_summary): Remove.
29413 (ipa_free_fn_summary) : New.
29414 * ipa-inline.c (ipa_inline): Update.
29415 (pass_ipa_inline): Do not generate summaries.
29416 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29418 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29419 and add pass_ipa_fn_summary.
29420 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29422 (make_pass_inline_parameters): Remove.
29424 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29426 * omp-low.c (struct omp_context): Remove "default_kind" member.
29429 * omp-offload.c (execute_oacc_device_lower): Remove the
29430 parallelism dimensions function attributes for unparallelized
29431 OpenACC kernels constructs.
29433 2017-05-23 Martin Liska <mliska@suse.cz>
29435 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29437 (cgraph_edge::make_speculative): Likewise.
29438 (cgraph_edge::resolve_speculation): Likewise.
29439 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29440 (cgraph_node::dump): Likewise.
29441 * cgraph.h: Likewise.
29442 * cgraphunit.c (analyze_functions): Likewise.
29443 (symbol_table::compile): Likewise.
29444 * ipa-cp.c (print_all_lattices): Likewise.
29445 (determine_versionability): Likewise.
29446 (initialize_node_lattices): Likewise.
29447 (ipcp_verify_propagated_values): Likewise.
29448 (estimate_local_effects): Likewise.
29449 (update_profiling_info): Likewise.
29450 (create_specialized_node): Likewise.
29451 (perhaps_add_new_callers): Likewise.
29452 (decide_about_value): Likewise.
29453 (decide_whether_version_node): Likewise.
29454 (identify_dead_nodes): Likewise.
29455 (ipcp_store_bits_results): Likewise.
29456 * ipa-devirt.c (dump_targets): Likewise.
29457 (ipa_devirt): Likewise.
29458 * ipa-icf.c (sem_item::dump): Likewise.
29459 (sem_function::equals): Likewise.
29460 (sem_variable::equals): Likewise.
29461 (sem_item_optimizer::read_section): Likewise.
29462 (sem_item_optimizer::execute): Likewise.
29463 (congruence_class::dump): Likewise.
29464 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29465 (dump_inline_summary): Likewise.
29466 (estimate_node_size_and_time): Likewise.
29467 (inline_analyze_function): Likewise.
29468 * ipa-inline-transform.c (inline_call): Likewise.
29469 * ipa-inline.c (report_inline_failed_reason): Likewise.
29470 (want_early_inline_function_p): Likewise.
29471 (edge_badness): Likewise.
29472 (update_edge_key): Likewise.
29473 (inline_small_functions): Likewise.
29474 * ipa-profile.c (ipa_profile): Likewise.
29475 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29476 (ipa_make_edge_direct_to_target): Likewise.
29477 (remove_described_reference): Likewise.
29478 (ipa_impossible_devirt_target): Likewise.
29479 (propagate_controlled_uses): Likewise.
29480 (ipa_print_node_params): Likewise.
29481 (ipcp_transform_function): Likewise.
29482 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29483 (propagate_pure_const): Likewise.
29484 * ipa-reference.c (generate_summary): Likewise.
29485 (read_write_all_from_decl): Likewise.
29486 (propagate): Likewise.
29487 (ipa_reference_read_optimization_summary): Likewise.
29488 * ipa-utils.c (ipa_merge_profiles): Likewise.
29489 * ipa.c (walk_polymorphic_call_targets): Likewise.
29490 (symbol_table::remove_unreachable_nodes): Likewise.
29491 (ipa_single_use): Likewise.
29492 * passes.c (execute_todo): Likewise.
29493 * predict.c (drop_profile): Likewise.
29494 * symtab.c (symtab_node::get_dump_name): New function.
29495 (symtab_node::dump_name): Likewise.
29496 (symtab_node::dump_asm_name): Likewise.
29497 (symtab_node::dump_references): Likewise.
29498 (symtab_node::dump_referring): Likewise.
29499 (symtab_node::dump_base): Likewise.
29500 (symtab_node::debug_symtab): Likewise.
29501 * tree-sra.c (convert_callers_for_node): Likewise.
29502 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29503 * value-prof.c (init_node_map): Likewise.
29505 2017-05-23 Martin Liska <mliska@suse.cz>
29507 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29508 and symtab_node::debug_symtab to symbol_table::debug.
29509 * cgraphunit.c (analyze_functions): Use the renamed function.
29510 (symbol_table::compile): Likewise.
29511 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29512 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29513 * passes.c (execute_todo): Likewise.
29514 * symtab.c (symbol_table::dump): New function.
29515 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29517 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29519 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29520 that nonconst implies exec.
29522 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29524 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29525 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29526 (inline_edge_summary_vec): Turn into ...
29527 (ipa_call_summaries): ... this one.
29528 (redirect_to_unreachable, edge_set_predicate,
29529 evaluate_properties_for_edge, inline_summary_alloc,
29530 reset_ipa_call_summary, reset_inline_summary,
29531 inline_summary_t::duplicate): Update.
29532 (inline_edge_duplication_hook): Turn to ...
29533 (ipa_call_summary_t::duplicate): ... this one.
29534 (inline_edge_removal_hook): Turn to ...
29535 (ipa_call_summary_t::remove): ... this one.
29536 (dump_inline_edge_summary): Turn to ...
29537 (dump_ipa_call_summary): ... this one.
29538 (estimate_function_body_sizes): Update.
29539 (inline_update_callee_summaries): Update.
29540 (remap_edge_change_prob): Update.
29541 (remap_edge_summaries): Update.
29542 (inline_merge_summary): Update.
29543 (do_estimate_edge_time): Update.
29544 (inline_generate_summary): Update.
29545 (inline_read_section): Update.
29546 (inline_read_summary): Update.
29547 (inline_free_summary): Update.
29548 * ipa-inline.c (can_inline_edge_p): Update.
29549 (compute_inlined_call_time): Update.
29550 (want_inline_small_function_p): Update.
29551 (edge_badness): Update.
29552 (early_inliner): Update.
29553 * ipa-inline.h (inline_edge_summary): Turn to ...
29554 (ipa_call_summary): ... this one.
29555 (ipa_call_summary_t): New class.
29556 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29557 (ipa_call_summaries): New.
29558 (inline_edge_summary): Remove.
29559 (estimate_edge_growth): Update.
29560 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29561 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29562 * ipa-split.c (execute_split_functions): Update.
29563 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29565 2017-05-23 Tom de Vries <tom@codesourcery.com>
29567 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29568 attributes): Document rdrand effective target.
29570 2017-05-23 Tom de Vries <tom@codesourcery.com>
29572 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29573 attributes): Sort alphabetically.
29575 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29577 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29579 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29582 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29583 V2DF/V2DI splat into two separate patterns, one that handles
29584 registers, and the other that only handles memory. Drop support
29585 for splatting from a GPR on ISA 2.07 and then splitting the
29586 splat into direct move and splat.
29587 (vsx_splat_<mode>_reg): Likewise.
29588 (vsx_splat_<mode>_mem): Likewise.
29590 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29592 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29594 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29596 PR middle-end/80809
29597 * omp-low.c (finish_taskreg_remap): New function.
29598 (finish_taskreg_scan): If unit size of ctx->record_type
29599 is non-constant, unshare the size expression and replace
29600 decls in it with possible outer var refs.
29602 PR middle-end/80809
29603 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29604 GOVD_SHARED rather than GOVD_PRIVATE with it.
29605 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29606 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29608 PR middle-end/80853
29609 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29610 as last argument to build_outer_var_ref for pointer bases of array
29611 section reductions.
29613 2017-05-19 Martin Sebor <msebor@redhat.com>
29615 * print-tree.c (print_node): Print DECL_READ_P flag.
29617 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29619 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29620 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29621 * cgraph.c: Likewise.
29622 * cgraphunit.c: Likewise.
29623 * gengtype.c: Likewise.
29624 * ipa-cp.c: Likewise.
29625 * ipa-devirt.c: Likewise.
29626 * ipa-icf.c: Likewise.
29627 * ipa-predicate.c: Likewise.
29628 * ipa-profile.c: Likewise.
29629 * ipa-prop.c: Likewise.
29630 * ipa-split.c: Likewise.
29632 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29633 edge_predicate_pool, dump_inline_hints,
29634 inline_summary::account_size_time, redirect_to_unreachable,
29635 edge_set_predicate, set_hint_predicate,
29636 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29637 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29638 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29639 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29640 ipa_call_summary_t::remove, initialize_growth_caches,
29641 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29642 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29643 mark_modified, unmodified_parm_1, unmodified_parm,
29644 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29645 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29646 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29647 will_be_nonconstant_predicate, record_modified_bb_info,
29648 get_minimal_bb, record_modified, param_change_prob,
29649 phi_result_unknown_predicate, predicate_for_phi_result,
29650 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29651 estimate_function_body_sizes, compute_inline_parameters,
29652 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29653 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29654 inline_update_callee_summaries, remap_edge_change_prob,
29655 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29656 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29657 inline_analyze_function, inline_summary_t::insert,
29658 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29659 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29660 inline_free_summary): Move to ipa-fnsummary.h
29661 (predicate_t): Remove.
29662 * ipa-fnsummary.c: New file.
29663 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29664 (enum inline_hints_vals, inline_hints, agg_position_info,
29665 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29666 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29667 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29668 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29669 inline_read_summary, inline_write_summary, inline_free_summary,
29670 inline_analyze_function, initialize_inline_failed,
29671 inline_merge_summary, inline_update_overall_summary,
29672 compute_inline_parameters): Move to ipa-fnsummary.h
29673 * ipa-fnsummary.h: New file.
29674 * ipa-inline-transform.h: Include ipa-inline.h.
29675 * ipa-inline.c: LIkewise.
29677 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29679 * ipa-inline.c (edge_badness): Use inlined_time instead of
29680 inline_summaries->get.
29682 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29684 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29686 2017-05-22 Nathan Sidwell <nathan@acm.org>
29688 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29689 (fdump-lang): Document 'raw' option.
29690 * dumpfile.h (TDI_tu): Delete.
29691 * dumpfile.c (dump_files): Remove translation-unit.
29692 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29694 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29696 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29697 command option from $(AWK) call.
29698 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29699 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29700 [FORMAT]: Remove handling of variable.
29701 * config/avr/t-multilib: Regenerate.
29703 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29705 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29707 (dump_inline_summary): Do not print self_time.
29708 (estimate_function_body_sizes): Do not set self_time.
29709 (compute_inline_parameters): Likewise.
29710 (inline_read_section, inline_write_summary): Do not stream self_time.
29711 * ipa-inline.h (inline_summary): Drop self_time.
29713 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29715 * ipa-inline-analysis.c (account_size_time): Rename to ...
29716 (inline_summary::account_size_time): ... this one.
29717 (reset_ipa_call_summary): Turn to ...
29718 (ipa_call_summary::reset): ... this one.
29719 (reset_inline_summary): Turn to ...
29720 (inline_summary::reset): ... this one.
29721 (inline_summary_t::remove): Update.
29722 (inline_summary_t::duplicate): Update.
29723 (ipa_call_summary_t::remove): Update.
29724 (dump_inline_summary): Update.
29725 (estimate_function_body_sizes): Update.
29726 (compute_inline_parameters): Update.
29727 (estimate_node_size_and_time): Update.
29728 (inline_merge_summary): Update.
29729 (inline_update_overall_summary): Update.
29730 (inline_read_section): Update.
29731 (inline_write_summary): Update.
29732 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29733 add account_size_time and reset member functions.
29734 (ipa_call_summary): Add reset function.
29735 * ipa-predicate.h (predicate::operator &): Constify.
29737 2017-05-22 Richard Biener <rguenther@suse.de>
29739 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29741 2017-05-19 Jason Merrill <jason@redhat.com>
29743 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29745 2017-05-19 Marek Polacek <polacek@redhat.com>
29748 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29749 TYPE_OVERFLOW_WRAPS checks.
29751 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29753 * tree-core.h (enum omp_clause_default_kind): Add
29754 "OMP_CLAUSE_DEFAULT_PRESENT".
29755 * tree-pretty-print.c (dump_omp_clause): Handle it.
29756 * gimplify.c (enum gimplify_omp_var_data): Add
29757 "GOVD_MAP_FORCE_PRESENT".
29758 (gimplify_adjust_omp_clauses_1): Map it to
29759 "GOMP_MAP_FORCE_PRESENT".
29760 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29762 * gimplify.c (oacc_default_clause): Clarify.
29764 2017-05-19 Nathan Sidwell <nathan@acm.org>
29766 LANG_HOOK_REGISTER_DUMPS
29767 * toplev.c (general_init): Call register dump lang hook.
29768 * doc/invoke.texi: Document -fdump-lang option family.
29769 * dumpfile.c (dump_files): Remove class dump here.
29770 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29771 * dumpfile.h (tree_dump_index): Remove TDI_class.
29772 * langhooks-def.h (lhd_register_dumps): Declare.
29773 (LANG_HOOKS_REGISTER_DUMPS): Define.
29774 (LANG_HOOKS_INITIALIZER): Add it.
29775 * langhooks.c (lhd_register_dumps): Define.
29776 * langhooks.h (struct lang_hooks): Add register_dumps.
29778 2017-05-19 Nathan Sidwell <nathan@acm.org>
29780 * context.h (context::set_passes): New.
29781 * context.c (context::context): Do not create pass manager.
29782 * toplev.c (general_init): Create pass manager here.
29784 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29786 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29787 use this splitter if two add or or instructions would also work for
29788 the constant we want to generate.
29790 2017-05-19 Richard Biener <rguenther@suse.de>
29793 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29794 predicate evaluation.
29796 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29798 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29800 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29803 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29805 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29806 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29807 (inline_edge_summary_vec): Turn into ...
29808 (ipa_call_summaries): ... this one.
29809 (redirect_to_unreachable, edge_set_predicate,
29810 evaluate_properties_for_edge, inline_summary_alloc,
29811 reset_ipa_call_summary, reset_inline_summary,
29812 inline_summary_t::duplicate): Update.
29813 (inline_edge_duplication_hook): Turn to ...
29814 (ipa_call_summary_t::duplicate): ... this one.
29815 (inline_edge_removal_hook): Turn to ...
29816 (ipa_call_summary_t::remove): ... this one.
29817 (dump_inline_edge_summary): Turn to ...
29818 (dump_ipa_call_summary): ... this one.
29819 (estimate_function_body_sizes): Update.
29820 (inline_update_callee_summaries): Update.
29821 (remap_edge_change_prob): Update.
29822 (remap_edge_summaries): Update.
29823 (inline_merge_summary): Update.
29824 (do_estimate_edge_time): Update.
29825 (inline_generate_summary): Update.
29826 (inline_read_section): Update.
29827 (inline_read_summary): Update.
29828 (inline_free_summary): Update.
29829 * ipa-inline.c (can_inline_edge_p): Update.
29830 (compute_inlined_call_time): Update.
29831 (want_inline_small_function_p): Update.
29832 (edge_badness): Update.
29833 (early_inliner): Update.
29834 * ipa-inline.h (inline_edge_summary): Turn to ...
29835 (ipa_call_summary): ... this one.
29836 (ipa_call_summary_t): New class.
29837 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29838 (ipa_call_summaries): New.
29839 (inline_edge_summary): Remove.
29840 (estimate_edge_growth): Update.
29841 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29842 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29843 * ipa-split.c (execute_split_functions): Update.
29844 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29846 2017-05-19 Richard Biener <rguenther@suse.de>
29848 PR middle-end/80764
29849 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29851 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29853 * config/rs6000/rs6000.c (struct machine_function): Add field
29854 fpr_is_wrapped_separately.
29855 (rs6000_get_separate_components): Use 64 components. Handle the
29856 new FPR components.
29857 (rs6000_components_for_bb): Handle the FPR components.
29858 (rs6000_emit_prologue_components): Handle the FPR components.
29859 (rs6000_emit_epilogue_components): Handle the FPR components.
29860 (rs6000_set_handled_components): Handle the FPR components.
29861 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29862 that are already separately shrink-wrapped.
29863 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29864 that are already separately shrink-wrapped.
29866 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29869 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29872 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29873 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29874 cases where the register allocator uses a move and an offsettable
29875 memory operation to/from a FPR register on ISA 2.06/2.07.
29876 (define_peephole2 for Altivec d-form store): Likewise.
29878 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29881 * config/i386/mmx.md (*mov<mode>_internal): Enable
29882 alternatives 11, 12, 13 and 14 also for 32bit targets.
29883 Remove alternatives 15, 16, 17 and 18.
29884 * config/i386/sse.md (vec_concatv2di): Change
29885 alternative (!x, *y) to (x, ?!*Yn).
29887 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29889 * dumpfile.h (enum dump_kind): Remove stray comma.
29891 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29893 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29894 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29895 predicate::num_conditions
29896 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29897 (CHANGED): turn into predicate::changed.
29898 (agg_position_info): Move to ipa-predicate.h
29899 (add_condition, predicate::add_clause, predicate::operator &=,
29900 predicate::or_with, predicate::evaluate, predicate::probability,
29901 dump_condition, dump_clause, predicate::dump,
29902 predicate::remap_after_duplication, predicate::remap_after_inlining,
29903 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29904 (evaluate_conditions_for_known_args): Update.
29905 (set_cond_stmt_execution_predicate): Update.
29906 * ipa-inline.h: Include ipa-predicate.h
29907 (condition, inline_param_summary, conditions, agg_position_info,
29908 predicate): Move to ipa-predicate.h
29909 * ipa-predicate.c: New file.
29910 * ipa-predicate.h: New file.
29912 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29914 * final.c (leaf_function_p): Check we are not in a sequence.
29916 2017-05-18 Martin Liska <mliska@suse.cz>
29918 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29919 * dumpfile.c (dump_register): Use new enum dump_kind.
29920 (get_dump_file_name): Likewise.
29921 (dump_enable_all): Likewise.
29922 (dump_switch_p_1): Likewise.
29923 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29924 * dumpfile.h (enum dump_kind): New enum type.
29925 (struct dump_file_info): Create constructor and
29926 format fields and comments.
29927 * passes.c (pass_manager::register_one_dump_file):
29929 * statistics.c (statistics_early_init): Likewise.
29930 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29931 TDF_TREE with TDF_SLIM.
29932 (gather_memory_references_ref): Likewise.
29934 2017-05-18 Martin Liska <mliska@suse.cz>
29936 * vec.h (struct vnull): Use it.
29938 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29940 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29941 (true_predicate, false_predicate, true_predicate_p,
29942 false_predicate_p): Remove.
29943 (single_cond_predicate, not_inlined_predicate): Turn to member function
29945 (add_condition): Update.
29946 (add_clause): Turn to...
29947 (predicate::add_clause): ... this one; update; allow passing NULL
29949 (and_predicates): Turn to ...
29950 (predicate::operator &=): ... this one.
29951 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29952 (or_predicates): Turn to ...
29953 (predicate::or_with): ... this one.
29954 (evaluate_predicate): Turn to ...
29955 (predicate::evaluate): ... this one.
29956 (predicate_probability): Turn to ...
29957 (predicate::probability): ... this one.
29958 (dump_condition): Update.
29959 (dump_predicate): Turn to ...
29960 (predicate::dump): ... this one.
29961 (account_size_time): Update.
29962 (edge_set_predicate): Update.
29963 (set_hint_predicate): UPdate.
29964 (evaluate_conditions_for_known_args): Update.
29965 (evaluate_properties_for_edge): Update.
29966 (remap_predicate_after_duplication): Turn to...
29967 (predicate::remap_after_duplication): ... this one.
29968 (remap_hint_predicate_after_duplication): Update.
29969 (inline_summary_t::duplicate): UPdate.
29970 (dump_inline_edge_summary): Update.
29971 (dump_inline_summary): Update.
29972 (set_cond_stmt_execution_predicate): Update.
29973 (set_switch_stmt_execution_predicate): Update.
29974 (compute_bb_predicates): Update.
29975 (will_be_nonconstant_expr_predicate): Update.
29976 (will_be_nonconstant_predicate): Update.
29977 (phi_result_unknown_predicate): Update.
29978 (predicate_for_phi_result): Update.
29979 (array_index_predicate): Update.
29980 (estimate_function_body_sizes): Update.
29981 (estimate_node_size_and_time): Update.
29982 (estimate_ipcp_clone_size_and_time): Update.
29983 (remap_predicate): Rename to ...
29984 (predicate::remap_after_inlining): ... this one.
29985 (remap_hint_predicate): Update.
29986 (inline_merge_summary): Update.
29987 (inline_update_overall_summary): Update.
29988 (estimate_size_after_inlining): Update.
29989 (read_predicate): Rename to ...
29990 (predicate::stream_in): ... this one.
29991 (read_inline_edge_summary): Update.
29992 (write_predicate): Rename to ...
29993 (predicate::stream_out): ... this one.
29994 (write_inline_edge_summary): Update.
29995 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29996 (clause_t): Turn to uint32_t
29997 (predicate): Turn to class; implement constructor and operators
29999 (size_time_entry): Update.
30000 (inline_summary): Update.
30001 (inline_edge_summary): Update.
30003 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30005 * fold-const.c (fold_binary_loc): Move transformation...
30006 * match.pd (C - X CMP X): ... here.
30008 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30010 * config/sparc/sparc.c (sparc_option_override): Set function
30011 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30012 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30014 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30016 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30018 2017-05-18 Marek Polacek <polacek@redhat.com>
30021 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30022 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30023 gimple_assign_load_p.
30025 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30027 PR middle-end/80692
30028 * real.c (do_compare): Give decimal_do_compare preference over
30029 comparing just the signs.
30031 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30033 * doc/md.texi (Canonicalization of Instructions): Describe the
30034 canonical form of instructions that inherently set a condition
30037 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30039 PR middle-end/80775
30040 * tree-cfg.c: Move deletion of unreachable case statements to after
30041 the merging of consecutive case labels.
30043 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30045 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30046 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30047 restoring of callee-saved registers.
30049 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30051 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30052 * config/visium/visium.c (single_set_and_flags): Likewise.
30053 * config/visium/visium.md (Substitutions): Likewise.
30055 2017-05-17 Martin Liska <mliska@suse.cz>
30057 * cfg.c: Introduce dump_flags_t type and
30058 use it instead of int type.
30060 * cfghooks.c: Likewise.
30061 * cfghooks.h (struct cfg_hooks): Likewise.
30062 * cfgrtl.c: Likewise.
30063 * cfgrtl.h: Likewise.
30064 * cgraph.c (cgraph_node::get_body): Likewise.
30065 * coretypes.h: Likewise.
30066 * domwalk.c: Likewise.
30067 * domwalk.h: Likewise.
30068 * dumpfile.c (struct dump_option_value_info): Likewise.
30069 (dump_enable_all): Likewise.
30070 (dump_switch_p_1): Likewise.
30071 (opt_info_switch_p): Likewise.
30072 * dumpfile.h (enum tree_dump_index): Likewise.
30073 (struct dump_file_info): Likewise.
30074 * genemit.c: Likewise.
30075 * generic-match-head.c: Likewise.
30076 * gengtype.c (open_base_files): Likewise.
30077 * gimple-pretty-print.c: Likewise.
30078 * gimple-pretty-print.h: Likewise.
30079 * graph.c (print_graph_cfg): Likewise.
30080 * graphite-scop-detection.c (dot_all_sese): Likewise.
30081 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30082 * loop-unroll.c (report_unroll): Likewise.
30083 * passes.c (pass_manager::register_one_dump_file): Likewise.
30084 * print-tree.c: Likewise.
30085 * statistics.c: Likewise.
30086 * tree-cfg.c: Likewise.
30087 * tree-cfg.h: Likewise.
30088 * tree-dfa.c: Likewise.
30089 * tree-dfa.h: Likewise.
30090 * tree-dump.c (dump_function): Likewise.
30091 * tree-dump.h (struct dump_info): Likewise.
30092 * tree-pretty-print.c: Likewise.
30093 * tree-pretty-print.h: Likewise.
30094 * tree-ssa-live.c: Likewise.
30095 * tree-ssa-live.h: Likewise.
30096 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30097 * tree-vect-loop.c: Likewise.
30098 * tree-vect-slp.c: Likewise.
30100 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30101 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30103 PR tree-optimization/80457
30104 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30105 of all arguments to a statement as scalar_to_vec operations.
30106 (vectorizable_call): Adjust call to vect_model_simple_cost for
30108 (vectorizable_conversion): Likewise.
30109 (vectorizable_assignment): Likewise.
30110 (vectorizable_shift): Likewise.
30111 (vectorizable_operation): Likewise.
30112 (vectorizable_comparison): Likewise.
30113 (vect_is_simple_cond): Record the def types for operands.
30114 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30115 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30116 for statement argument count.
30118 2017-05-16 Carl Love <cel@us.ibm.com>
30120 * config/rs6000/rs6000-c: Add support for built-in functions
30121 vector unsigned long long vec_bperm (vector unsigned long long,
30122 vector unsigned char)
30123 vector signed long long vec_mule (vector signed int,
30125 vector unsigned long long vec_mule (vector unsigned int,
30126 vector unsigned int)
30127 vector signed long long vec_mulo (vector signed int,
30129 vector unsigned long long vec_mulo (vector unsigned int,
30130 vector unsigned int)
30131 vector signed char vec_sldw (vector signed char,
30132 vector signed char,
30134 vector unsigned char vec_sldw (vector unsigned char,
30135 vector unsigned char,
30137 vector signed short vec_sldw (vector signed short,
30138 vector signed short,
30140 vector unsigned short vec_sldw (vector unsigned short,
30141 vector unsigned short,
30143 vector signed int vec_sldw (vector signed int,
30146 vector unsigned int vec_sldw (vector unsigned int,
30147 vector unsigned int,
30149 vector signed long long vec_sldw (vector signed long long,
30150 vector signed long long,
30152 vector unsigned long long vec_sldw (vector unsigned long long,
30153 vector unsigned long long,
30155 * config/rs6000/rs6000-c: Add support for built-in functions
30156 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30157 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30158 * doc/extend.texi: Update the built-in documentation for the
30159 new built-in functions.
30161 2017-05-16 Marek Polacek <polacek@redhat.com>
30165 * tree.c (save_expr): Don't fold the expression.
30167 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30169 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30170 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30171 and (?*y,m). Update insn attributes.
30173 2017-05-16 Martin Liska <mliska@suse.cz>
30175 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30176 flags argument of print_gimple_stmt, print_gimple_expr,
30177 print_generic_stmt and print_generic_expr.
30178 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30179 * coretypes.h: Likewise.
30180 * except.c (dump_eh_tree): Likewise.
30181 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30182 * gimple-pretty-print.h: Likewise.
30183 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30184 (backprop::push_to_worklist): Likewise.
30185 (backprop::pop_from_worklist): Likewise.
30186 (backprop::process_use): Likewise.
30187 (backprop::intersect_uses): Likewise.
30188 (note_replacement): Likewise.
30189 * gimple-ssa-store-merging.c
30190 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30191 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30192 (pass_store_merging::execute): Likewise.
30193 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30194 (ssa_base_cand_dump_callback): Likewise.
30195 (dump_incr_vec): Likewise.
30196 (replace_refs): Likewise.
30197 (replace_mult_candidate): Likewise.
30198 (create_add_on_incoming_edge): Likewise.
30199 (create_phi_basis): Likewise.
30200 (insert_initializers): Likewise.
30201 (all_phi_incrs_profitable): Likewise.
30202 (introduce_cast_before_cand): Likewise.
30203 (replace_one_candidate): Likewise.
30204 * gimplify.c (gimplify_expr): Likewise.
30205 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30206 (set_rename): Likewise.
30207 (rename_uses): Likewise.
30208 (copy_loop_phi_nodes): Likewise.
30209 (add_close_phis_to_merge_points): Likewise.
30210 (copy_loop_close_phi_args): Likewise.
30211 (copy_cond_phi_args): Likewise.
30212 (graphite_copy_stmts_from_block): Likewise.
30213 (translate_pending_phi_nodes): Likewise.
30214 * graphite-poly.c (print_pdr): Likewise.
30215 (dump_gbb_cases): Likewise.
30216 (dump_gbb_conditions): Likewise.
30217 (print_scop_params): Likewise.
30218 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30219 (build_cross_bb_scalars_use): Likewise.
30220 (gather_bbs::before_dom_children): Likewise.
30221 * hsa-dump.c (dump_hsa_immed): Likewise.
30222 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30223 (get_replacement_map): Likewise.
30224 * ipa-inline-analysis.c (dump_condition): Likewise.
30225 (estimate_function_body_sizes): Likewise.
30226 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30227 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30228 * ipa-prop.c (ipa_dump_param): Likewise.
30229 (ipa_print_node_jump_functions_for_edge): Likewise.
30230 (ipa_modify_call_arguments): Likewise.
30231 (ipa_modify_expr): Likewise.
30232 (ipa_dump_param_adjustments): Likewise.
30233 (ipa_dump_agg_replacement_values): Likewise.
30234 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30235 * ipa-pure-const.c (check_stmt): Likewise.
30236 (pass_nothrow::execute): Likewise.
30237 * ipa-split.c (execute_split_functions): Likewise.
30238 * omp-offload.c (dump_oacc_loop_part): Likewise.
30239 (dump_oacc_loop): Likewise.
30240 * trans-mem.c (tm_log_emit): Likewise.
30241 (tm_memopt_accumulate_memops): Likewise.
30242 (dump_tm_memopt_set): Likewise.
30243 (dump_tm_memopt_transform): Likewise.
30244 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30245 (print_loop): Likewise.
30246 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30247 (chkp_gather_checks_info): Likewise.
30248 (chkp_get_check_result): Likewise.
30249 (chkp_remove_check_if_pass): Likewise.
30250 (chkp_use_outer_bounds_if_possible): Likewise.
30251 (chkp_reduce_bounds_lifetime): Likewise.
30252 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30253 (chkp_mark_completed_bounds): Likewise.
30254 (chkp_register_incomplete_bounds): Likewise.
30255 (chkp_mark_invalid_bounds): Likewise.
30256 (chkp_maybe_copy_and_register_bounds): Likewise.
30257 (chkp_build_returned_bound): Likewise.
30258 (chkp_get_bound_for_parm): Likewise.
30259 (chkp_build_bndldx): Likewise.
30260 (chkp_get_bounds_by_definition): Likewise.
30261 (chkp_generate_extern_var_bounds): Likewise.
30262 (chkp_get_bounds_for_decl_addr): Likewise.
30263 * tree-chrec.c (chrec_apply): Likewise.
30264 * tree-data-ref.c (dump_data_reference): Likewise.
30265 (dump_subscript): Likewise.
30266 (dump_data_dependence_relation): Likewise.
30267 (analyze_overlapping_iterations): Likewise.
30268 * tree-inline.c (expand_call_inline): Likewise.
30269 (tree_function_versioning): Likewise.
30270 * tree-into-ssa.c (dump_defs_stack): Likewise.
30271 (dump_currdefs): Likewise.
30272 (dump_names_replaced_by): Likewise.
30273 (dump_update_ssa): Likewise.
30274 (update_ssa): Likewise.
30275 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30276 * tree-parloops.c (build_new_reduction): Likewise.
30277 (try_create_reduction_list): Likewise.
30278 (ref_conflicts_with_region): Likewise.
30279 (oacc_entry_exit_ok_1): Likewise.
30280 (oacc_entry_exit_single_gang): Likewise.
30281 * tree-pretty-print.h: Likewise.
30282 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30283 (get_scalar_evolution): Likewise.
30284 (add_to_evolution): Likewise.
30285 (get_loop_exit_condition): Likewise.
30286 (analyze_evolution_in_loop): Likewise.
30287 (analyze_initial_condition): Likewise.
30288 (analyze_scalar_evolution): Likewise.
30289 (instantiate_scev): Likewise.
30290 (number_of_latch_executions): Likewise.
30291 (gather_chrec_stats): Likewise.
30292 (final_value_replacement_loop): Likewise.
30293 (scev_const_prop): Likewise.
30294 * tree-sra.c (dump_access): Likewise.
30295 (disqualify_candidate): Likewise.
30296 (create_access): Likewise.
30297 (reject): Likewise.
30298 (maybe_add_sra_candidate): Likewise.
30299 (create_access_replacement): Likewise.
30300 (analyze_access_subtree): Likewise.
30301 (analyze_all_variable_accesses): Likewise.
30302 (sra_modify_assign): Likewise.
30303 (initialize_constant_pool_replacements): Likewise.
30304 (find_param_candidates): Likewise.
30305 (decide_one_param_reduction): Likewise.
30306 (replace_removed_params_ssa_names): Likewise.
30307 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30308 * tree-ssa-copy.c (dump_copy_of): Likewise.
30309 (copy_prop_visit_cond_stmt): Likewise.
30310 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30311 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30312 (record_equivalences_from_stmt): Likewise.
30313 * tree-ssa-dse.c (compute_trims): Likewise.
30314 (delete_dead_call): Likewise.
30315 (delete_dead_assignment): Likewise.
30316 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30317 (forward_propagate_into_cond): Likewise.
30318 (pass_forwprop::execute): Likewise.
30319 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30320 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30322 (move_computations_worker): Likewise.
30323 (execute_sm): Likewise.
30324 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30325 (remove_exits_and_undefined_stmts): Likewise.
30326 (remove_redundant_iv_tests): Likewise.
30327 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30328 (adjust_iv_update_pos): Likewise.
30329 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30330 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30331 (value_replacement): Likewise.
30332 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30333 * tree-ssa-pre.c (print_pre_expr): Likewise.
30334 (get_representative_for): Likewise.
30335 (create_expression_by_pieces): Likewise.
30336 (insert_into_preds_of_block): Likewise.
30337 (eliminate_insert): Likewise.
30338 (eliminate_dom_walker::before_dom_children): Likewise.
30339 (eliminate): Likewise.
30340 (remove_dead_inserted_code): Likewise.
30341 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30342 * tree-ssa-reassoc.c (get_rank): Likewise.
30343 (eliminate_duplicate_pair): Likewise.
30344 (eliminate_plus_minus_pair): Likewise.
30345 (eliminate_not_pairs): Likewise.
30346 (undistribute_ops_list): Likewise.
30347 (eliminate_redundant_comparison): Likewise.
30348 (update_range_test): Likewise.
30349 (optimize_range_tests_var_bound): Likewise.
30350 (optimize_vec_cond_expr): Likewise.
30351 (rewrite_expr_tree): Likewise.
30352 (rewrite_expr_tree_parallel): Likewise.
30353 (linearize_expr): Likewise.
30354 (break_up_subtract): Likewise.
30355 (linearize_expr_tree): Likewise.
30356 (attempt_builtin_powi): Likewise.
30357 (attempt_builtin_copysign): Likewise.
30358 (transform_stmt_to_copy): Likewise.
30359 (transform_stmt_to_multiply): Likewise.
30360 (dump_ops_vector): Likewise.
30361 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30362 (print_scc): Likewise.
30363 (set_ssa_val_to): Likewise.
30364 (visit_reference_op_store): Likewise.
30365 (visit_use): Likewise.
30366 (sccvn_dom_walker::before_dom_children): Likewise.
30367 (run_scc_vn): Likewise.
30368 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30370 (expr_hash_elt::print): Likewise.
30371 (const_and_copies::pop_to_marker): Likewise.
30372 (const_and_copies::record_const_or_copy_raw): Likewise.
30373 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30374 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30375 (dump_predicates): Likewise.
30376 (find_uninit_use): Likewise.
30377 (warn_uninitialized_phi): Likewise.
30378 (pass_late_warn_uninitialized::execute): Likewise.
30379 * tree-ssa.c (verify_vssa): Likewise.
30380 (verify_ssa): Likewise.
30381 (maybe_optimize_var): Likewise.
30382 * tree-vrp.c (dump_value_range): Likewise.
30383 (dump_all_value_ranges): Likewise.
30384 (dump_asserts_for): Likewise.
30385 (register_edge_assert_for_2): Likewise.
30386 (vrp_visit_cond_stmt): Likewise.
30387 (vrp_visit_switch_stmt): Likewise.
30388 (vrp_visit_stmt): Likewise.
30389 (vrp_visit_phi_node): Likewise.
30390 (simplify_cond_using_ranges_1): Likewise.
30391 (fold_predicate_in): Likewise.
30392 (evrp_dom_walker::before_dom_children): Likewise.
30393 (evrp_dom_walker::push_value_range): Likewise.
30394 (evrp_dom_walker::pop_value_range): Likewise.
30395 (execute_early_vrp): Likewise.
30397 2017-05-16 Richard Biener <rguenther@suse.de>
30399 * dwarf2out.c (loc_list_from_tree_1): Do not create
30400 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30402 2017-05-16 Richard Biener <rguenther@suse.de>
30404 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30406 (note_variable_value_in_expr): If we resolved the decl ref
30407 do not push to the stack.
30409 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30411 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30412 operations in fast-math mode.
30413 (vaddq_f16): Likewise.
30414 (vmul_f16): Likewise.
30415 (vmulq_f16): Likewise.
30416 (vsub_f16): Likewise.
30417 (vsubq_f16): Likewise.
30418 * config/arm/neon.md (add<mode>3): New.
30420 (fma:<VH:mode>3): New. Also remove outdated comment.
30423 2017-05-16 Martin Liska <mliska@suse.cz>
30427 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30428 (odr_types_equivalent_p): Likewise.
30430 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30432 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30434 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30437 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30438 non-interunit SSE move alternatives with '?'.
30439 (zero-extendsidi peephole2): New peephole to skip intermediate
30440 general register in SSE zero-extend sequence.
30442 2017-05-15 Jeff Law <law@redhat.com>
30444 * reorg.c (relax_delay_slots): Create a new variable to hold
30445 the temporary target rather than clobbering TARGET_LABEL.
30447 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30448 missing argument to extract_bit_field call.
30449 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30451 2017-05-15 Martin Liska <mliska@suse.cz>
30454 * gcc.c (process_command): Do not allow empty argument of -o option.
30456 2017-05-15 Renlin Li <renlin.li@arm.com>
30458 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30459 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30460 * config/aarch64/constraints.md (Usf): Add long call check.
30461 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30462 (call_value): Likewise.
30463 (sibcall): Likewise.
30464 (sibcall_value): Likewise.
30466 (call_value_insn): New.
30467 (sibcall_insn): Update rtx pattern.
30468 (sibcall_value_insn): Likewise.
30469 (call_internal): Remove.
30470 (call_value_internal): Likewise.
30471 (sibcall_internal): Likewise.
30472 (sibcall_value_internal): Likewise.
30473 (call_reg): Likewise.
30474 (call_symbol): Likewise.
30475 (call_value_reg): Likewise.
30476 (call_value_symbol): Likewise.
30478 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30481 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30483 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30485 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30486 compatible with CCGOCmode and with CCZmode.
30488 2017-05-14 Martin Sebor <msebor@redhat.com>
30490 PR middle-end/77671
30491 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30492 (gimple_fold_builtin_snprintf): Same.
30493 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30494 (gimple_fold_builtin_snprintf): Same.
30495 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30496 of character types.
30497 (is_call_safe): New function.
30498 (try_substitute_return_value): Call it.
30499 (try_simplify_call): New function.
30500 (pass_sprintf_length::handle_gimple_call): Call it.
30502 2017-05-14 Martin Sebor <msebor@redhat.com>
30504 PR middle-end/80669
30505 * builtins.c (expand_builtin_stpncpy): Simplify.
30507 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30509 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30510 * config/i386/i386.h
30511 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30512 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30513 (struct machine_function): Add new members call_ms2sysv,
30514 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30515 (struct machine_frame_state): New fields sp_realigned and
30516 sp_realigned_offset.
30517 * config/i386/i386.c
30518 (enum xlogue_stub): New enum.
30519 (enum xlogue_stub_sets): New enum.
30520 (class xlogue_layout): New class.
30521 (struct ix86_frame): New fields stack_realign_allocate_offset,
30522 stack_realign_offset and outlined_save_offset. Modify comments to
30523 detail stack layout when using out-of-line stubs.
30524 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30525 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30526 -mcall-ms2sysv-xlogues.
30527 (stub_managed_regs): New static variable.
30528 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30529 registers managed by out-of-line stub.
30530 (disable_call_ms2sysv_xlogues): New function.
30531 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30532 m->call_ms2sysv when appropriate and compute frame layout for
30534 (sp_valid_at, fp_valid_at): New inline functions.
30535 (choose_basereg): New function.
30536 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30538 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30539 Use align parameter of choose_baseaddr to generated aligned SSE movs
30541 (pro_epilogue_adjust_stack): Modify to track
30542 machine_frame_state::sp_realigned.
30543 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30544 (ix86_nsaved_sseregs): Likewise.
30545 (ix86_emit_save_regs): Likewise.
30546 (ix86_emit_save_regs_using_mov): Likewise.
30547 (ix86_emit_save_sse_regs_using_mov): Likewise.
30548 (get_scratch_register_on_entry): Likewise.
30549 (gen_frame_set): New function.
30550 (gen_frame_load): Likewise.
30551 (gen_frame_store): Likewise.
30552 (emit_outlined_ms2sysv_save): Likewise.
30553 (emit_outlined_ms2sysv_restore): Likewise.
30554 (ix86_expand_prologue): Modify stack re-alignment code and call
30555 emit_outlined_ms2sysv_save when appropriate.
30556 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30557 parameter rtx_insn *insn, which allows the function to be used to only
30558 generate the notes.
30559 (ix86_expand_epilogue): Modify validity checks of frame and stack
30560 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30561 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30562 * config/i386/predicates.md
30563 (save_multiple): New predicate.
30564 (restore_multiple): Likewise.
30565 * config/i386/sse.md
30566 (save_multiple<mode>): New pattern.
30567 (save_multiple_realign<mode>): Likewise.
30568 (restore_multiple<mode>): Likewise.
30569 (restore_multiple_and_return<mode>): Likewise.
30570 (restore_multiple_leave_return<mode>): Likewise.
30571 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30573 2017-05-14 Julia Koval <julia.koval@intel.com>
30575 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30576 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30577 (__builtin_ia32_xsetbv): New builtins.
30578 * config/i386/i386.c (ix86_expand_special_args_builtin):
30580 (ix86_expand_builtin): Special expand for new intrinsics.
30581 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30582 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30583 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30585 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30587 * cfganal.c (inverted_post_order_compute): Change argument type
30589 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30590 * df-core.c (rest_of_handle_df_initialize): Adjust.
30591 (rest_of_handle_df_finish): Likewise.
30592 (df_analyze_1): Likewise.
30593 (df_analyze): Likewise.
30594 (loop_inverted_post_order_compute): Change argument to be a vec *.
30595 (df_analyze_loop): Adjust.
30596 (df_get_n_blocks): Likewise.
30597 (df_get_postorder): Likewise.
30598 * df.h (struct df_d): Change field to be a vec.
30599 * lcm.c (compute_laterin): Adjust.
30600 (compute_available): Likewise.
30601 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30602 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30603 * tree-ssa-pre.c (compute_antic): Likewise.
30605 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30607 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30608 (depth_first_search::depth_first_search): Change structure init
30609 function to this constructor.
30610 (depth_first_search::add_bb): Rename function to this member.
30611 (depth_first_search::execute): Likewise.
30612 (flow_dfs_compute_reverse_finish): Adjust.
30614 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30616 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30617 (longest_simple_path): Likewise.
30618 * shrink-wrap.c (spread_components): Likewise.
30619 (disqualify_problematic_components): Likewise.
30620 (emit_common_heads_for_components): Likewise.
30621 (emit_common_tails_for_components): Likewise.
30622 (insert_prologue_epilogue_for_components): Likewise.
30624 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30626 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30629 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30631 * df-core.c (df_set_blocks): Start using auto_bitmap.
30632 (df_compact_blocks): Likewise.
30633 * df-problems.c (df_rd_confluence_n): Likewise.
30634 * df-scan.c (df_insn_rescan_all): Likewise.
30635 (df_process_deferred_rescans): Likewise.
30636 (df_update_entry_block_defs): Likewise.
30637 (df_update_exit_block_uses): Likewise.
30638 (df_entry_block_bitmap_verify): Likewise.
30639 (df_exit_block_bitmap_verify): Likewise.
30640 (df_scan_verify): Likewise.
30641 * lra-constraints.c (lra_constraints): Likewise.
30642 (undo_optional_reloads): Likewise.
30643 (lra_undo_inheritance): Likewise.
30644 * lra-remat.c (calculate_gen_cands): Likewise.
30645 (do_remat): Likewise.
30646 * lra-spills.c (assign_spill_hard_regs): Likewise.
30647 (spill_pseudos): Likewise.
30648 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30649 (bitmap_set_subtract_values): Likewise.
30651 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30653 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30654 management with auto_bitmap.
30655 (fix_inter_tick): Likewise.
30656 (fix_recovery_deps): Likewise.
30657 * ira.c (add_store_equivs): Likewise.
30658 (find_moveable_pseudos): Likewise.
30659 (split_live_ranges_for_shrink_wrap): Likewise.
30660 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30661 (rtx_reuse_manager::seen_def_p): Likewise.
30662 (rtx_reuse_manager::set_seen_def): Likewise.
30663 * print-rtl.h (class rtx_reuse_manager): Likewise.
30665 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30667 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30669 (migrate_btr_def): Likewise.
30670 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30671 * df-core.c (loop_post_order_compute): Likewise.
30672 (loop_inverted_post_order_compute): Likewise.
30673 * hsa-common.h: Likewise.
30674 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30675 * init-regs.c (initialize_uninitialized_regs): Likewise.
30676 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30677 (inline_small_functions): Likewise.
30678 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30679 * ira.c (combine_and_move_insns): Likewise.
30680 (build_insn_chain): Likewise.
30681 * loop-invariant.c (find_invariants): Likewise.
30682 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30683 * predict.c (tree_predict_by_opcode): Likewise.
30684 (predict_paths_leading_to): Likewise.
30685 (predict_paths_leading_to_edge): Likewise.
30686 (estimate_loops_at_level): Likewise.
30687 (estimate_loops): Likewise.
30688 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30689 (spread_components): Likewise.
30690 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30691 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30692 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30693 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30694 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30695 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30696 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30697 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30698 (mark_threaded_blocks): Likewise.
30699 (thread_through_all_blocks): Likewise.
30700 * tree-ssa.c (verify_ssa): Likewise.
30701 (execute_update_addresses_taken): Likewise.
30702 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30704 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30706 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30708 (post_order_compute): Likewise.
30709 (inverted_post_order_compute): Likewise.
30710 (pre_and_rev_post_order_compute_fn): Likewise.
30712 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30714 * genrecog.c (int_set::int_set): Explicitly construct our
30715 auto_vec base class.
30716 * vec.h (auto_vec::auto_vec): New constructor.
30718 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30720 * bitmap.h (class auto_bitmap): New constructor taking
30721 bitmap_obstack * argument.
30723 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30725 * bitmap.h (class auto_bitmap): Change type of m_bits to
30726 bitmap_head, and adjust ctor / dtor and member operators.
30728 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30730 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30731 when returned register mode doesn't match original mode.
30733 2017-05-12 Jeff Law <law@redhat.com>
30734 Jakub Jelinek <jakub@redhat.com>
30736 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30737 we look for cc setter after the compare-elim changes.
30738 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30739 within the vector to match what compare-elim now expects.
30740 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30741 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30743 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30744 after the compare-elim changes.
30745 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30746 the vector to match what compare-elim now expects.
30747 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30748 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30749 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30750 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30751 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30753 * config/visium/visium.c (single_set_and_flags): Fix where
30754 we look for cc setter after the compare-elim changes.
30755 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30756 with the vector to match what compare-elim now expects.
30757 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30758 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30759 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30760 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30761 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30762 (neg<mode>2_insn_set_overflow): Likewise.
30764 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30766 PR middle-end/79794
30767 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30768 maybe_expand_insn call, set ops[0].target. If still set after call,
30769 set alt_rtl. Add extra arg to recursive calls.
30770 (extract_bit_field): Add alt_rtl argument. Pass to
30772 * expmed.h (extract_bit_field): Fix prototype.
30773 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30774 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30775 to extract_bit_field_calls.
30776 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30777 Pass alt_rtl to extract_bit_field calls.
30778 * calls.c (store_unaligned_arguments_into_psuedos)
30779 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30780 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30782 * optabs.h (struct expand_operand): Add target bitfield.
30784 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30786 * compare-elim.c (try_eliminate_compare): Canonicalize
30787 operation with embedded compare to
30788 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30789 (set (reg) (operation)].
30791 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30793 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30796 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30797 cost of adding a carry flag for ADC instruction.
30798 [case MINUS]: Ignore the cost of subtracting a carry flag
30799 for SBB instruction.
30801 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30803 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30805 * config/rs6000/bmiintrin.h: New file.
30806 * config/rs6000/bmi2intrin.h: New file.
30807 * config/rs6000/x86intrin.h: New file.
30809 2017-05-12 Jeff Law <law@redhat.com>
30811 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30814 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30816 PR middle-end/80707
30817 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30819 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30821 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30822 early expansion of vector divide builtins.
30823 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30824 builtins identified as having unsigned arguments.
30826 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30828 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30829 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30830 expansion of vector logical operations (and, andc, or, xor,
30833 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30835 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30836 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30838 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30840 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30841 early GIMPLE expansion of vector multiplies.
30843 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30845 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30846 TARGET_HAVE_MOVT conditional.
30847 (movt splitter): Likewise.
30849 2017-05-12 Richard Biener <rguenther@suse.de>
30851 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30852 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30853 Fold all stmts not inplace.
30855 2017-05-12 Richard Biener <rguenther@suse.de>
30857 PR tree-optimization/80713
30858 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30859 inserted_exprs bit for not removed stmts.
30861 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30863 PR middle-end/69921
30864 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30865 parallelized" attribute for parallelized OpenACC kernels.
30866 * omp-offload.c (execute_oacc_device_lower): Use it.
30868 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30869 Set "oacc kernels" attribute.
30870 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30871 parameter. Adjust all users.
30872 (oacc_fn_attrib_kernels_p): Remove function.
30873 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30874 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30875 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30876 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30877 assert "oacc kernels" attribute is set.
30879 2017-05-11 Carl Love <cel@us.ibm.com>
30881 * config/rs6000/rs6000-c: Add support for built-in functions
30882 vector unsigned char vec_popcnt (vector signed char)
30883 vector unsigned char vec_popcnt (vector unsigned char)
30884 vector unsigned short vec_popcnt (vector signed short)
30885 vector unsigned short vec_popcnt (vector unsigned short)
30886 vector unsigned int vec_popcnt (vector signed int)
30887 vector unsigned int vec_popcnt (vector unsigned int)
30888 vector unsigned long long vec_popcnt (vector signed long long)
30889 vector unsigned long long vec_popcnt (vector unsigned long long)
30890 vector signed long long vec_slo (vector signed long long,
30891 vector signed char)
30892 vector signed long long vec_slo (vector signed long long,
30893 vector unsigned char)
30894 vector unsigned long long vec_slo (vector unsigned long long,
30895 vector signed char)
30896 vector unsigned long long vec_slo (vector unsigned long long,
30897 vector unsigned char)
30898 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30899 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30900 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30901 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30902 * doc/extend.texi: Update the built-in documentation file for the
30903 new built-in functions.
30905 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30907 * attribs.h (sorted_attr_string): Move machine independent
30908 functions for target clone support from the i386 port to common
30909 code. Rename ix86_function_versions to common_function_versions.
30910 Rename make_name to make_unique_name.
30911 (common_function_versions): Likewise.
30912 (make_unique_name): Likewise.
30913 (make_dispatcher_decl): Likewise.
30914 (is_function_default_version): Likewise.
30915 * attribs.c (attr_strcmp): Likewise.
30916 (sorted_attr_string): Likewise.
30917 (common_function_versions): Likewise.
30918 (make_unique_name): Likewise.
30919 (make_dispatcher_decl): Likewise.
30920 (is_function_default_version): Likewise.
30921 * config/i386/i386.c (attr_strcmp): Likewise.
30922 (sorted_attr_string): Likewise.
30923 (ix86_function_versions): Likewise.
30924 (make_name): Likewise.
30925 (make_dispatcher_decl): Likewise.
30926 (is_function_default_version): Likewise.
30927 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30929 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30932 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30933 Account for direct move costs for vec_construct of integer
30936 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30939 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30940 (UNSPEC_STX_ATOMIC): Ditto.
30941 (loaddi_via_sse): New insn.
30942 (storedi_via_sse): Ditto.
30943 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30944 Update corresponding peephole2 patterns.
30945 (atomic_storedi_fpu): Ditto.
30947 2017-05-11 Julia Koval <julia.koval@intel.com>
30949 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30950 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30952 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30953 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30954 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30956 2017-05-11 Nathan Sidwell <nathan@acm.org>
30958 * graphite-poly.c: Include dumpfile.h.
30960 * dumpfle.h (dump_function): Declare here ...
30961 * tree-dump.h (dump_function): ... not here.
30962 * dumpfile.c: #include tree-cfg.h.
30963 (dump_function): Move here from ...
30964 * tree-dump.c (dump_function): ... here.
30965 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30966 * graphite-poly.c: Don't include tree-dump.h.
30967 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30968 * print-tree.c: Likewise.
30969 * stor-layout.c: Likewise.
30970 * tree-nested.c: Likewise.
30972 * dumpfile.c (dump_start): Use TDF_FLAGS.
30973 (dump_enable_all): Fix TDF_KIND check thinko.
30975 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30977 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30978 array entries to represent two legal parameterizations of the
30979 overloaded __builtin_cmpb function, as represented by the
30980 P6_OV_BUILTIN_CMPB constant.
30981 (altivec_resolve_overloaded_builtin): Add special case handling
30982 for the __builtin_cmpb function, as represented by the
30983 P6_OV_BUILTIN_CMPB constant.
30984 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30985 (BU_P6_64BIT_2): New macro.
30986 (BU_P6_OVERLOAD_2): New macro
30987 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30988 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30989 (CMPB): Add overload support to represent both 32-bit and 64-bit
30990 compare-bytes function.
30991 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30992 support for TARGET_CMPB.
30993 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30994 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30995 documentation of the __builtin_cmpb overloaded built-in function.
30997 2017-05-11 Richard Biener <rguenther@suse.de>
30999 PR tree-optimization/80705
31000 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31001 bases are not vectorizable.
31003 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31005 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31006 when counting register pressure.
31008 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31010 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31011 (iv_ca_more_deps): Renamed to ...
31012 (iv_ca_compare_deps): ... this.
31013 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31015 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31017 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31019 (determine_group_iv_costs): ... here.
31020 (find_inv_vars_cb): Record inv var if it's not recorded before.
31022 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31024 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31025 (get_shiftadd_cost): Ditto.
31027 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31029 * tree-ssa-address.c: Include header file.
31030 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31032 (add_to_parts): Refactor.
31033 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31034 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31037 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31039 PR tree-optimization/53090
31040 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31042 (extract_cond_operands): Detect condition with IV on both sides
31043 and return COMP_IV_EXPR_2.
31044 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31045 (rewrite_use_compare): Simplify by removing call to function
31046 extract_cond_operands.
31048 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31050 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31051 (extract_cond_operands): Detect condition comparing against non-
31052 invariant bound and return appropriate enum value.
31053 (find_interesting_uses_cond): Update use of extract_cond_operands.
31054 Handle its return value accordingly.
31055 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31057 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31059 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31060 nonlinear iv_use computation in loop invariant sensitive way.
31062 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31064 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31065 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31067 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31069 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31070 (dump_cand): Support iv_cand.inv_exprs.
31071 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31073 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31076 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31078 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31080 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31081 as local function. Include necessary header files.
31082 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31084 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31086 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31088 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31090 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31091 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31092 RSHIFT_EXPR and BIT_NOT_EXPR.
31094 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31096 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31097 (adjust_setup_cost): New parameter supporting round up adjustment.
31098 (struct address_cost_data): Delete.
31099 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31100 (split_address_cost, ptr_difference_cost): Delete.
31101 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31102 (struct ainc_cost_data): New struct.
31103 (get_address_cost_ainc): New function.
31104 (get_address_cost, get_computation_cost): Reimplement.
31105 (determine_group_iv_cost_address): Record inv_expr for all uses of
31107 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31108 (iv_ca_has_deps): Reimplemented to ...
31109 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31111 (iv_ca_extend): Call iv_ca_more_deps.
31113 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31115 * tree-ssa-address.c (struct mem_address): Move to header file.
31116 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31117 * tree-ssa-address.h (struct mem_address): Move from C file.
31118 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31120 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31122 * tree-affine.h (aff_combination_type): New interface.
31123 (aff_combination_zero_p): Remove static.
31124 (aff_combination_const_p): New interface.
31125 (aff_combination_singleton_var_p): New interfaces.
31127 2017-05-11 Richard Biener <rguenther@suse.de>
31129 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31130 Skip unreachable blocks and destinations.
31131 (eliminate): Move stmt removal and fixup ...
31132 (fini_eliminate): ... here. Skip inserted exprs.
31133 (pass_pre::execute): Move fini_pre after fini_eliminate.
31134 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31135 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31136 PRE to get rid of dead code that has invalid SSA form and
31137 split critical edges again.
31139 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31141 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31143 2017-05-11 Richard Biener <rguenther@suse.de>
31145 * passes.c (execute_function_todo): Verify loops if they are
31146 said to be up-to-date.
31147 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31148 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31150 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31153 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31154 handle calling assemble_external ourself.
31157 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31158 modes with zero size. Enhance comment.
31160 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31162 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31163 built-ins for vec_xl and vec_xst with short and char pointer
31166 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31168 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31169 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31170 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31171 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31172 (_mm_maskz_min_round_ss): New intrinsics.
31173 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31174 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31175 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31176 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31177 (__builtin_ia32_minss_mask_round): New builtins.
31178 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31179 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31180 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31182 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31183 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31185 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31188 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31190 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31191 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31192 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31193 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31194 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31195 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31196 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31197 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31198 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31199 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31200 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31201 (__builtin_ia32_mulss_mask_round): New builtins.
31202 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31203 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31204 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31206 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31207 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31209 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31212 2017-05-10 Julia Koval <julia.koval@intel.com>
31214 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31215 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31216 (_mm256_setr_m128i): New intrinsics.
31218 2017-05-10 Julia Koval <julia.koval@intel.com>
31220 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31221 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31222 (_mm_maskz_rcp14_ss): New intrinsics.
31223 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31224 (__builtin_ia32_rcp14ss_mask): New builtins.
31225 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31227 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31229 PR tree-optimization/51513
31230 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31231 (assert_unreachable_fallthru_edge_p): Use it.
31232 (group_case_labels_stmt): Likewise.
31233 * tree-cfg.h: Prototype it.
31234 * stmt.c: Include cfghooks.h and tree-cfg.h.
31235 (emit_case_dispatch_table) <gap_label>: New local variable.
31236 Use it to fill dispatch table gaps.
31237 Test for default_label before updating probabilities.
31238 (expand_case) <default_label>: Remove unneeded initialization.
31239 Test for unreachable default case statement and remove its edge.
31240 Set default_label accordingly.
31241 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31243 2017-05-10 Carl Love <cel@us.ibm.com>
31245 * config/rs6000/rs6000-c: Add support for built-in functions
31246 vector signed char vec_neg (vector signed char)
31247 vector signed short int vec_neg (vector short int)
31248 vector signed int vec_neg (vector signed int)
31249 vector signed long long vec_neg (vector signed long long)
31250 vector float vec_neg (vector float)
31251 vector double vec_neg (vector double)
31252 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31254 * config/rs6000/altivec.h: Add define for vec_neg
31255 * doc/extend.texi: Update the built-in documentation for the
31256 new built-in functions.
31258 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31260 PR tree-optimization/77644
31261 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31263 2017-05-10 Nathan Sidwell <nathan@acm.org>
31265 * dumpfile.h (TDI_lang_all): New.
31266 (TDF_KIND): New. Renumber others
31267 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31269 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31271 (get_dump_file_name): Adjust suffix generation.
31272 (dump_enable_all): Use TDF_KIND.
31273 * doc/invoke.texi (-fdump-lang-all): Document.
31275 * dumpfile.h: Tabify.
31277 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31280 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31281 Move member access before delete.
31283 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31285 * tree-inline.c (expand_call_inline): Split block at stmt
31288 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31291 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31292 are now unused after splitting mov{sf,sd}_hardfloat.
31293 (f32_lr2): Likewise.
31294 (f32_lm): Likewise.
31295 (f32_lm2): Likewise.
31296 (f32_li): Likewise.
31297 (f32_li2): Likewise.
31298 (f32_lv): Likewise.
31299 (f32_sr): Likewise.
31300 (f32_sr2): Likewise.
31301 (f32_sm): Likewise.
31302 (f32_sm2): Likewise.
31303 (f32_si): Likewise.
31304 (f32_si2): Likewise.
31305 (f32_sv): Likewise.
31306 (f32_dm): Likewise.
31307 (f32_vsx): Likewise.
31308 (f32_av): Likewise.
31309 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31310 For movsf, order stores so the VSX stores occur before the GPR
31311 store which encourages the register allocator to use a traditional
31312 FPR instead of a GPR. For movsd, order the stores so that the GPR
31313 store comes before the VSX stores to allow the power6 to work.
31314 This is due to the power6 not having a 32-bit integer store
31315 instruction from a FPR.
31316 (movsf_hardfloat): Likewise.
31317 (movsd_hardfloat): Likewise.
31319 2017-05-09 Martin Sebor <msebor@redhat.com>
31321 PR translation/80280
31322 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31325 PR translation/80280
31326 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31327 data member added in r247778.
31328 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31330 2017-05-09 Nathan Sidwell <nathan@acm.org>
31332 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31334 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31337 2017-05-09 Marek Polacek <polacek@redhat.com>
31339 * doc/invoke.texi: Fix typo.
31341 2017-05-09 Richard Biener <rguenther@suse.de>
31343 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31344 (vrp_val_is_min): Likewise.
31345 (set_value_range_to_value): Likewise.
31346 (set_value_range_to_nonnegative): Likewise.
31347 (gimple_assign_nonzero_p): Likewise.
31348 (gimple_stmt_nonzero_p): Likewise.
31349 (vrp_int_const_binop): Likewise. Remove unreachable case.
31350 (adjust_range_with_scev): Adjust comments.
31351 (compare_range_with_value): Likewise.
31352 (extract_range_from_phi_node): Likewise.
31353 (test_for_singularity): Likewise.
31355 2017-05-09 Richard Biener <rguenther@suse.de>
31357 * tree-vrp.c (get_single_symbol): Add assert that we don't
31358 get overflowed constants as invariant part.
31359 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31360 checks. Use wi::cmp instead of recursing for integer constants.
31361 (compare_values): Just ignore whether we assumed undefined
31362 overflow instead of failing the compare.
31363 (extract_range_for_var_from_comparison_expr): Add comment before the
31364 TREE_NO_WARNING sets.
31365 (test_for_singularity): Likewise.
31366 (extract_range_from_comparison): Do not disable optimization
31367 when we assumed undefined overflow.
31368 (extract_range_basic): Remove init of unused var.
31370 2017-05-09 Richard Biener <rguenther@suse.de>
31372 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31373 (extract_range_from_multiplicative_op_1): Adjust.
31374 (extract_range_from_binary_expr_1): Use int_const_binop.
31376 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31379 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31380 rs6000_store_data_bypass_p in seven define_bypass directives and
31381 in several comments.
31382 * config/rs6000/rs6000-protos.h: Add prototype for
31383 rs6000_store_data_bypass_p function.
31384 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31385 function implements slightly different (rs6000-specific) semantics
31386 than store_data_bypass_p, returning false rather than aborting
31387 with assertion error when arguments do not satisfy the
31388 requirements of store data bypass.
31389 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31390 rs6000_store_data_bypass_p.
31392 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31394 * config/xtensa/xtensa-protos.h
31395 (xtensa_initial_elimination_offset): New declaration.
31396 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31397 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31398 macro definition, add case for FRAME_POINTER_REGNUM when
31399 FRAME_GROWS_DOWNWARD.
31400 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31401 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31402 xtensa_initial_elimination_offset.
31404 2017-05-08 Nathan Sidwell <nathan@acm.org>
31406 * doc/invoke.texi: Alphabetize -fdump options.
31408 2017-05-08 Martin Sebor <msebor@redhat.com>
31410 PR translation/80280
31411 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31413 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31415 * target.def (compute_frame_layout): New optional target hook.
31416 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31417 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31418 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31420 * reload1.c (verify_initial_elim_offsets): Likewise.
31421 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31422 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31423 (arm_get_frame_offsets): Split up into this ...
31424 (arm_compute_frame_layout): ... and this function.
31426 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31428 * config/aarch64/constraints.md (Usa): New constraint.
31429 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31431 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31433 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31434 with_multilib_list after it has been checked.
31436 2017-05-08 Richard Biener <rguenther@suse.de>
31438 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31439 (bitmap_set_subtract_values): Likewise.
31441 2017-05-08 Richard Biener <rguenther@suse.de>
31443 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31444 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31446 (gimple_stmt_nonzero_warnv_p): Rename to ...
31447 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31449 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31450 (extract_range_basic): Adjust, do not disable propagation on
31451 strict overflow sensitive simplification.
31452 (vrp_visit_cond_stmt): Likewise.
31454 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31456 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31457 body size unconditionally.
31459 2017-05-07 Jeff Law <law@redhat.com>
31462 2017-05-06 Jeff Law <law@redhat.com>
31463 PR tree-optimization/78496
31464 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31467 PR tree-optimization/78496
31468 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31469 (simplify_stmt_using_ranges): Call it.
31470 (vrp_dom_walker::before_dom_children): Extract equivalences
31471 from an ASSERT_EXPR with an equality comparison against a
31474 2017-05-06 Jeff Law <law@redhat.com>
31476 PR tree-optimization/78496
31477 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31480 PR tree-optimization/78496
31481 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31482 (simplify_stmt_using_ranges): Call it.
31483 (vrp_dom_walker::before_dom_children): Extract equivalences
31484 from an ASSERT_EXPR with an equality comparison against a
31487 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31489 * lra-constraints.c (lra_copy_reg_equiv): New function.
31490 (split_reg): Use it to copy equivalence information from the
31491 original register to the spill register.
31493 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31495 PR rtl-optimization/75964
31496 * simplify-rtx.c (simplify_const_relational_operation): Remove
31497 invalid handling of comparisons of integer ABS.
31499 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31501 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31502 initialize to zero.
31503 (init_regs): Remove declaration.
31504 (function_arg_advance_32): Initialize error_p as boolean variable.
31506 2017-05-05 Nathan Sidwell <nathan@acm.org>
31508 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31509 lines. Use for (;;).
31511 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31513 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31514 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31515 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31516 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31517 VF=2 that require versioning.
31519 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31521 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31524 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31526 * diagnostic.h (diagnostic_override_option_index): Convert from
31527 macro to inline function.
31529 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31531 * diagnostic.c (last_module_changed_p): New function.
31532 (set_last_module): New function.
31533 (diagnostic_report_current_module): Convert macro usage to
31534 the above functions.
31535 * diagnostic.h (diagnostic_context::last_module): Strengthen
31536 from const line_map * to const line_map_ordinary *.
31537 (diagnostic_last_module_changed): Delete macro.
31538 (diagnostic_set_last_module): Delete macro.
31540 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31542 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31543 with diagnostic_report_diagnostic.
31544 (diagnostic_n_impl_richloc): Likewise.
31545 * diagnostic.h (report_diagnostic): Delete macro.
31546 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31547 with diagnostic_report_diagnostic.
31548 * substring-locations.c (format_warning_va): Likewise.
31550 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31552 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31553 save/restor of format_spec. Move option-printing code to...
31554 (print_option_information): ...this new function, and
31555 reimplement by simply printing to the pretty_printer,
31556 rather than appending to the format string.
31558 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31560 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31561 handling logic into...
31562 (update_effective_level_from_pragmas): ...this new function.
31564 2017-05-04 Andrew Waterman <andrew@sifive.com>
31566 * config/riscv/riscv.opt (mstrict-align): New option.
31567 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31568 (SLOW_UNALIGNED_ACCESS): Define.
31569 (riscv_slow_unaligned_access): Declare.
31570 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31572 (riscv_slow_unaligned_access): New variable.
31573 (rocket_tune_info): Set slow_unaligned_access to true.
31574 (optimize_size_tune_info): Set slow_unaligned_access to false.
31575 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31576 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31577 (riscv_option_override): Set riscv_slow_unaligned_access.
31578 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31580 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31582 * config/riscv/riscv.md: Unify indentation.
31584 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31589 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31591 (extendsi<mode>2): Add support for doing sign extension via
31592 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31593 don't have ISA 3.0 instructions.
31594 (extendsi<mode>2 splitter): Likewise.
31595 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31596 generate the normal insns since SImode can now go in vector
31597 registers. Disallow the special UNSPECs needed for previous
31598 machines to hide SImode being used. Add new insns
31599 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31600 (fix_trunc<mode>si2_stfiwx): Likewise.
31601 (fix_trunc<mode>si2_internal): Likewise.
31602 (fixuns_trunc<mode>si2): Likewise.
31603 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31604 (fctiw<u>z_<mode>_smallint): Likewise.
31605 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31606 of floating point to 32-bit integer from doing a direct move to
31607 the GPR registers to do a store.
31608 (fctiwz_<mode>): Break long line.
31610 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31612 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31613 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31614 (addr_list, addr_offset_valid_p): New.
31615 (split_address_groups): Check offset validity with above function.
31616 (gt-tree-ssa-loop-ivopts.h): Include header file.
31618 2017-05-05 Nathan Sidwell <nathan@acm.org>
31620 * config.gcc (arm*-*-*): Add missing 'fi'.
31622 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31624 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31625 included in -fopt-info does not matter.
31626 * doc/optinfo.texi (-fopt-info): Fix description of default
31627 behavour. Explicitly say order of options included in -fopt-info
31630 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31632 * config.gcc: Allow combinations of aprofile and rmprofile values for
31633 --with-multilib-list.
31634 * config/arm/t-multilib: New file.
31635 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31636 variables. Remove setting of ISA and floating-point ABI in
31637 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31638 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31639 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31640 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31642 * config/arm/t-rmprofile: Likewise except for the matches changes.
31643 * doc/install.texi (--with-multilib-list): Document the combination of
31644 aprofile and rmprofile values and warn about pitfalls in doing that.
31646 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31648 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31649 (movdi_aarch64): Likewise.
31651 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31653 PR tree-optimization/80632
31654 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31656 (build_arrays): Initialize it for virtual phis.
31657 (fix_phi_nodes): Use it for virtual phis.
31659 PR tree-optimization/80558
31660 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31661 [x, y] op z into [x op, y op z] for op & or | if conditions
31664 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31665 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31668 * config/arm/arm.md (use_literal_pool): Remove.
31669 (64-bit immediate split): No longer takes cost into consideration
31670 if arm_disable_literal_pool is enabled.
31671 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31672 used when arm_disable_literal_pool is enabled.
31673 (arm_max_const_double_inline_cost): Remove use of
31674 arm_disable_literal_pool.
31675 (push_minipool_fix): Add assert.
31676 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31677 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31678 (no_literal_pool_sf_immediate): New.
31680 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31682 PR tree-optimization/80613
31683 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31684 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31686 2017-05-05 Richard Biener <rguenther@suse.de>
31688 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31690 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31692 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31693 of this flag from insn conditions due to removal from r247495.
31695 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31697 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31699 (arm_early_store_addr_dep_ptr): Likewise.
31700 * config/arm/aarch-common-protos.h
31701 (arm_early_load_addr_dep_ptr): Add prototype.
31702 (arm_early_store_addr_dep_ptr): Likewise.
31703 * config/arm/cortex-a53.md: Add new bypasses.
31705 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31707 * tree.c (next_type_uid): Change type to unsigned.
31708 (type_hash_canon): Decrement back next_type_uid if
31709 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31710 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31713 2017-05-04 Martin Sebor <msebor@redhat.com>
31715 * builtins.c: Fix a trivial typo in a comment.
31717 PR middle-end/79234
31718 * builtins.c (check_sizes): Adjust to handle reading past the end.
31719 Avoid printing excessive upper bound of ranges. Use %E to print
31720 tree nodes instead of converting them to %wu.
31721 (expand_builtin_memchr): New function.
31722 (compute_dest_size): Rename...
31723 (compute_objsize): ...to this.
31724 (expand_builtin_memcpy): Adjust.
31725 (expand_builtin_mempcpy): Adjust.
31726 (expand_builtin_strcat): Adjust.
31727 (expand_builtin_strcpy): Adjust.
31728 (check_strncat_sizes): Adjust.
31729 (expand_builtin_strncat): Adjust.
31730 (expand_builtin_strncpy): Adjust and simplify.
31731 (expand_builtin_memset): Adjust.
31732 (expand_builtin_bzero): Adjust.
31733 (expand_builtin_memcmp): Adjust.
31734 (expand_builtin): Handle memcmp.
31735 (maybe_emit_chk_warning): Check strncat just once.
31737 2017-05-04 Martin Sebor <msebor@redhat.com>
31739 PR preprocessor/79214
31740 PR middle-end/79222
31741 PR middle-end/79223
31742 * builtins.c (check_sizes): Add inlining context and issue
31743 warnings even when -Wno-system-headers is set.
31744 (check_strncat_sizes): Same.
31745 (expand_builtin_strncat): Same.
31746 (expand_builtin_memmove): New function.
31747 (expand_builtin_stpncpy): Same.
31748 (expand_builtin): Handle memmove and stpncpy.
31750 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31752 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31753 which is not used any more.
31755 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31757 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31759 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31761 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31762 (cortexa53_tunings): Likewise.
31763 (cortexa57_tunings): Likewise.
31764 (cortexa72_tunings): Likewise.
31765 (cortexa73_tunings): Likewise.
31767 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31769 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31770 Set loop alignment to 8.
31772 2017-05-04 Martin Sebor <msebor@redhat.com>
31774 PR translation/80280
31775 * builtins.c (expand_builtin_object_size): Add missing quoting to
31776 %D and like directives.
31777 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31778 (hsa_type_for_tree_type): Same.
31779 (verify_function_arguments): Same.
31780 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31781 * varasm.c (get_section): Same.
31784 2017-05-04 Martin Sebor <msebor@redhat.com>
31786 PR translation/80280
31787 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31789 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31791 * config/aarch64/aarch64.c (generic_addrcost_table):
31792 Change HI/TI mode setting.
31794 2017-05-04 Martin Jambor <mjambor@suse.cz>
31796 PR tree-optimization/80622
31797 * tree-sra.c (comes_initialized_p): New function.
31798 (build_accesses_from_assign): Only set write lazily when
31799 comes_initialized_p is false.
31800 (analyze_access_subtree): Use comes_initialized_p.
31801 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31802 instead of testing for PARM_DECL.
31804 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31806 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31807 constraint on operand 0 to allow more general addressing modes.
31808 Adjust output template.
31809 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31811 * config/aarch64/aarch64-protos.h
31812 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31813 * config/aarch64/constraints.md (Dp): New address constraint.
31814 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31817 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31819 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31820 update use of estimate_ipcp_clone_size_and_time.
31821 (estimate_local_effects): Update use of
31822 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31823 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31824 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31825 Return nonspecialized time.
31827 2017-05-04 Richard Biener <rguenther@suse.de>
31829 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31830 for the last VUSE which def dominates the PHI. Directly call
31832 (get_continuation_for_phi_1): Remove.
31834 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31836 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31837 to explain the use of truncating division. Cap the number of
31838 iterations to the maximum given by nb_iterations_upper_bound,
31841 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31843 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31844 * configure: Regenerate.
31845 * config.in: Regenerate.
31846 * config/i386/driver-mingw32.c: new file.
31847 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31848 * config.host: Link driver-mingw32.o on MinGW host.
31849 * doc/install.texi: Document new --enable-mingw-wildcard configure
31852 2017-05-04 Marek Polacek <polacek@redhat.com>
31854 PR tree-optimization/80612
31855 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31857 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31858 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31860 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31861 (movt splitter): Likewise.
31862 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31863 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31864 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31865 block for Thumb-1 with MOVT.
31866 (thumb2_legitimate_address_p): Move code block ...
31867 (can_avoid_literal_pool_for_label_p): ... into this new function.
31868 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31870 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31871 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31872 "M-profile targets with the MOVT instruction".
31874 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31876 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31877 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31878 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31880 2017-05-04 Martin Liska <mliska@suse.cz>
31882 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31883 variable cond_code.
31885 2017-05-04 Richard Biener <rguenther@suse.de>
31887 * tree.c (array_at_struct_end_p): Handle arrays at struct
31888 end with flexarrays more conservatively. Refactor and treat
31889 arrays of arrays or aggregates more strict. Fix
31890 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31891 * tree.c (array_at_struct_end_p): Adjust prototype.
31892 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31893 * gimple-fold.c (get_range_strlen): Likewise.
31894 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31896 2017-05-04 Richard Biener <rguenther@suse.de>
31898 PR tree-optimization/31130
31899 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31901 (supports_overflow_infinity): Likewise.
31902 (is_negative_overflow_infinity): Likewise.
31903 (is_positive_overflow_infinity): Likewise.
31904 (is_overflow_infinity): Likewise.
31905 (stmt_overflow_infinity): Likewise.
31906 (overflow_infinity_range_p): Likewise.
31907 (usable_range_p): Remove as always returning true.
31908 (make_overflow_infinity): Remove.
31909 (negative_overflow_infinity): Likewise.
31910 (positive_overflow_infinity): Likewise.
31911 (avoid_overflow_infinity): Likewise.
31912 (set_value_range): Adjust accordingly.
31913 (set_value_range_to_nonnegative): Likewise, remove now unused
31914 overflow_infinity arg.
31915 (vrp_operand_equal_p): Adjust.
31916 (update_value_range): Likewise.
31917 (range_int_cst_singleton_p): Likewise.
31918 (operand_less_p): Likewise.
31919 (compare_values_warnv): Likewise.
31920 (extract_range_for_var_from_comparison_expr): Likewise.
31921 (vrp_int_const_binop): Likewise.
31922 (zero_nonzero_bits_from_vr): Likewise.
31923 (extract_range_from_multiplicative_op_1): Likewise.
31924 (extract_range_from_binary_expr_1): Likewise.
31925 (extract_range_from_unary_expr): Likewise.
31926 (extract_range_from_comparison): Likewise.
31927 (extract_range_basic): Likewise.
31928 (adjust_range_with_scev): Likewise.
31929 (compare_ranges): Likewise.
31930 (compare_range_with_value): Likewise.
31931 (dump_value_range): Likewise.
31932 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31934 (simplify_cond_using_ranges): Adjust.
31936 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31938 * brig-builtins.def: Added a builtin for class_f64.
31939 * builtin-types.def: Added a builtin type needed by class_f64.
31941 2017-05-03 Jason Merrill <jason@redhat.com>
31943 * timevar.def: Add TV_CONSTEXPR.
31945 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31947 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31949 2017-05-03 Martin Jambor <mjambor@suse.cz>
31951 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31952 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31953 * ipa-cp.c (ipcp_propagate_stage): Do not call
31954 ipa_update_after_lto_read.
31955 * ipa-inline.c (ipa_inline): Likewise.
31957 2017-05-03 Martin Jambor <mjambor@suse.cz>
31959 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31960 tag. Added a default constructor and a destructor.
31961 (ipa_edge_args_sum_t): New class;
31962 (ipa_edge_args_sum): Declare.
31963 (ipa_edge_args_vector): Remove declaration.
31964 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31965 (ipa_free_edge_args_substructures): Remove declaration.
31966 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31967 (ipa_edge_args_info_available_for_edge_p): Likewise.
31968 * ipa-prop.c (ipa_edge_args_vector): Removed.
31969 (edge_removal_hook_holder): Likewise.
31970 (edge_duplication_hook_holder): Likewise.
31971 (ipa_edge_args_sum): New variable.
31972 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31973 ipa_edge_args_vector.
31974 (ipa_free_edge_args_substructures): Likewise.
31975 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31976 ipa_edge_args_vector.
31977 (ipa_edge_removal_hook): Turned into method
31978 ipa_edge_args_sum_t::remove.
31979 (ipa_edge_duplication_hook): Turned into method
31980 ipa_edge_args_sum_t::duplicate.
31981 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31982 registering edge hooks.
31983 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31984 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31985 ipa_edge_args_sum instead of ipa_edge_args_vector.
31986 * ipa-profile.c (ipa_profile): Likewise.
31988 2017-05-03 Martin Jambor <mjambor@suse.cz>
31990 * symbol-summary.h (function_summary): New method exists.
31991 (function_summary::symtab_removal): Deallocate through release.
31992 (call_summary): New class.
31993 (gt_ggc_mx): New overload.
31994 (gt_pch_nx): Likewise.
31995 (gt_pch_nx): Likewise.
31997 2017-05-03 Jeff Law <law@redhat.com>
31999 PR tree-optimization/78496
32000 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32001 from simplify_cond_using_ranges. Split off code to walk
32002 backwards through casts into ...
32003 (simplify_cond_using_ranges_2): New function.
32004 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32005 (execute_vrp): After identifying jump threads, call
32006 simplify_cond_using_ranges_2.
32008 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32011 * ipa-inline.h (inline_summary): Add ctor.
32012 (create_ggc): Do not use ggc_cleared_alloc.
32014 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32015 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32017 * gcc.c (handle_braces): Support escaping in switch matching
32019 * doc/invoke.texi (Spec Files): Document it.
32020 Remove superfluous @code markup in items.
32022 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32024 * diagnostic-show-locus.c (struct column_range): New struct.
32025 (get_affected_columns): New function.
32026 (get_printed_columns): New function.
32027 (struct correction): New struct.
32028 (correction::ensure_capacity): New function.
32029 (correction::ensure_terminated): New function.
32030 (struct line_corrections): New struct.
32031 (line_corrections::~line_corrections): New dtor.
32032 (line_corrections::add_hint): New function.
32033 (layout::print_trailing_fixits): Reimplement in terms of the new
32035 (selftest::test_overlapped_fixit_printing): New function.
32036 (selftest::diagnostic_show_locus_c_tests): Call it.
32038 2017-05-03 Nathan Sidwell <nathan@acm.org>
32040 Canonicalize canonical type hashing
32041 * tree.h (type_hash_canon_hash): Declare.
32042 * tree.c (type_hash_list, attribute_hash_list): Move into
32043 type_hash_canon_hash.
32044 (build_type_attribute_qual_variant): Break out hash code calc into
32045 type_hash_canon_hash.
32046 (type_hash_canon_hash): New. Generic type hash computation.
32047 (build_range_type_1, build_array_type_1, build_function_type,
32048 build_method_type_directly, build_offset_type, build_complex_type,
32049 make_vector_type): Call it.
32051 2017-05-03 Richard Biener <rguenther@suse.de>
32053 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32054 When all DRs have unknown misaligned do not always peel
32055 when there is a store but apply the same costing model as if
32056 there were only loads.
32058 2017-05-03 Richard Biener <rguenther@suse.de>
32061 PR tree-optimization/80492
32062 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32063 compare_base_decls returning dont-know properly.
32065 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32067 * config/arm/iterators.md (CCSI): New mode iterator.
32068 (arch): New mode attribute.
32069 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32070 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32071 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32072 code iterator for success result mode.
32073 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32074 the corresponding new insn generators.
32076 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32079 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32080 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32082 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32084 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32085 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32086 (DDR_A): Wrap DDR argument in brackets.
32087 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32088 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32089 (DDR_REVERSED_P): Likewise.
32091 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32093 PR tree-optimization/79472
32094 * tree-switch-conversion.c (struct switch_conv_info): Add
32095 contiguous_range and default_case_nonstandard fields.
32096 (collect_switch_conv_info): Compute contiguous_range and
32097 default_case_nonstandard fields, don't clear final_bb if
32098 contiguous_range and only the default case doesn't have the required
32100 (check_all_empty_except_final): Set default_case_nonstandard instead
32101 of failing if contiguous_range and the default case doesn't have empty
32103 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32104 and only the default case doesn't have the required constants. Skip
32106 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32107 if default_case_nonstandard.
32108 (build_constructors): Build constant 1 just once. Assert that default
32109 values aren't inserted in between cases if contiguous_range. Skip
32111 (build_arrays): Skip virtual phis.
32112 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32113 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32114 Handle virtual phis.
32115 (gen_inbound_check): Handle default_case_nonstandard case.
32116 (process_switch): Adjust check_final_bb caller. Call
32117 gather_default_values with the first non-default case instead of
32118 default case if default_case_nonstandard.
32120 2017-05-02 Nathan Sidwell <nathan@acm.org>
32122 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32123 check. Fix formatting.
32125 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32127 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32128 errors when comparing specialized and unspecialized times.
32130 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32132 * diagnostic-show-locus.c
32133 (layout::should_print_annotation_line_p): Make private.
32134 (layout::print_annotation_line): Make private.
32135 (layout::annotation_line_showed_range_p): Make private.
32136 (layout::show_ruler): Make private.
32137 (layout::print_source_line): Make private. Pass in line and
32138 line_width, rather than calling location_get_source_line. Drop
32140 (layout::print_leading_fixits): New method.
32141 (layout::print_any_fixits): Rename to...
32142 (layout::print_trailing_fixits): ...this, and make private.
32143 Don't print newline fixits.
32144 (diagnostic_show_locus): Move logic for printing one row into...
32145 (layout::print_line): ...this new function. Move the
32146 location_get_source_line call and error-handling from
32147 print_source_line to here. Call print_leading_fixits, and rename
32148 print_any_fixits to print_trailing_fixits.
32149 (selftest::test_fixit_insert_containing_newline): Update now that
32150 newlines are partially supported.
32151 (selftest::test_fixit_insert_containing_newline_2): New test.
32152 (selftest::test_fixit_replace_containing_newline): Update comments.
32153 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32154 * edit-context.c (class added_line): New class.
32155 (class edited_line): Describe newline handling in comment.
32156 (edited_line::actually_edited_p): New method.
32157 (edited_line::print_content): Delete redundant decl.
32158 (edited_line::m_predecessors): New field.
32159 (edited_file::print_content): Call edited_line::print_content.
32160 (edited_file::print_diff): Update to support newlines.
32161 (edited_file::print_diff_hunk): Likewise.
32162 (edited_file::print_run_of_changed_lines): New function.
32163 (edited_file::print_diff_line): Convert to...
32164 (print_diff_line): ...this.
32165 (edited_file::get_effective_line_count): New function.
32166 (edited_line::edited_line): Initialize new field m_predecessors.
32167 (edited_line::~edited_line): Clean up m_predecessors.
32168 (edited_line::apply_fixit): Handle newlines.
32169 (edited_line::get_effective_line_count): New function.
32170 (edited_line::print_content): New function.
32171 (edited_line::print_diff_lines): New function.
32172 (selftest::test_applying_fixits_insert_containing_newline): New
32174 (selftest::test_applying_fixits_replace_containing_newline): New
32176 (selftest::insert_line): New function.
32177 (selftest::test_applying_fixits_multiple_lines): Add example of
32179 (selftest::edit_context_c_tests): Call the new tests.
32181 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32183 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32184 parameter cand. Update dump information.
32185 (get_computation_cost): Update uses.
32187 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32189 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32190 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32191 (get_computation_at, rewrite_use_address): Update use of
32192 get_computation_aff.
32194 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32196 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32197 (get_computation): Delete.
32198 (get_computation_cost): Implement like get_computation_cost_at.
32199 Use get_computation_at.
32200 (get_computation_cost_at): Delete.
32201 (rewrite_use_nonlinear_expr): Use get_computation_at.
32202 (rewrite_use_compare, remove_unused_ivs): Ditto.
32204 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32206 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32208 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32210 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32211 (ivopts_global_cost_for_size): Rename parameter and update uses.
32212 (iv_ca_recount_cost): Update uses.
32213 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32214 candidates seperately in n_invs and n_cands.
32215 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32217 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32219 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32220 (find_inv_vars_cb): New.
32221 (find_depends): Renamed to ...
32222 (find_inv_vars): ... this.
32223 (add_candidate_1, force_var_cost): Call find_inv_vars.
32224 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32226 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32228 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32229 inv_vars. Add inv_exprs.
32230 (struct iv_cand): Rename depends_on to inv_vars.
32231 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32232 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32233 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32234 (dump_cand): Dump inv_vars.
32235 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32236 (record_invariant, find_depends, add_candidate_1): Ditto.
32237 (set_group_iv_cost, force_var_cost): Ditto.
32238 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32239 (get_computation_cost_at, get_computation_cost): Ditto.
32240 (determine_group_iv_cost_generic): Ditto.
32241 (determine_group_iv_cost_address): Ditto.
32242 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32243 (determine_group_iv_costs): Ditto.
32244 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32245 (iv_ca_set_remove_invariants): Renamed to ...
32246 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32247 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32248 (iv_ca_set_add_invariants): Renamed to ...
32249 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32250 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32251 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32252 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32253 (create_new_ivs): Remove useless dump.
32255 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32257 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32259 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32260 (iv_ca_set_no_cp, create_new_iv): Ditto.
32262 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32264 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32266 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32268 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32269 function tree_check2.
32271 2017-05-02 Martin Liska <mliska@suse.cz>
32273 * doc/gcov.texi: Add missing preposition.
32274 * gcov.c (function_info::function_info): Properly fill up
32275 all member variables.
32277 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32279 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32281 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32283 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32285 2017-05-02 Martin Liska <mliska@suse.cz>
32288 * lto-streamer-in.c (lto_read_tree_1): Remove
32289 LTO_STREAMER_DEBUG.
32290 * lto-streamer.c (struct tree_hash_entry): Likewise.
32291 (struct tree_entry_hasher): Likewise.
32292 (tree_entry_hasher::hash): Likewise.
32293 (tree_entry_hasher::equal): Likewise.
32294 (lto_streamer_init): Likewise.
32295 (lto_orig_address_map): Likewise.
32296 (lto_orig_address_get): Likewise.
32297 (lto_orig_address_remove): Likewise.
32298 * lto-streamer.h: Likewise.
32299 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32300 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32302 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32304 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32305 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32306 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32307 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32308 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32309 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32310 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32311 (mm_maskz_sub_ss): New intrinsics.
32312 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32313 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32314 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32315 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32316 (__builtin_ia32_subss_mask_round): New builtins.
32317 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32318 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32319 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32321 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32322 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32324 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32327 2017-05-02 Martin Jambor <mjambor@suse.cz>
32329 PR tree-optimization/78687
32330 * tree-sra.c (access): New field parent.
32331 (process_subtree_disqualification): New function.
32332 (disqualify_candidate): Call it.
32333 (build_accesses_from_assign): Reset write flag if creating an
32335 (build_access_subtree): Fill in parent field and also prpagate
32336 down grp_write flag.
32337 (create_artificial_child_access): New parameter set_grp_write, set
32338 grp_write to its value.
32339 (propagate_subaccesses_across_link): Also propagate grp_write flag
32341 (propagate_all_subaccesses): Push the closest parent back to work
32342 queue if add_access_to_work_queue returned true.
32344 2017-05-02 Richard Biener <rguenther@suse.de>
32346 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32347 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32348 -fstrict-overflow documentation.
32349 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32350 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32351 flag_strict_overflow.
32352 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32353 * lto-opts.c (lto_write_options): Do not stream it.
32354 * lto-wrapper.c (merge_and_complain): Do not handle it.
32355 * opts.c (default_options_table): Do not set -fstrict-overflow.
32356 (finish_options): Likewise do not clear it when sanitizing.
32357 * simplify-rtx.c (simplify_const_relational_operation): Do not
32358 test flag_strict_overflow.
32360 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32362 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32363 using enabled attribute.
32364 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32365 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32366 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32367 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32368 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32369 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32370 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32371 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32372 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32373 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32375 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32377 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32379 2017-05-02 Richard Biener <rguenther@suse.de>
32381 PR tree-optimization/80591
32383 2017-04-10 Richard Biener <rguenther@suse.de>
32385 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32388 2017-05-02 Richard Biener <rguenther@suse.de>
32390 PR tree-optimization/80549
32391 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32392 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32393 headers if they do not have a preheader.
32395 2017-05-02 Martin Liska <mliska@suse.cz>
32398 * common.opt: Fix typo.
32399 * doc/invoke.texi: Likewise.
32401 2017-05-01 Jan Beulich <jbeulich@suse.com>
32403 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32404 swapping, add (x,x,m,x,n) alternative.
32406 2017-05-01 Nathan Sidwell <nathan@acm.org>
32408 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32409 unnecessary unadjusted_alignment check.
32411 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32414 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32416 * gimplify.c (gimplify_cilk_detach): New function.
32417 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32418 * tree-core.h: Document EXPR_CILK_SPAWN.
32419 * tree.h (EXPR_CILK_SPAWN): Define.
32421 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32423 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32424 to use new fixit_hint representation, using the "replace" logic.
32425 (get_line_span_for_fixit_hint): Likewise.
32426 (layout::print_any_fixits): Likewise.
32427 (selftest::test_one_liner_many_fixits): Rename to...
32428 (selftest::test_one_liner_many_fixits_1): ...this, and update
32429 comment and expected output to reflect that the multiple fix-it
32430 hints are now consolidated into one insertion.
32431 (selftest::test_one_liner_many_fixits_2): New test.
32432 (selftest::test_diagnostic_show_locus_one_liner): Update for
32434 (selftest::test_fixit_consolidation): Update for fix-it API
32436 * diagnostic.c (print_parseable_fixits): Likewise.
32437 * edit-context.c (edited_line::m_line_events): Convert from
32438 auto_vec <line_event *> to auto_vec <line_event>.
32439 (class line_event): Convert from abstract base class to a concrete
32440 class, taking over the role of replace_event.
32441 (class insert_event): Delete.
32442 (class replace_event): Rename to class line_event. Convert to
32444 (edit_context::add_fixits): Reimplement.
32445 (edit_context::apply_insert): Delete.
32446 (edit_context::apply_replace): Rename to...
32447 (edit_context::apply_fixit): ...this. Convert to half-open range.
32448 (edited_file::apply_insert): Delete.
32449 (edited_file::apply_replace): Rename to...
32450 (edited_file::apply_fixit): ...this.
32451 (edited_line::~edited_line): Drop deletion of events.
32452 (edited_line::apply_insert): Delete.
32453 (edited_line::apply_replace): Rename to...
32454 (edited_line::apply_fixit): ...this. Convert to half-open range.
32455 Update for change to type of m_line_events.
32456 * edit-context.h (edit_context::apply_insert): Delete.
32457 (edit_context::apply_replace): Rename to...
32458 (edit_context::apply_fixit): ...this.
32460 2017-05-01 Martin Sebor <msebor@redhat.com>
32462 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32465 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32468 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32469 __get_cpuid_max returns 0.
32470 (__get_cpuid_count): Ditto.
32472 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32474 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32475 replacement expression is another instance of one of its arguments.
32477 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32480 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32481 check for stack push/pop autoinc.
32482 * config/i386/i386.c (ix86_agi_dependent): Return false
32483 if the only reason why modified_in_p returned true is that
32484 addr is SP based and set_insn is a push or pop.
32486 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32488 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32491 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32494 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32495 (account_size_time): Use two predicates - exec_pred and
32497 (evaluate_conditions_for_known_args): Compute both clause and
32499 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32500 (inline_summary_t::duplicate): Update.
32501 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32503 (compute_inline_parameters): Likewise.
32504 (estimate_edge_size_and_time): Update caluclation of time.
32505 (estimate_node_size_and_time): Compute both time and nonspecialized
32507 (estimate_ipcp_clone_size_and_time): Update.
32508 (inline_merge_summary): Update.
32509 (do_estimate_edge_time): Update.
32510 (do_estimate_edge_size): Update.
32511 (do_estimate_edge_hints): Update.
32512 (inline_read_section, inline_write_summary): Stream both new predicates.
32513 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32515 (compute_inlined_call_time): Cleanup.
32516 (big_speedup_p): Update.
32517 (edge_badness): Update.
32518 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32519 (size_time_entry): Replace predicate by exec_predicate and
32520 nonconst_predicate.
32521 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32522 (estimate_edge_time): Return also nonspec_time.
32523 (reset_edge_growth_cache): Update.
32525 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32527 PR rtl-optimization/80491
32528 * ifcvt.c (noce_process_if_block): When looking for x setter
32529 with missing else_bb, don't check only the insn right before
32530 cond_earliest, but look for the last insn that x is modified in
32531 within the same bb.
32533 PR rtl-optimization/80491
32534 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32536 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32538 PR tree-optimization/80487
32539 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32541 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32543 PR tree-optimization/79697
32544 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32545 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32546 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32548 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32549 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32551 2017-04-28 Martin Sebor <msebor@redhat.com>
32553 PR tree-optimization/80523
32554 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32555 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32557 (maybe_warn, format_directive, parse_directive): Use new functions.
32558 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32560 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32562 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32564 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32566 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32567 target_header_dir): Set correctly.
32568 * configure: Regenerated.
32569 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32570 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32571 instead of SYSTEM_HEADER_DIR.
32573 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32575 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32576 (estimate_local_effects): Likewise.
32577 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32578 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32579 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32580 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32581 do_estimate_edge_time, estimate_edge_time): Likewise.
32582 * ipa-inline-analysis.c (estimate_node_size_and_time,
32583 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32584 (estimate_time_after_inlining): Remove.
32586 2017-04-28 Martin Liska <mliska@suse.cz>
32588 * doc/gcov.texi: Enhance documentation of gcov.
32590 2017-04-28 Martin Liska <mliska@suse.cz>
32592 * doc/gcov.texi: Sort options in alphabetic order.
32593 * doc/gcov-dump.texi: Likewise.
32594 * doc/gcov-tool.texi: Likewise.
32595 * gcov.c (print_usage): Likewise.
32596 * gcov-dump.c (print_usage): Likewise.
32597 * gcov-tool.c (print_merge_usage_message): Likewise.
32598 (print_rewrite_usage_message): Likewise.
32599 (print_overlap_usage_message): Likewise.
32601 2017-04-28 Martin Liska <mliska@suse.cz>
32603 PR gcov-profile/53915
32604 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32606 2017-04-28 Martin Liska <mliska@suse.cz>
32608 PR gcov-profile/79891
32609 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32610 is marked by compiler as living on a line.
32611 (get_cycles_count): Remove usage of the union.
32612 (output_intermediate_file): Likewise.
32613 (find_source): Fix GNU coding style.
32614 (accumulate_line_counts): Remove old non-all block mode.
32615 (output_lines): Remove usage of the union.
32616 * profile.c (output_location): Include all BBs, even if
32617 belonging to a same line (and file) as a previous BB.
32619 2017-04-28 Martin Liska <mliska@suse.cz>
32621 * gcov.c (process_args): Handle new argument 'w'.
32622 (read_graph_file): Assign ID to BBs.
32623 (output_branch_count): Display BB # if verbose flag is set.
32624 (output_lines): Likewise for arcs.
32625 (print_usage): Add '--verbose' option help.
32626 * doc/gcov.texi: Document --verbose (-w) option.
32628 2017-04-28 Martin Liska <mliska@suse.cz>
32630 * gcov.c (struct block_location_info): New struct.
32631 (process_file): Fill up the new structure.
32632 (read_graph_file): Replace usage of encoding by the newly added
32634 (add_line_counts): Likewise.
32635 (accumulate_line_counts): Remove usage of the union.
32636 (function_info::function_info): New function.
32637 (function_info::~function_info): Likewise.
32638 (process_file): Call delete instead of release_function.
32639 (release_function): Release the function.
32640 (release_structures): Call delete instead of release_function.
32641 (solve_flow_graph): Replace usage of num_blocks.
32642 (find_exception_blocks): Likewise.
32643 (output_lines): Fix GNU coding style.
32645 2017-04-28 Martin Liska <mliska@suse.cz>
32648 * coverage.c (coverage_remove_note_file): New function.
32649 * coverage.h: Declare the function.
32650 * toplev.c (finalize): Clean if an error has been seen.
32652 2017-04-28 Martin Liska <mliska@suse.cz>
32654 PR gcov-profile/80031
32655 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32656 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32657 * gcov.c (read_graph_file): Read just number of blocks.
32658 * profile.c (branch_prob): Do not stream 0 flags per a basic
32661 2017-04-28 Martin Liska <mliska@suse.cz>
32663 * gcov-dump.c (tag_*): Add new argument to declarations.
32664 (dump_gcov_file): Likewise.
32665 (tag_blocks): Add and use new argument depth.
32666 (tag_arcs): Likewise.
32667 (tag_lines): Likewise.
32668 (tag_counters): Likewise.
32669 (tag_summary): Likewise.
32670 (dump_working_sets): Use depth to do a proper indentation.
32672 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32675 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32676 * symtab.c (symtab_node::debug_symtab): Move definition here.
32678 2017-04-28 Richard Biener <rguenther@suse.de>
32680 * lto-streamer.h (LTO_major_version): Bump to 7.
32682 2017-04-28 Richard Biener <rguenther@suse.de>
32684 * tree-vrp.c (assert_info): New struct.
32685 (add_assert_info): New helper.
32686 (register_edge_assert_for_2): Refactor to add asserts to a vector
32688 (register_edge_assert_for_1): Likewise.
32689 (register_edge_assert_for): Likewise.
32690 (finish_register_edge_assert_for): New helper actually registering
32691 asserts where live on edge.
32692 (find_conditional_asserts): Adjust.
32693 (find_switch_asserts): Likewise.
32694 (evrp_dom_walker::try_find_new_range): Generalize.
32695 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32697 2017-04-27 Marek Polacek <polacek@redhat.com>
32700 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32701 arg10 and arg11 to itype.
32703 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32705 * doc/extend.texi (Object Size Checking): Improve grammar.
32707 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32710 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32711 that the logic for permitting reciprocal estimates matches that
32714 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32717 * tree.c (type_cache_hasher::equal): Only compare
32718 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32719 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32720 non-aggregate element types.
32721 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32722 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32724 2017-04-27 Richard Biener <rguenther@suse.de>
32726 PR middle-end/80533
32727 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32728 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32729 keeping a reference to a trailing array.
32731 2017-04-27 Richard Biener <rguenther@suse.de>
32733 PR middle-end/80539
32734 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32735 being in loop-closed SSA form conservatively.
32736 (chrec_fold_multiply_poly_poly): Likewise.
32738 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32740 PR middle-end/79665
32741 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32742 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32744 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32747 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32748 (aarch64_function_arg_alignment): Return unsigned int again, but still
32749 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32750 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32751 Don't emit -Wpsabi note.
32752 (aarch64_function_arg_boundary): Likewise.
32753 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32756 2017-04-26 Nathan Sidwell <nathan@acm.org>
32758 * tree.h (crc32_unsigned_n): Declare.
32759 (crc32_unsigned, crc32_unsigned): Make inline.
32760 * tree.c (crc32_unsigned_bits): Replace with ...
32761 (crc32_unsigned_n): ... this.
32762 (crc32_unsigned, crc32_byte): Remove.
32763 (crc32_string): Remove unnecessary braces.
32765 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32767 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32768 * ipa-inline-analysis.c (MAX_TIME): Remove.
32769 (account_size_time): Use sreal for time.
32770 (dump_inline_summary): Update.
32771 (estimate_function_body_sizes): Update.
32772 (estimate_edge_size_and_time): Update.
32773 (estimate_calls_size_and_time): Update.
32774 (estimate_node_size_and_time): Update.
32775 (inline_merge_summary): Update.
32776 (inline_update_overall_summary): Update.
32777 (estimate_time_after_inlining): Update.
32778 (inline_read_section): Update.
32779 (inline_write_summary): Update.
32780 * ipa-inline.c (compute_uninlined_call_time): Update.
32781 (compute_inlined_call_time): Update.
32782 (recursive_inlining): Update.
32783 (inline_small_functions): Update.
32784 (dump_overall_stats): Update.
32785 * ipa-inline.h: Include sreal.h.
32786 (size_time_entry): Turn time to sreal.
32787 (inline_summary): Turn self_time nad time to sreal.
32789 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32791 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32793 (sreal::stream_out, sreal::stream_in): New.
32794 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32796 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32798 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32801 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32804 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32806 (dimode_scalar_chain::compute_convert_gain): Ditto.
32807 (dimode_scalar_chain::make_vector_copies): Ditto.
32808 (dimode_scalar_chain::convert_reg): Ditto.
32809 (dimode_scalar_chain::convert_insn): Ditto.
32810 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32811 (VI248_AVX512BW_1): New mode iterator.
32812 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32813 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32816 2017-04-25 Martin Sebor <msebor@redhat.com>
32818 PR tree-optimization/80497
32819 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32820 constants are representable in HOST_WIDE_INT.
32821 (parse_directive): Ditto.
32823 2017-04-25 Martin Sebor <msebor@redhat.com>
32826 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32827 (new_zero_array): Adjust signature.
32828 (dom_info::dom_init): Used unsigned rather that size_t.
32829 (dom_info::dom_info): Same.
32831 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32832 Jakub Jelinek <jakub@redhat.com>
32835 * config/arm/arm.c: Include gimple.h.
32836 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32837 returns negative, increment ncrn only if it returned positive.
32838 (arm_needs_doubleword_align): Return int instead of bool,
32839 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32840 members, but if there is any such non-FIELD_DECL
32841 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32842 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32843 returns negative, increment nregs only if it returned positive.
32844 (arm_setup_incoming_varargs): Likewise.
32845 (arm_function_arg_boundary): Emit -Wpsabi note if
32846 arm_needs_doubleword_align returns negative, return
32847 DOUBLEWORD_ALIGNMENT only if it returned positive.
32849 2017-04-25 Marek Polacek <polacek@redhat.com>
32852 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32853 first argument to type.
32855 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32858 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32859 type checks to test for compatibility instead of equality.
32861 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32862 Jakub Jelinek <jakub@redhat.com>
32865 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32867 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32868 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32869 the alignment computation, but return their maximum in warn_alignment.
32870 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32871 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32873 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32874 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32877 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32879 * config/arc/simdext.md (dmpyh): Fix typo.
32881 2017-04-25 Richard Biener <rguenther@suse.de>
32883 PR tree-optimization/80492
32884 * alias.c (compare_base_decls): Handle registers with asm
32885 specification conservatively.
32886 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32887 compare_base_decls returning dont-know properly.
32889 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32891 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32892 (legitimate_offset_address_p): New function.
32893 (arc_legitimate_address_p): Use above function.
32895 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32897 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32899 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32901 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32902 ACCH registers whenever they are available.
32904 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32906 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32907 double regs fix when not used.
32909 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32911 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32913 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32914 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32916 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32918 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32919 when emitting short ADD instructions.
32921 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32923 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32925 (cmpsi_cc_c_insn): Likewise.
32926 (cbranchsi4_scratch): Compute proper instruction length using
32927 compact_hreg_operand.
32928 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32930 2017-04-25 Richard Biener <rguenther@suse.de>
32932 PR middle-end/80509
32933 * passes.c (pass_manager::pass_manager): Initialize
32934 m_name_to_pass_map.
32936 2017-04-25 Richard Biener <rguenther@suse.de>
32938 PR tree-optimization/79201
32939 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32941 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32944 * config/s390/vector.md: Split MEM->GPR vector moves for
32945 non-s_operand addresses.
32947 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32950 * config/s390/predicates.md (reload_const_wide_int_operand): New
32952 * config/s390/s390.md ("movti"): Remove d/P alternative.
32953 ("movti_bigconst"): New pattern definition.
32955 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32958 * s390-protos.h (s390_expand_cs_hqi): Removed.
32959 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32960 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32961 modes as well as CCZ1mode and CCZmode.
32962 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32963 signature of s390_emit_compare_and_swap.
32964 (s390_expand_cs_hqi): Likewise, make static.
32965 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32966 compare-and-swap, in some cases.
32967 (s390_expand_cs): Wrapper function.
32968 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32970 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32971 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32972 patterns for small and large integers. Forbid symref memory operands.
32973 Move expander to s390.c. Require cc register.
32974 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32975 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32976 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32977 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32978 symref memory operands. Remove CC mode and call s390_match_ccmode
32980 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32982 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32984 * config/s390/s390.md (define_peephole2): New peephole to help
32985 combining the load-and-test pattern with volatile memory.
32987 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32989 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32990 with CCZmode for TARGET_Z196.
32992 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32994 PR rtl-optimization/80501
32995 * combine.c (make_compound_operation_int): Set subreg_code to SET
32996 even for AND with mask of the sign bit of mode.
32998 PR rtl-optimization/80500
32999 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33000 sum's initial value.
33002 2017-04-25 Julian Brown <julian@codesourcery.com>
33003 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33005 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33007 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33009 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33011 2017-04-25 Julian Brown <julian@codesourcery.com>
33012 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33014 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33015 (thunderx2t99_sha): New Reservation.
33017 2017-04-25 Julian Brown <julian@codesourcery.com>
33018 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33020 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33021 type for 1-element load.
33023 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33025 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33027 2017-04-24 Martin Jambor <mjambor@suse.cz>
33029 PR tree-optimization/80293
33030 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33031 char arrays not totally scalarizable if it is false.
33032 (analyze_all_variable_accesses): Pass correct value in the new
33033 parameter. Add a statistics counter.
33035 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33037 PR middle-end/79931
33038 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33040 2017-04-24 Richard Biener <rguenther@suse.de>
33042 PR tree-optimization/80494
33043 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33044 out for complex types.
33046 2017-04-24 Richard Biener <rguenther@suse.de>
33048 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33049 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33050 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33051 (DFS): Adjust and never fail.
33052 (sccvn_dom_walker::fail): Remove.
33053 (sccvn_dom_walker::before_dom_children): Adjust.
33054 (run_scc_vn): Likewise and never fail.
33055 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33056 (pass_fre::execute): Likewise.
33058 2017-04-24 Richard Biener <rguenther@suse.de>
33060 PR tree-optimization/79725
33061 * tree-ssa-sink.c (statement_sink_location): Return whether
33062 failure reason was zero uses. Move that check later.
33063 (sink_code_in_bb): Deal with zero uses by removing the stmt
33066 2017-04-24 Richard Biener <rguenther@suse.de>
33069 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33070 pointer-based references.
33072 2017-04-24 Richard Biener <rguenther@suse.de>
33075 * pass_manager.h (pass_manager::operator new): Remove.
33076 (pass_manager::operator delete): Likewise.
33077 * passes.c (pass_manager::operator new): Remove.
33078 (pass_manager::operator delete): Likewise.
33079 (pass_manager::pass_manager): Zero individual pass members.
33081 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33084 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33085 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33086 Check "XEXP (src, 1)" operand here.
33087 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33088 Check "XEXP (src, 1)" operand here.
33089 (dimode_scalar_chain::make_vector_copies): Detect count register
33090 of a shift instruction. Zero extend count register from QImode
33091 to DImode to satisfy vector shift pattern count operand predicate.
33092 Substitute vector shift count operand with a DImode copy.
33093 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33096 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33098 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33099 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33100 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33101 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33102 (UNSPEC_NOREX_MEM): Remove definition.
33104 2017-04-21 Richard Biener <rguenther@suse.de>
33106 PR tree-optimization/79547
33107 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33108 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33109 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33110 without any constraints.
33112 2017-04-21 Richard Biener <rguenther@suse.de>
33114 PR tree-optimization/78847
33115 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33117 2017-04-21 Richard Biener <rguenther@suse.de>
33119 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33120 (build_distinct_type_copy): Likewise.
33121 (build_variant_type_copy): Likewise.
33122 * tree.c (build_qualified_type): Pass down mem-stat info.
33123 (build_distinct_type_copy): Likewise.
33124 (build_variant_type_copy): Likewise.
33126 2017-04-21 Richard Biener <rguenther@suse.de>
33128 PR tree-optimization/80237
33129 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33131 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33132 for a simplified result.
33134 2016-04-21 Richard Biener <rguenther@suse.de>
33136 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33137 sth as strict as a simple_iv but a chrec without symbols and an
33138 operand defined in the loop we are peeling (and not some subloop).
33139 (propagate_constants_for_unrolling): Propagate all constants.
33141 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33144 * config/i386/i386.c (print_reg): Remove assert for disalowed
33145 regno values, call output_operand_lossage instead.
33147 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33150 * config/i386/constraints.md (Yc): New register constraint.
33151 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33152 Use Yc constraint for alternative 2 of operand 0. Remove
33153 preferred_for_speed attribute.
33155 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33157 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33158 lastprivate clauses in SIMT case.
33160 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33162 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33164 2017-04-20 Richard Biener <rguenther@suse.de>
33166 PR tree-optimization/57796
33167 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33168 as N scalar stores.
33169 (vect_model_load_cost): Cost gathers as N scalar loads.
33171 2017-04-20 Richard Biener <rguenther@suse.de>
33173 * ggc-page.c (ggc_allocated_p): Rename to ...
33174 (safe_lookup_page_table_entry): ... this and return the lookup
33176 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33178 2017-04-20 Richard Biener <rguenther@suse.de>
33180 PR tree-optimization/80453
33181 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33182 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33183 from the conditions.
33184 (vn_phi_eq): Pass them down.
33185 (vn_phi_lookup): Record them.
33186 (vn_phi_insert): Likewise.
33188 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33190 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33191 uninitialized variable warning to avoid buffer overrun.
33193 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33196 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33197 is suppressed for '{ 0 }' in C.
33199 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33201 * BASE-VER: Set to 8.0.0.
33203 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33205 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33206 priority .init_array and .fini_array section with SECTION_NOTYPE
33209 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33211 PR middle-end/80423
33212 * tree.h (build_array_type): Add typeless_storage default argument.
33213 * tree.c (type_cache_hasher::equal): Also compare
33214 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33215 (build_array_type): Add typeless_storage argument, set
33216 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33218 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33219 (build_array_type): Likewise. Add typeless_storage argument.
33221 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33222 Jakub Jelinek <jakub@redhat.com>
33224 PR tree-optimization/80426
33225 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33226 operation on symbolic operands, also compute the overflow for the
33227 invariant part when the operation degenerates into a negation.
33229 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33232 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33233 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33236 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33238 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33241 * config/avr/avr.c (tree.h): Include it.
33242 (cgraph.h): Include it.
33243 (avr_encode_section_info): Don't warn for uninitialized progmem
33244 variable if it's just an alias.
33246 2017-04-19 Richard Biener <rguenther@suse.de>
33249 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33250 when needed by AutoPGO.
33252 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33255 * doc/lto.texi: Remove an extra 'that'.
33257 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33259 PR rtl-optimization/80429
33260 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33261 are only used in debug insns.
33263 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33264 Vladimir Makarov <vmakarov@redhat.com>
33266 * config/sparc/predicates.md (input_operand): Add comment. Return
33267 true for any memory operand when LRA is in progress.
33268 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33270 2017-04-18 Jeff Law <law@redhat.com>
33273 * mips.md ({return,simple_return}_internal): Do not overwrite
33276 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33278 PR tree-optimization/80443
33279 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33280 instead of adding 1, subtract -1 and similarly instead of subtracting
33283 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33285 PR rtl-optimization/80357
33286 * haifa-sched.c (tmp_bitmap): New variable.
33287 (model_recompute): Handle duplicate use records.
33288 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33289 (free_global_sched_pressure_data): Free it.
33291 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33294 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33295 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33296 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33297 instead of SYSTEM_HEADER_DIR.
33299 2017-04-18 Jeff Law <law@redhat.com>
33301 PR middle-end/80422
33302 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33303 predecessors after walking up the insn chain.
33305 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33308 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33309 sizetype type into debug info.
33311 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33314 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33315 unneeded test for TARGET_UPPER_REGS_SF.
33316 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33318 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33321 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33322 instead of gsi_after_labels.
33324 2017-04-18 Jeff Law <law@redhat.com>
33326 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33330 2017-04-13 Jeff Law <law@redhat.com>
33331 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33332 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33334 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33337 * config/avr/avr.c (intl.h): Include it.
33338 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33340 2017-04-18 Martin Liska <mliska@suse.cz>
33342 PR gcov-profile/78783
33343 * gcov-tool.c (gcov_output_files): Validate that destination
33344 file is either removed by the tool or by a user.
33346 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33347 Guy Benyei <guybe@mellanox.com>
33349 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33350 block, and do not negate it, the stored id is already negative.
33352 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33354 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33356 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33359 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33360 masks of options that should be turned off if the VSX vector
33361 options are turned off.
33362 (OTHER_P8_VECTOR_MASKS): Likewise.
33363 (OTHER_VSX_VECTOR_MASKS): Likewise.
33364 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33365 rs6000_disable_incompatible_switches to validate no type switches
33367 (rs6000_incompatible_switch): New function to disallow turning on
33368 other vector options if -mno-vsx, -mno-power8-vector, or
33369 -mno-power9-vector are specified.
33371 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33373 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33375 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33377 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33378 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33379 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33380 (ARG_POINTER_CFA_OFFSET): Likewise.
33382 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33384 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33385 conditions to take advantage of various optimizations.
33387 2017-04-13 Jeff Law <law@redhat.com>
33389 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33390 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33391 (zero_extendsidi2_dext): Likewise.
33393 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33396 * fold-const.c (fold_ternary_loc): Revert
33397 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33400 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33402 PR rtl-optimization/80343
33403 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33404 new scratch pseudo.
33406 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33409 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33410 to ubsan_encode_value.
33412 2017-04-13 Jeff Law <law@redhat.com>
33414 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33415 appearing in DEBUG_INSNs.
33417 2017-04-13 Martin Liska <mliska@suse.cz>
33419 PR gcov-profile/80413
33420 * gcov-io.c (gcov_write_string): Copy to buffer just when
33421 allocated size is greater than zero.
33423 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33426 * dwarf2out.c (decls_for_scope): Ignore declarations of
33427 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33429 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33432 * ipa-visibility.c (non_local_p): Fix typos.
33433 (localize_node): When localizing symbol in same comdat group,
33434 dissolve the group only when we know external symbols are going
33436 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33438 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33440 PR tree-optimization/79390
33441 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33442 order does not result in usable sequence, retry with reversed operand
33448 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33449 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33450 op0 instead of fold_convert_loc (loc, type, arg0).
33452 2017-04-12 Jeff Law <law@redhat.com>
33454 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33455 has a delay slot in the generated code.
33457 * config/cris/cris.md (cris_preferred_reload_class): Return
33458 GENNONACR_REGS rather than GENERAL_REGS.
33460 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33463 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33464 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33465 signedness of the result type.
33467 2017-04-12 Richard Biener <rguenther@suse.de>
33468 Jeff Law <law@redhat.com>
33470 PR tree-optimization/80359
33471 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33472 trim stores to TARGET_MEM_REFs.
33474 2017-04-12 Richard Biener <rguenther@suse.de>
33476 PR tree-optimization/79390
33477 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33478 threading case even more.
33480 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33483 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33484 for quad_address_p for TImode, instead of just not indexed_address.
33486 2017-04-12 Richard Biener <rguenther@suse.de>
33487 Bernd Edlinger <bernd.edlinger@hotmail.de>
33489 PR middle-end/79671
33490 * alias.c (component_uses_parent_alias_set_from): Handle
33491 TYPE_TYPELESS_STORAGE.
33492 (get_alias_set): Likewise.
33493 * tree-core.h (tree_type_common): Add typeless_storage flag.
33494 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33495 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33496 for types containing members with TYPE_TYPELESS_STORAGE.
33497 (place_field): Likewise.
33498 (layout_type): Likewise for ARRAY_TYPE.
33499 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33500 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33501 TYPE_TYPELESS_STORAGE.
33502 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33504 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33507 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33508 first argument to type.
33510 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33514 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33515 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33516 (rs6000_expand_binop_builtin): Likewise.
33517 (rs6000_expand_ternop_builtin): Likewise; also add missing
33518 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33519 vshasigma built-ins.
33520 * doc/extend.texi: Document that vec_xxpermdi's third argument
33521 must be a constant.
33523 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33525 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33526 Use shift_const cost parameter when calculating gain of STV shifts.
33528 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33530 PR rtl-optimization/70478
33531 * lra-constraints.c (process_alt_operands): Check memory for
33532 disfavoring memory insn operand.
33534 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33536 PR middle-end/80100
33537 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33538 left shift in unsigned HOST_WIDE_INT type.
33540 PR rtl-optimization/80385
33541 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33542 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33545 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33546 if they have any depend clauses.
33548 2017-04-11 Martin Liska <mliska@suse.cz>
33551 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33552 * ipa-split.c (split_function): Create a local comdat symbol
33553 if caller is in a comdat group.
33555 2017-04-11 Martin Liska <mliska@suse.cz>
33558 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33561 2017-04-11 Martin Sebor <msebor@redhat.com>
33563 PR middle-end/80364
33564 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33565 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33567 (directive::set_width, directive::set_precision, format_character):
33569 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33572 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33575 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33576 conflict, set target->arch_name instead of target->cpu_name.
33578 2017-04-11 Richard Biener <rguenther@suse.de>
33580 PR tree-optimization/80374
33581 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33582 build_zero_cst, remove fold_convertible_p check again.
33584 2017-04-11 Martin Liska <mliska@suse.cz>
33587 * ubsan.c (instrument_object_size): Do not instrument register
33590 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33593 * config/i386/i386-builtin-types.def
33594 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33595 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33596 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33597 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33598 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33599 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33600 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33601 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33602 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33603 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33604 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33605 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33606 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33607 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33608 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33609 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33610 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33611 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33612 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33613 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33614 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33615 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33616 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33617 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33618 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33619 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33620 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33621 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33622 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33623 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33624 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33625 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33626 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33627 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33628 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33629 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33630 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33631 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33632 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33633 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33634 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33635 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33636 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33637 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33638 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33639 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33641 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33642 flag to second_arg_count, handle 4 argument function type _COUNT
33643 aliases, handle second_arg_count on second argument rather than last.
33645 2017-04-10 Jeff Law <law@redhat.com>
33647 PR tree-optimization/80374
33648 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33649 record anything if we can not convert integer_zero_node to the
33652 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33655 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33656 Enhance special handling given to the TARGET_P9_MINMAX option in
33657 relation to certain other options.
33659 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33661 PR tree-optimization/80153
33662 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33663 remove POINTER_PLUS_EXPR's base part directly, rather than through
33666 2017-04-10 Richard Biener <rguenther@suse.de>
33667 Bin Cheng <bin.cheng@arm.com>
33669 PR tree-optimization/80153
33670 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33671 the first element of pointer type aff_tree. Build result expr in
33673 (add_elt_to_tree): Convert to type unconditionally. Remove other
33674 fold_convert calls.
33675 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33676 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33678 2017-04-10 Richard Biener <rguenther@suse.de>
33680 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33683 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33685 PR rtl-optimization/70478
33686 * lra-constraints.c (curr_small_class_check): New.
33687 (update_and_check_small_class_inputs): New.
33688 (process_alt_operands): Update curr_small_class_check. Disfavor
33689 alternative insn memory operands. Check available regs for small
33692 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33695 * config/mips/mips.opt (-mvirt): Update description.
33696 * doc/invoke.texi (-mvirt): Likewise.
33698 2017-04-10 Richard Biener <rguenther@suse.de>
33700 PR middle-end/80362
33701 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33702 looking for NEGATE_EXPR in -A / -B to A / B folding.
33704 2017-04-10 Martin Liska <mliska@suse.cz>
33706 PR gcov-profile/80224
33707 * gcov.c (print_usage): Fix usage string.
33708 (get_gcov_intermediate_filename): Remove.
33709 (output_gcov_file): Use both for normal and intermediate format.
33710 (generate_results): Do not initialize special file for
33711 intermediate format.
33713 2017-04-10 Richard Biener <rguenther@suse.de>
33715 PR tree-optimization/80304
33716 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33719 2017-04-10 Nathan Sidwell <nathan@acm.org>
33722 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33723 (rs6000_init_builtins): Use it.
33725 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33727 * config/arm/arm.md (<mrc>): Add mode to SET source.
33728 (<mrrc>): Likewise.
33730 2017-04-10 Richard Biener <rguenther@suse.de>
33732 PR middle-end/80344
33733 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33735 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33738 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33739 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33740 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33741 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33742 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33743 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33744 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33745 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33746 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33747 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33748 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33749 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33750 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33751 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33752 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33753 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33754 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33755 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33756 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33757 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33758 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33759 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33760 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33762 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33764 PR rtl-optimization/70478
33765 * lra-constraints.c: Reverse the last patch.
33767 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33769 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33770 Add comment for WCHAR_T.
33772 2017-04-08 Martin Liska <mliska@suse.cz>
33775 2017-04-07 Martin Liska <mliska@suse.cz>
33778 * ipa-split.c (split_function): Add function part to a same comdat
33781 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33784 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33786 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33788 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33789 (vec_pairing): ...to this.
33790 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33791 (rs6000_sched_init): Adjust for name change.
33792 (struct rs6000_sched_context): Likewise.
33793 (rs6000_init_sched_context): Likewise.
33794 (rs6000_set_sched_context): Likewise.
33796 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33802 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33804 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33805 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33806 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33808 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33810 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33812 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33814 PR rtl-optimization/70703
33815 * ira-color.c (update_conflict_hard_regno_costs): Use
33816 int64_t instead of HOST_WIDE_INT.
33818 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33820 PR rtl-optimization/70478
33821 * lra-constraints.c (process_alt_operands): Disfavor alternative
33822 insn memory operands.
33824 2017-04-07 Jeff Law <law@redhat.com>
33826 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33827 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33829 2017-04-07 Martin Liska <mliska@suse.cz>
33832 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33833 Show error message instead of an ICE.
33835 2017-04-07 Martin Liska <mliska@suse.cz>
33838 * ipa-split.c (split_function): Add function part to a same comdat
33841 2017-04-07 Richard Biener <rguenther@suse.de>
33843 PR middle-end/80341
33844 * tree.c (get_unwidened): Also handle ! for_type case for
33846 * convert.c (do_narrow): Split out from ...
33847 (convert_to_integer_1): ... here. Do not pass final truncation
33848 type to get_unwidened for TRUNC_DIV_EXPR.
33850 2017-04-07 Richard Biener <rguenther@suse.de>
33852 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33854 (aff_combination_const): Adjust.
33855 (aff_combination_scale): Likewise.
33856 (aff_combination_add_elt): Likewise.
33857 (aff_combination_add_cst): Likewise.
33858 (aff_combination_convert): Likewise.
33859 (add_elt_to_tree): Likewise. Remove unused argument.
33860 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33862 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33864 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33866 * config/arm/arm.c (arm_default_short_enums): Use
33867 ARM_DEFAULT_SHORT_ENUMS.
33868 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33870 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33873 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33874 members with redundant out-of-class redeclaration.
33876 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33879 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33880 * config/i386/i386.md (*zero_extendsidi2):
33881 Add (?*x,*x) and (?*v,*v) alternatives.
33883 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33886 * config/i386/i386.c (ix86_expand_builtin)
33887 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33888 mode from insn data. Convert operands to insn operand mode.
33889 Copy operands that don't satisfy insn predicate to a register.
33891 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33893 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33896 2017-04-06 Richard Biener <rguenther@suse.de>
33898 PR tree-optimization/80334
33899 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33900 preserve alignment of accesses.
33902 2017-04-06 Richard Biener <rguenther@suse.de>
33904 PR tree-optimization/80262
33905 * tree-sra.c (build_ref_for_offset): Preserve address-space
33907 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33908 Drop useless address-space information on MEM_REF offsets.
33910 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33912 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33914 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33916 PR rtl-optimization/70703
33917 * ira-color.c (update_conflict_hard_regno_costs): Use
33918 HOST_WIDE_INT instead of long.
33920 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33923 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33924 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33926 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33927 for x86_64 target. Handle -m3dnowa option.
33929 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33931 PR rtl-optimization/70703
33932 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33933 (update_conflict_hard_regno_costs): Use long instead of unsigned
33934 arithmetic for cost calculation.
33936 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33937 Bernd Edlinger <bernd.edlinger@hotmail.de>
33940 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33943 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33946 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33947 ptr_mode with Pmode throughout.
33948 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33949 into probe_stack_range and use DImode.
33951 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33954 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33955 call_eh_return is true.
33957 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33959 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33960 Initialize last_match_fntype_index.
33962 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33965 * tree-nvr.c: Include internal-fn.h.
33966 (pass_return_slot::execute): Ignore internal calls without
33969 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33970 Richard Biener <rguenther@suse.de>
33973 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33974 captures used multiple times, except for the last use.
33975 * generic-match-head.c: Include gimplify.h.
33977 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33979 PR tree-optimization/79390
33980 * target.h (struct noce_if_info): Declare.
33981 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33982 * target.def (noce_conversion_profitable_p): New target hook.
33983 * ifcvt.h (struct noce_if_info): New type, moved from ...
33984 * ifcvt.c (struct noce_if_info): ... here.
33985 (noce_conversion_profitable_p): Renamed to ...
33986 (default_noce_conversion_profitable_p): ... this. No longer
33988 (noce_try_store_flag_constants, noce_try_addcc,
33989 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33990 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33991 instead of noce_conversion_profitable_p.
33992 * config/i386/i386.c: Include ifcvt.h.
33993 (ix86_option_override_internal): Don't override
33994 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33995 (ix86_noce_conversion_profitable_p): New function.
33996 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33997 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33998 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33999 * doc/tm.texi: Regenerated.
34001 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34003 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34006 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34009 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34010 instructions for small multiply cores.
34012 2017-04-04 Jeff Law <law@redhat.com>
34014 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34016 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34017 that are not set by the loop over the elements.
34019 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34022 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34023 int mode, convert_modes it to mode as unsigned, otherwise use
34024 lowpart_subreg to mode rather than SImode.
34025 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34026 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34027 Use DImode instead of SImode for the shift count operand.
34028 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34031 2017-04-04 Richard Biener <rguenther@suse.de>
34033 PR middle-end/80281
34034 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34035 arithmetic done for the negate or the plus. Simplify.
34036 (A - (-B) -> A + B): Likewise.
34037 * fold-const.c (split_tree): Make sure to not negate pointers.
34039 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34041 PR rtl-optimization/60818
34042 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34043 a compare of comparisons with the thing compared if this results
34044 in a different machine mode.
34046 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34048 * alias.c (base_alias_check): Fix typo in comment.
34049 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34050 * cgraphunit.c (symbol_table::compile): Likewise.
34051 * collect2.c (maybe_run_lto_and_relink): Likewise.
34052 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34053 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34054 * config/avr/avr.c (avr_map_op_t): Likewise.
34055 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34056 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34057 * config/epiphany/epiphany.md (movcc): Likewise.
34058 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34059 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34061 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34062 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34063 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34064 * config/sh/sh.c (sh_rtx_costs): Likewise.
34065 * fold-const.c (fold_truth_andor): Likewise.
34066 * genautomata.c (collapse_flag): Likewise.
34067 * gengtype.h (struct type::u::s): Likewise.
34068 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34069 * input.c (FORMAT_AMOUNT): Likewise.
34070 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34071 (known_aggs_to_agg_replacement_list): Likewise.
34072 * ipa-inline-analysis.c: Likewise.
34073 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34074 * ipa-polymorphic-call.c
34075 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34076 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34077 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34079 * modulo-sched.c (apply_reg_moves): Likewise.
34080 * omp-expand.c (build_omp_regions_1): Likewise.
34081 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34082 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34083 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34084 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34085 * value-prof.c: Likewise.
34086 * var-tracking.c (val_reset): Likewise.
34088 2017-04-03 Richard Biener <rguenther@suse.de>
34090 PR tree-optimization/80275
34091 * fold-const.c (split_address_to_core_and_offset): Handle
34094 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34096 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34097 descriptors is at least equal to that of functions.
34099 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34101 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34103 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34106 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34107 (mov<IMOD4:mode>): New expander.
34108 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34110 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34112 PR rtl-optimization/79405
34113 * fwprop.c (propagations_left): New variable.
34114 (forward_propagate_into): Decrement it.
34115 (fwprop_init): Initialize it.
34116 (fw_prop): If the variable has reached zero, stop propagating.
34117 (fwprop_addr): Ditto.
34119 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34122 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34123 a FUNCTION_DECL, pass it as decl instead of origin to
34126 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34128 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34131 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34134 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34135 TARGET_VSX_SMALL_INTEGER.
34137 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34139 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34140 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34142 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34144 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34145 extraction from odd-numbered MSA register.
34147 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34149 PR middle-end/80173
34150 * expmed.c (store_bit_field_1): Don't attempt to create
34151 a word subreg out of hard registers wider than word if they
34152 have HARD_REGNO_NREGS of 1 for their mode.
34154 PR middle-end/80163
34155 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34156 conversions to integer types wider than word and pointer.
34159 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34160 (rtx_equal_for_cselib_p): Pass 0 to it.
34161 * cselib.c (cselib_hasher::equal): Likewise.
34162 (rtx_equal_for_cselib_1): Add depth argument. If depth
34163 is 128, don't look up VALUE locs and punt. Increment
34164 depth in recursive calls when walking VALUE locs.
34166 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34168 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34169 (make_gcov_file_name): Use the canonical path name for generating
34171 (read_line): Fix handling of files with ascii null bytes.
34173 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34175 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34176 to initialise a vector register instead
34177 of using a const_int.
34179 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34181 PR translation/80189
34182 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34183 diagnostic messages.
34185 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34188 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34189 (dfp_diex_<mode>): Update mode of operand 1.
34190 * doc/extend.texi (dxex, dxexq): Document change to return type.
34191 (diex, diexq): Document change to argument type.
34193 2017-03-30 Martin Jambor <mjambor@suse.cz>
34196 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34197 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34198 it reflects the signature changes performed at the callee side.
34199 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34200 to cgraph_build_function_type_skip_args.
34201 (build_function_decl_skip_args): Adjust call to the above function.
34203 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34206 * config/i386/sse.md
34207 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34208 register as dest whenever it is a MEM not rtx_equal_p to the
34209 corresponding dup operand, and when forcing into reg move the
34210 reg into the memory afterwards.
34211 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34212 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34213 for the force_reg mode.
34214 (avx512vl_vextractf128<mode>): Use register as dest either
34215 always when a MEM, or when it is a MEM not rtx_equal_p to the
34216 corresponding dup operand, or even not when it is a CONST_VECTOR
34217 depending on the mode and lo vs. hi.
34218 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34220 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34221 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34222 Likewise. Require that operands[2] is even.
34223 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34224 Remove extraneous parens. Require that operands[2] is a multiple
34226 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34227 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34228 disallow memory then.
34230 2017-03-30 Richard Biener <rguenther@suse.de>
34232 PR tree-optimization/77498
34233 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34234 to non-constants over backedges.
34236 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34238 PR rtl-optimization/80233
34239 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34240 as last_combined_insn. Do not test for BARRIER_P separately.
34242 2017-03-29 Andreas Schwab <schwab@suse.de>
34245 * calls.c (prepare_call_address): Convert funexp to Pmode before
34246 copying to temp reg.
34248 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34250 PR tree-optimization/80158
34251 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34252 Handle possible future case of more than one alternate
34254 (replace_rhs_if_not_dup): Likewise.
34255 (replace_one_candidate): Likewise.
34257 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34259 PR rtl-optimization/80193
34260 * ira.c (ira): Do not check allocation for LRA.
34262 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34264 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34265 (nvptx_output_simt_exit): Declare.
34266 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34267 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34268 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34269 (nvptx_declare_function_name): ...here. Emit declaration of local
34270 memory space buffer for omp_simt_enter insn.
34271 (nvptx_output_unisimt_switch): New.
34272 (nvptx_output_softstack_switch): New.
34273 (nvptx_output_simt_enter): New.
34274 (nvptx_output_simt_exit): New.
34275 * config/nvptx/nvptx.h (struct machine_function): New fields
34276 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34277 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34278 (UNSPECV_SIMT_EXIT): Ditto.
34279 (omp_simt_enter_insn): New insn.
34280 (omp_simt_enter): New expansion.
34281 (omp_simt_exit): New insn.
34282 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34284 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34285 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34286 (expand_GOMP_SIMT_EXIT): New.
34287 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34288 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34289 (GOMP_SIMT_EXIT): Ditto.
34290 * target-insns.def (omp_simt_enter): New insn.
34291 (omp_simt_exit): Ditto.
34292 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34294 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34295 (lower_rec_input_clauses): Likewise.
34296 (lower_lastprivate_clauses): Handle SIMT privatization.
34298 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34299 (ompdevlow_adjust_simt_enter): New.
34300 (find_simtpriv_var_op): New.
34301 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34302 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34304 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34305 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34306 (copy_decl_for_dup_finish): Ditto.
34308 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34310 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34313 * config/i386/i386.c (ix86_option_override_internal): Always
34314 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34316 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34318 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34320 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34322 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34323 mark new edge's irreducible flag accordign to it.
34324 (vect_do_peeling): Check loop preheader edge's irreducible flag
34325 and pass it to function slpeel_add_loop_guard.
34327 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34329 PR tree-optimization/80218
34330 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34331 Update block frequencies and counts.
34333 2017-03-28 Richard Biener <rguenther@suse.de>
34335 PR tree-optimization/78644
34336 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34337 of a simplification result we may not use it at all.
34339 2017-03-28 Richard Biener <rguenther@suse.de>
34342 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34343 without arguments, generate default definition of a SSA name.
34345 2017-03-28 Richard Biener <rguenther@suse.de>
34347 PR middle-end/80222
34348 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34349 TYPE_REF_CAN_ALIAS_ALL references.
34350 * fold-const.c (fold_indirect_ref_1): Likewise.
34352 2017-03-28 Martin Liska <mliska@suse.cz>
34355 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34356 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34358 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34359 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34361 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34362 (EXTRA_SPECS): Define.
34363 (SUBTARGET_EXTRA_SPECS): Likewise.
34364 (SUBTARGET_CPP_SPEC): Likewise.
34365 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34366 SUBTARGET_EXTRA_SPECS.
34367 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34369 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34371 * config/arc/simdext.md (vst64_insn): Update pattern.
34372 (vld32wh_insn): Likewise.
34373 (vld32wl_insn): Likewise.
34374 (vld64_insn): Likewise.
34375 (vld32_insn): Likewise.
34377 2017-03-28 Marek Polacek <polacek@redhat.com>
34380 * fold-const.c (fold_comparison): Use protected_set_expr_location
34381 instead of SET_EXPR_LOCATION.
34383 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34385 * tree.c (add_expr): Avoid name lookup warning.
34387 2017-03-27 Jeff Law <law@redhat.com>
34389 PR tree-optimization/80216
34390 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34391 function name. Limit recursion depth.
34392 (record_temporary_equivalences): Corresponding changes.
34394 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34396 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34399 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34402 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34404 * cfgcleanup.c (reg_note_cfa_p): New array.
34405 (insns_have_identical_cfa_notes): New function.
34406 (old_insns_match_p): Don't cross-jump in between /f
34407 and non-/f instructions. If both i1 and i2 are frame related,
34408 verify all CFA notes, their order and content.
34410 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34413 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34414 HImode and SImode with zero extend to DImode to one insn.
34415 (bswap<mode>2_extenddi): Likewise.
34416 (bswapsi2_extenddi): Likewise.
34417 (bswaphi2_extendsi): Likewise.
34418 (bswaphi2): Combine bswap HImode and SImode into one insn.
34419 Separate memory insns from swapping register.
34420 (bswapsi2): Likewise.
34421 (bswap<mode>2): Likewise.
34422 (bswaphi2_internal): Delete, no longer used.
34423 (bswapsi2_internal): Likewise.
34424 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34425 store, and gpr<-gpr swap insns.
34426 (bswap<mode>2_store): Likewise.
34427 (bswaphi2_reg): Register only splitter, combine with the splitter.
34428 (bswaphi2 splitter): Likewise.
34429 (bswapsi2_reg): Likewise.
34430 (bswapsi2 splitter): Likewise.
34431 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34432 the insns into load, store, and register/register insns.
34433 (bswapdi2_ldbrx): Likewise.
34434 (bswapdi2_load): Likewise.
34435 (bswapdi2_store): Likewise.
34436 (bswapdi2_reg): Likewise.
34438 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34440 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34441 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34443 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34446 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34448 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34449 special handling for target option conflicts between dform
34450 options (-mpower9-dform, -mpower9-dform-vector,
34451 -mpower9-dform-scalar) and -mno-direct-move.
34453 2017-03-27 Richard Biener <rguenther@suse.de>
34455 PR tree-optimization/80181
34456 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34458 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34460 * config/arc/predicates.md (move_double_src_operand): Replace the
34461 call to move_double_src_operand with a call to address_operand.
34463 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34465 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34466 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34467 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34469 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34471 * config/arc/predicates.md (long_immediate_loadstore_operand):
34472 Consider scaled addresses cases.
34474 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34476 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34477 restored when in interrupt.
34478 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34479 doesn't have delay slot.
34481 2017-03-27 Richard Biener <rguenther@suse.de>
34484 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34485 inlined thunk clones.
34487 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34490 * asan.c (instrument_derefs): Copy over last operand from
34491 original COMPONENT_REF to the new COMPONENT_REF with
34492 DECL_BIT_FIELD_REPRESENTATIVE.
34493 * ubsan.c (instrument_object_size): Likewise.
34495 2017-03-27 Richard Biener <rguenther@suse.de>
34497 PR tree-optimization/80170
34498 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34499 sure DR/SCEV didnt fold in constants we do not see when looking
34500 at the reference base alignment.
34502 2017-03-27 Richard Biener <rguenther@suse.de>
34504 PR middle-end/80171
34505 * gimple-fold.c (fold_ctor_reference): Properly guard against
34506 NULL return value from canonicalize_constructor_val.
34508 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34511 * config/i386/i386.c (ix86_expand_builtin)
34512 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34513 flags reg setting and flags reg using instructions.
34514 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34515 clobbering instructions to zero extend op2.
34517 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34519 * doc/install.texi (Configuration) <--with-aix-soname>:
34520 Update link to AIX ld.
34522 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34524 PR rtl-optimization/80160
34525 PR rtl-optimization/80159
34526 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34527 reg_alternate_class into account.
34529 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34532 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34533 to consider in curr_insn_transform.
34535 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34537 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34538 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34539 and emit_mode_inner.
34541 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34543 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34544 argument to the overloaded builtin variants. Use the new flag to
34545 deprecate certain builtin variants.
34546 * config/s390/s390-builtin-types.def: Add new builtin types.
34547 * config/s390/s390-builtins.h: Support new flags field for
34548 overloaded builtins.
34549 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34550 (s390_macro_to_expand): Enable vector float data type.
34551 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34552 builtins by incrementing the __VEC__ version number.
34553 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34555 (s390_resolve_overloaded_builtin): Emit error messages depending
34556 on the builtin flags.
34557 * config/s390/s390.c (s390_expand_builtin): Support additional
34558 flags argument. Change error message to match the messages
34559 emitted in s390-c.c.
34560 * config/s390/s390.md: New UNSPEC_* constants.
34561 (op_type): Add new instruction types.
34562 * config/s390/vecintrin.h: Add new builtins and test data class
34564 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34565 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34566 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34567 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34568 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34569 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34571 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34572 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34573 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34574 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34576 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34577 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34578 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34579 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34581 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34582 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34583 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34584 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34585 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34586 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34587 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34589 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34590 ("vec_scatter_element<V_HW_4:mode>_DI")
34591 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34592 ("vec_fpint<mode>", "vflls")
34593 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34594 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34595 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34596 ("*vec_cmphe<mode>_cc"): ... these.
34598 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34599 mode constant instead of magic value.
34601 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34603 * config/s390/s390.c (s390_expand_vec_compare): Support other
34604 vector floating point modes than just V2DF.
34605 (s390_expand_vcond): Likewise.
34606 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34607 (s390_cannot_change_mode_class): Prevent mode changes between TF
34608 and V1TF in vector registers.
34609 * config/s390/s390.md (DF, SF): New mode attributes.
34610 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34611 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34612 SFmode support for VRs.
34613 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34615 (VFT, VF_HW): New mode iterators.
34616 (vw, sdx): New mode attributes.
34617 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34618 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34619 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34620 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34621 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34622 also the new vector floating point modes. Renaming to ...
34624 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34625 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34626 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34627 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34628 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34629 ("vec_unordered<mode>"): ... these.
34631 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34632 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34633 ("*vec_extendv2df"): New insn definitions.
34635 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34637 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34638 ("mulditi3_2", "*muldi3_sign"): New patterns.
34639 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34640 rename the pattern definition.
34642 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34644 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34646 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34648 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34650 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34651 instruction if possible.
34652 * config/s390/vector.md (vec_halfnumelts): New mode
34654 ("*vec_vllezlf<mode>"): New pattern.
34656 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34658 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34659 ("popcountv4si2", "popcountv2di2"): Rename to ...
34660 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34661 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34663 ("popcount<mode>2_vxe"): New pattern.
34665 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34667 * common/config/s390/s390-common.c (processor_flags_table): Add
34669 * config.gcc: Add arch12.
34670 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34671 Default to arch12 for unknown CPU model numbers.
34672 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34673 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34674 PROCESSOR_max sanity check.
34675 * config/s390/s390-opts.h (enum processor_type): Add
34677 * config/s390/s390.c (processor_table): Add arch12.
34678 (s390_expand_builtin): Add check for B_VXE flag.
34679 (s390_issue_rate): Add PROCESSOR_ARCH12.
34680 (s390_get_sched_attrmask): Likewise.
34681 (s390_get_unit_mask): Likewise.
34682 (s390_sched_score): Enable z13 scheduling for arch12.
34683 (s390_sched_reorder): Likewise.
34684 (s390_sched_variable_issue): Likewise.
34685 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34687 (s390_tune_attr): Use z13 scheduling also for arch12.
34688 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34689 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34690 (TARGET_VXE_P): New macros.
34691 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34692 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34693 * config/s390/s390.opt: Add arch12 as processor_type.
34695 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34697 * config/s390/s390.md
34698 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34699 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34700 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34702 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34703 Rename expanders to ...
34705 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34706 ("fixuns_truncdddi2_emu"): ... these.
34708 ("fixuns_trunc<mode>si2_emu"): New expander.
34710 ("*fixuns_truncdfdi2_z13"): Rename to ...
34711 ("*fixuns_truncdfdi2_vx"): ... this.
34713 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34715 * config/s390/2964.md: Remove the single element vector compare
34716 instructions which are no longer used.
34717 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34719 (s390_canonicalize_comparison): Remove handling of DFmode
34721 (s390_expand_vec_compare_scalar): Remove function.
34722 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34723 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34725 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34727 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34729 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34731 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34732 will anyway by matched by mov<mode>_64dfp.
34734 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34736 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34737 vlef/vstef. Add missing operand to vleif.
34739 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34741 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34742 pair for all vector types with 64 bit elements.
34743 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34744 * config/s390/vector.md (V_HW_64): ... here.
34745 (V_128_NOSINGLE): New mode iterator.
34746 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34747 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34748 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34749 ("*vec_load_pairv2di"): Change to ...
34750 ("*vec_load_pair<mode>"): ... this one.
34752 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34754 * config/s390/constraints.md: Add comments.
34755 (jKK): Reject element sizes > 8 bytes.
34756 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34758 * config/s390/s390.md: Add the s_operand checks formerly in
34759 s390_split_ok_p to various splitters where they are still
34761 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34762 for 128 bit vectors. Plus two splitters.
34764 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34766 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34769 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34772 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34773 error if the boundary argument is not constant.
34775 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34777 PR rtl-optimization/80112
34778 * loop-doloop.c (doloop_condition_get): Don't check condition
34779 if cmp isn't SET with IF_THEN_ELSE src.
34781 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34783 PR tree-optimization/80158
34784 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34785 replacing a candidate statement, also replace it for the
34786 candidate's alternate interpretation.
34787 (replace_rhs_if_not_dup): Likewise.
34788 (replace_one_candidate): Likewise.
34790 2017-03-24 Richard Biener <rguenther@suse.de>
34792 PR tree-optimization/80167
34793 * graphite-isl-ast-to-gimple.c
34794 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34796 (translate_isl_ast_to_gimple::get_rename): Likewise.
34798 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34800 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34801 handling of certain combinations of target options, including the
34802 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34803 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34805 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34808 * config/arm/arm.md (*load_multiple): Add reload_completed to
34809 matching condition.
34811 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34812 Richard Biener <rguenth@suse.de>
34814 PR tree-optimization/79908
34815 PR tree-optimization/80136
34816 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34817 been cast away, gimplify_and_add suffices.
34819 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34821 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34823 2017-03-23 Richard Biener <rguenther@suse.de>
34825 PR tree-optimization/80032
34826 * gimplify.c (gimple_push_cleanup): Forced unconditional
34827 cleanups still have to go to the conditional_cleanups
34830 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34832 PR tree-optimization/80072
34833 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34835 (next_operand_entry_id): Change type to unsigned int.
34836 (sort_by_operand_rank): Make sure to return the right return value
34837 even if unsigned fields are bigger than INT_MAX.
34838 (struct oecount): Change cnt and id type to unsigned int.
34839 (oecount_hasher::equal): Formatting fix.
34840 (oecount_cmp): Make sure to return the right return value
34841 even if unsigned fields are bigger than INT_MAX.
34842 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34845 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34846 TREE_READONLY on result if writing it more than once.
34849 * doc/invoke.texi (-fsanitize=thread): Document that with
34850 -fnon-call-exceptions atomics are not able to throw
34854 * tsan.c: Include tree-eh.h.
34855 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34856 maybe_clean_or_replace_eh_stmt where needed.
34857 (instrument_memory_accesses): Add cfg_changed argument.
34858 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34859 if it returned true.
34860 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34862 PR rtl-optimization/63191
34863 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34864 wrapper function, moved the whole old content into ...
34865 (ix86_delegitimize_address_1): ... this. New inline function.
34866 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34867 true as last argument instead of ix86_delegitimize_address.
34869 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34871 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34872 cortexa57_branch_cost.
34874 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34876 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34878 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34881 * doc/md.texi (Constraints): Document wA constraint.
34882 * config/rs6000/constraints.md (wA): New.
34883 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34884 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34885 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34886 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34888 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34891 * gimplify.c (is_oacc_declared): New function.
34892 (oacc_default_clause): Use it to set default flags for acc declared
34893 variables inside parallel regions.
34894 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34895 declared variables.
34896 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34897 declare attribute to any decl as necessary.
34899 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34902 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34903 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34904 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34905 (arm_arch_lpae): This.
34906 * config/arm/arm.c (arm_arch7ve): Rename into ...
34907 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34908 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34911 2017-03-22 Martin Liska <mliska@suse.cz>
34914 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34915 error message instead of an ICE.
34917 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34919 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34921 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34923 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34925 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34928 2017-03-21 Martin Sebor <msebor@redhat.com>
34930 * doc/extend.texi: Use "cannot" instead of "can't."
34931 * doc/hostconfig.texi: Same.
34932 * doc/install.texi: Same.
34933 * doc/invoke.texi: Same.
34934 * doc/loop.texi: Same.
34935 * doc/md.texi: Same.
34936 * doc/objc.texi: Same.
34937 * doc/rtl.texi: Same.
34938 * doc/tm.texi: Same.
34939 * doc/tm.texi.in: Same.
34940 * doc/trouble.texi: Same.
34942 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34945 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34946 (collect_checksum_attributes): Set it.
34947 (die_checksum_ordered): Use it.
34949 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34951 PR tree-optimization/79908
34952 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34953 change: For a VA_ARG whose LHS has been cast away, use
34954 force_gimple_operand to construct the side effects.
34956 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34958 PR translation/80001
34959 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34960 more amenable to translation.
34961 (oacc_loop_auto_partitions): Likewise.
34963 2017-03-21 Marek Polacek <polacek@redhat.com>
34964 Martin Sebor <msebor@redhat.com>
34966 PR tree-optimization/80109
34967 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34968 on INTEGRAL_TYPE_P.
34970 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34971 Segher Boessenkool <segher@kernel.crashing.org>
34974 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34975 check reg_used_between_p between insn and one of succ or succ2
34976 depending on if succ is artificial insn not inserted into insn
34979 2017-03-21 Martin Liska <mliska@suse.cz>
34981 PR gcov-profile/80081
34982 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34983 * doc/gcc.texi: Include gcov-dump stuff.
34984 * doc/gcov-dump.texi: New file.
34986 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34988 PR rtl-optimization/79150
34989 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34990 conditional jump, if the jump is the last insn of the loop.
34992 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34993 Richard Biener <rguenth@suse.de>
34995 PR tree-optimization/79908
34996 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34997 been cast away, use force_gimple_operand to construct the side
35000 2017-03-21 Martin Liska <mliska@suse.cz>
35002 PR libfortran/79956
35003 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35006 2017-03-21 Brad Spengler <spender@grsecurity.net>
35009 * plugin.c (htab_hash_plugin): New function.
35010 (add_new_plugin): Use it and adjust.
35011 (parse_plugin_arg_opt): Adjust.
35012 (init_one_plugin): Likewise.
35014 2017-03-21 Richard Biener <rguenther@suse.de>
35016 PR tree-optimization/80032
35017 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35018 if set force the cleanup to happen unconditionally.
35019 (gimplify_target_expr): Push inserted clobbers with force_uncond
35020 to avoid them being removed by control-dependent DCE.
35022 2017-03-21 Richard Biener <rguenther@suse.de>
35024 PR tree-optimization/80122
35025 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35026 va_arg_pack_len when the inlined call stmt requires pack
35028 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35030 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35033 * tsan.c (instrument_builtin_call): If the memory model argument
35034 is not a constant, assume it is valid.
35037 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35040 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35042 PR rtl-optimization/79910
35043 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35044 if its dest is used by an insn before I2 (other than the combined
35045 insns themselves, which are properly handled already).
35047 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35050 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35052 * combine.c (record_used_regs): New static function.
35053 (try_combine): Handle situations where there is an additional
35054 instruction between I2 and I3 which needs to have a LOG_LINK
35058 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35060 * combine.c (try_combine): Delete redundant i1 test. Call
35061 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35063 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35066 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35067 alternatives 13/14.
35069 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35071 PR tree-optimization/80054
35072 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35073 the optimization if a PHI or any of its arguments is not dominated
35074 by the candidate's basis. Use gphi* rather than gimple* as
35076 (replace_profitable_candidates): Clean up a gimple* variable that
35077 should be a gphi* variable.
35079 2017-03-20 Martin Sebor <msebor@redhat.com>
35082 * doc/extend.texi (attribute constructor): Document present limitation.
35084 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35087 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35088 __POWER9_VECTOR__ #ifdef control, change template definition to
35089 use Power9-specific built-in function.
35090 (vec_any_eq): Likewise.
35091 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35092 to control outcomes from this test.
35093 (vector_ae_<mode>p): For VEC_F modes, likewise.
35095 2017-03-20 Ian Lance Taylor <iant@google.com>
35097 * config/i386/i386.c (ix86_function_regparm): Save an extra
35098 register for -fsplit-stack with DECL_STATIC_CHAIN.
35100 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35103 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35104 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35106 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35108 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35110 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35113 2017-03-20 Marek Polacek <polacek@redhat.com>
35116 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35118 2017-03-20 Richard Biener <rguenther@suse.de>
35120 PR tree-optimization/80113
35121 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35122 allocate extra SSA name for PHI def.
35123 (add_close_phis_to_outer_loops): Likewise.
35124 (add_close_phis_to_merge_points): Likewise.
35125 (copy_loop_close_phi_args): Likewise.
35126 (copy_cond_phi_nodes): Likewise.
35128 2017-03-20 Martin Liska <mliska@suse.cz>
35130 PR middle-end/79753
35131 * tree-chkp.c (chkp_build_returned_bound): Do not build
35132 returned bounds for a LHS that's not a BOUNDED_P type.
35134 2017-03-20 Martin Liska <mliska@suse.cz>
35138 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35139 COMPLEX_CST and VECTOR_CST.
35141 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35144 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35145 target operand. A new splitter adds the clobber statement in case
35146 the target operand is dead anyway.
35148 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35150 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35151 to age-old versions of binutils and glibc.
35153 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35155 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35157 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35159 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35161 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35163 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35164 requirement for binutils 2.13.
35166 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35168 * combine.c (try_combine): Delete redundant i1 test. Call
35169 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35171 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35173 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35174 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35176 <riscv64-*-elf>: Re-arrange section
35177 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35178 <riscv32-*-linux>: Likewise.
35179 <riscv64-*-elf>: Likewise
35180 <riscv64-*-linux>: Likewise.
35182 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35185 * aarch64.opt(verbose-cost-dump): Fix typo.
35187 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35190 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35191 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35193 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35195 * reload.c (find_reloads): When reloading a nonoffsettable address,
35196 use RELOAD_OTHER for it and its address reloads.
35198 PR rtl-optimization/79910
35199 * combine.c (record_used_regs): New static function.
35200 (try_combine): Handle situations where there is an additional
35201 instruction between I2 and I3 which needs to have a LOG_LINK
35204 2017-03-17 Jeff Law <law@redhat.com>
35206 PR tree-optimization/71437
35207 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35208 conditional in the hash table first.
35209 (vrp_dom_walker::before_dom_children): Extract condition from
35210 ASSERT_EXPR. Record condition, its inverion and any implied
35211 conditions as well.
35213 2017-03-17 Marek Polacek <polacek@redhat.com>
35214 Markus Trippelsdorf <markus@trippelsdorf.de>
35216 PR tree-optimization/80079
35217 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35220 2017-03-17 Richard Biener <rguenther@suse.de>
35222 PR middle-end/80075
35223 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35224 Properly verify the LHS before the RHS possibly claims to be
35226 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35229 2017-03-17 Martin Jambor <mjambor@suse.cz>
35231 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35232 (List of -O2 options): Likewise.
35233 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35236 2017-03-17 Tom de Vries <tom@codesourcery.com>
35238 * gcov-dump.c (print_usage): Print bug_report_url.
35240 2017-03-17 Richard Biener <rguenther@suse.de>
35242 PR middle-end/80050
35243 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35244 (parser::peek): Likewise.
35246 2017-03-17 Richard Biener <rguenther@suse.de>
35248 PR tree-optimization/80048
35249 * sese.c (free_sese_info): Properly release rename_map and
35250 copied_bb_map elements.
35252 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35254 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35255 Add linked-list forward and backlinks. Insert on
35256 construction, remove on destruction.
35257 (class pass_store_merging): Add m_stores_head field.
35258 (pass_store_merging::terminate_and_process_all_chains):
35259 Iterate over m_stores_head list.
35260 (pass_store_merging::terminate_all_aliasing_chains):
35262 (pass_store_merging::execute): Check for debug stmts first.
35263 Push new chains onto the m_stores_head stack.
35265 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35268 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35269 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35270 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35272 2017-03-16 Jeff Law <law@redhat.com>
35274 PR tree-optimization/71437
35275 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35276 member function. Implementation moved into after_dom_children
35277 member function and into the threader's thread_outgoing_edges
35279 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35280 some code into new thread_outgoing_edges.
35281 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35282 definition. Simplify marker handling (do it here). Assume we always
35283 have the available expression and the const/copies tables.
35284 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35286 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35287 * tree-vrp.c (equiv_stack): No longer file scoped.
35288 (vrp_dom_walker): New class.
35289 (vrp_dom_walker::before_dom_children): New member function.
35290 (vrp_dom_walker::after_dom_children): Likewise.
35291 (identify_jump_threads): Setup domwalker. Use it rather than
35292 walking edges in a random order by hand. Simplify setup/finalization.
35293 (finalize_jump_threads): Remove.
35294 (vrp_finalize): Do not call identify_jump_threads here.
35295 (execute_vrp): Do it here instead and call thread_through_all_blocks
35298 PR tree-optimization/71437
35299 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35301 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35303 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35304 (dom_opt_dom_walker::thread_across_edge): Remove
35305 handle_dominating_asserts argument. All callers changed.
35306 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35307 changes. Remove calls to lhs_of_dominating_assert. Other
35308 uses of handle_dominating_asserts turn into unconditional code
35309 (simplify_control_stmt_condition_1): Likewise.
35310 (simplify_control_stmt_condition): Likewise.
35311 (thread_through_normal_block, thread_across_edge): Likewise.
35312 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35313 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35314 object if it is not an SSA_NAME.
35315 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35316 before calling into the VRP specific simplifiers.
35317 (identify_jump_threads): Remove handle_dominating_asserts
35320 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35323 * tree-diagnostic.c (default_tree_printer): No longer static.
35324 * tree-diagnostic.h (default_tree_printer): New prototype.
35326 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35328 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35329 Change ins into fmov.
35331 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35333 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35334 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35335 Use h_con constraint for operand 1.
35336 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35337 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35339 2017-03-15 Jeff Law <law@redhat.com>
35341 PR tree-optimization/71437
35342 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35343 (record_temporary_equivalences): Use it.
35345 PR tree-optimization/71437
35346 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35347 tree-ssa-scopedtables.
35348 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35349 (record_conditions, record_cond, vuse_eq): Likewise.
35350 (record_edge_info): Adjust to API tweak of record_conditions.
35351 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35352 (record_temporary_equivalences, optimize_stmt): Likewise.
35353 (eliminate_redundant_computations): Likewise.
35354 (record_equivalences_from_stmt): Likewise.
35355 * tree-ssa-scopedtables.c: Include options.h and params.h.
35356 (vuse_eq): New function, moved from tree-ssa-dom.c
35357 (build_and_record_new_cond): Likewise.
35358 (record_conditions): Likewise. Accept vector of conditions rather
35359 than edge_equivalence structure for first argument.
35360 for the first argument.
35361 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35362 from tree-ssa-dom.c.
35363 (avail_exprs_stack::record_cond): Likewise.
35364 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35365 from tree-ssa-dom.c.
35366 (avail_exprs_stack): Add new member functions lookup_avail_expr
35368 (record_conditions): Declare.
35370 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35373 * lra-constraints.c (process_alt_operands): Increase reject for
35374 reloading an input/output operand.
35376 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35379 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35380 insns to convert from signed/unsigned char/short to IEEE 128-bit
35382 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35384 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35387 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35388 subreg of inner mode for values already in registers.
35390 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35392 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35393 iteration reg is used after the loop.
35395 2017-03-14 Martin Sebor <msebor@redhat.com>
35397 PR tree-optimization/79800
35398 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35399 precision in negative-positive range.
35400 (format_floating): Call non-const overload with adjusted precision.
35402 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35405 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35408 2017-03-14 Martin Sebor <msebor@redhat.com>
35410 PR middle-end/80020
35411 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35412 * builtins.def (aligned_alloc): Use it.
35415 * Makefile.in (GTFILES): Add calls.c.
35416 * calls.c: Include "gt-calls.h".
35418 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35420 PR rtl-optimization/79728
35421 * regs.h (struct target_regs): New field
35422 x_contains_allocatable_regs_of_mode.
35423 (contains_allocatable_regs_of_mode): New macro.
35424 * reginfo.c (init_reg_sets_1): Initialize it, and change
35425 contains_reg_of_mode so it includes global regs as well.
35426 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35427 rather than contains_regs_of_mode.
35429 2017-03-14 Martin Liska <mliska@suse.cz>
35431 * doc/invoke.texi: Document options that can't be combined with
35432 -fcheck-pointer-bounds.
35434 2017-03-14 Martin Liska <mliska@suse.cz>
35436 PR middle-end/79831
35437 * doc/invoke.texi (-Wchkp): Document the option.
35439 2017-03-14 Martin Liska <mliska@suse.cz>
35441 * Makefile.in: Install gcov-dump.
35443 2017-03-14 Martin Liska <mliska@suse.cz>
35445 * multiple_target.c (expand_target_clones): Bail out for
35446 an invalid attribute.
35448 2017-03-14 Richard Biener <rguenther@suse.de>
35450 * alias.c (struct alias_set_entry): Pack properly.
35451 * cfgloop.h (struct loop): Likewise.
35452 * cse.c (struct set): Likewise.
35453 * ipa-utils.c (struct searchc_env): Likewise.
35454 * loop-invariant.c (struct invariant): Likewise.
35455 * lra-remat.c (struct cand): Likewise.
35456 * recog.c (struct change_t): Likewise.
35457 * rtl.h (struct address_info): Likewise.
35458 * symbol-summary.h (function_summary): Likewise.
35459 * tree-loop-distribution.c (struct partition): Likewise.
35460 * tree-object-size.c (struct object_size_info): Likewise.
35461 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35462 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35463 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35464 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35465 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35466 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35467 (struct _stmt_vec_info): Likewise.
35469 2017-03-14 Martin Liska <mliska@suse.cz>
35472 * multiple_target.c (create_dispatcher_calls): Check that
35473 a target can create a function dispatcher.
35475 2017-03-14 Martin Liska <mliska@suse.cz>
35478 * multiple_target.c (expand_target_clones): Drop local.local
35479 flag for default implementation.
35481 2017-03-14 Richard Biener <rguenther@suse.de>
35483 PR tree-optimization/80030
35484 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35486 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35488 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35489 gcc_fallthrough() instead of __attribute__((fallthrough));
35491 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35493 * doc/gcc.texi: Remove "up" link to (DIR).
35494 * doc/gccint.texi: Ditto.
35496 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35498 * doc/install.texi (Specific) <avr>: Remove reference to
35501 2017-03-13 Jeff Law <law@redhat.com>
35503 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35504 attribute rather than comments.
35506 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35507 match_scratch operand is highest.
35509 2017-03-13 Martin Liska <mliska@suse.cz>
35511 PR middle-end/78339
35512 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35513 is a CHKP clone, use original declaration.
35515 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35517 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35518 (arc_conditional_register_usage): Use a different allocation order
35519 when optimizing for size.
35520 * common/config/arc/arc-common.c (arc_option_optimization_table):
35521 Section anchors default on when optimizing for size.
35523 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35525 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35527 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35529 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35530 * config/arc/arc.md (cpu_facility): Add cd variant.
35531 (*movqi_insn): Add code density variant.
35532 (*movhi_insn): Likewise.
35533 (*movqi_insn): Likewise.
35534 (*addsi3_mixed): Likewise.
35535 (subsi3_insn): Likewise.
35537 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35539 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35541 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35543 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35544 expressions with MINUS and UNARY ops.
35546 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35549 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35551 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35552 between vec_select and vector argument.
35553 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35554 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35555 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35556 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35557 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35558 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35560 2017-03-13 Richard Biener <rguenther@suse.de>
35563 * params.def (vect-max-peeling-for-alignment): Fix typo.
35565 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35567 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35568 issue that only occurred with binutils below 2.18.
35570 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35572 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35573 refer to binutils 2.11/2.12 minimum.
35575 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35577 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35578 ftp.kernel.org and simplify binutils requirement.
35580 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35582 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35584 (Optimize Options): Ditto. Also remove redundancy.
35586 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35588 PR translation/79848
35589 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35591 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35592 to G_ to avoid double translation.
35594 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35596 PR translation/79923
35597 * auto-profile.c (get_combined_location): Convert leading
35598 character of diagnostics to lower case and remove trailing period.
35599 (read_profile): Likewise for various diagnostics.
35600 * config/arm/arm.c (arm_option_override): Remove trailing period
35601 from various diagnostics.
35602 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35603 (msp430_expand_delay_cycles): Likewise.
35605 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35608 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35609 full command-line argument, rather than just "str".
35610 (aarch64_validate_march): Likewise.
35611 (aarch64_validate_mtune): Likewise.
35613 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35615 PR rtl-optimization/78911
35616 * lra-assigns.c (must_not_spill_p): New function.
35617 (spill_for): Use it.
35619 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35621 PR tree-optimization/79981
35622 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35623 ATOMIC_COMPARE_EXCHANGE ifn result.
35624 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35625 IFN_ATOMIC_COMPARE_EXCHANGE.
35627 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35630 * opts.c (parse_sanitizer_options): Add missing question mark to
35631 "did you mean" message.
35633 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35635 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35637 (VMULEUH_UNS): Likewise.
35638 (VMULOUB_UNS): Likewise.
35639 (VMULOUH_UNS): Likewise.
35640 * config/rs6000/rs6000.c (builtin_function_type): Remove
35641 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35643 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35646 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35647 x with result of extra_parsing_for_operand_code_0.
35648 (function_reader::extra_parsing_for_operand_code_0): Convert
35649 return type from void to rtx, returning x. When reading
35650 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35651 larger size containing struct block_symbol.
35653 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35655 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35656 -mfloat128-hardware without -m64.
35658 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35661 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35662 entries to the case statement that marks unsigned arguments to
35663 overloaded functions.
35665 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35667 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35668 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35670 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35673 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35674 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35676 2017-03-10 Martin Liska <mliska@suse.cz>
35680 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35681 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35684 2017-03-10 Olivier Hainque <hainque@adacore.com>
35686 * tree-switch-conversion (array_value_type): Start by resetting
35687 candidate type to it's main variant.
35689 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35691 PR rtl-optimization/79909
35692 * combine.c (try_combine): Use simplify_replace_rtx on individual
35693 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35694 of the whole CALL_INSN_FUNCTION_USAGE.
35696 PR tree-optimization/79972
35697 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35698 get_range_info on SSA_NAMEs. Formatting fixes.
35700 2017-03-10 Richard Biener <rguenther@suse.de>
35701 Jakub Jelinek <jakub@redhat.com>
35703 PR tree-optimization/77975
35704 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35705 edge to be constant.
35706 (get_val_for): For constant x return it. Formatting fix.
35707 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35708 would use the same bases as the current one.
35710 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35712 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35713 instead of vec_select for V1TImode.
35714 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35716 (VSX_LE_128): Add V1TI to this mode iterator.
35717 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35718 (*vsx_le_perm_store_<mode>): Likewise.
35719 (pre-reload splitter for VSX stores): Likewise.
35720 (post-reload splitter for VSX stores): Likewise.
35721 (*vsx_xxpermdi2_le_<mode>): Likewise.
35722 (*vsx_lxvd2x2_le_<mode>): Likewise.
35723 (*vsx_stxvd2x2_le_<mode>): Likewise.
35725 2017-03-09 Michael Eager <eager@eagercon.com>
35727 Correct failures with --enable-checking=yes,rtl.
35729 * config/microblaze/microblaze.c (microblaze_expand_shift):
35730 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35731 test for const0_rtx.
35732 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35733 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35735 2017-03-09 Richard Biener <rguenther@suse.de>
35737 PR tree-optimization/79977
35738 * graphite-scop-detection.c (scop_detection::merge_sese):
35739 Handle the case of extra exits to blocks dominating the entry.
35741 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35743 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35746 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35748 PR rtl-optimization/79949
35749 * lra-constraints.c (process_alt_operands): Check memory when
35750 trying to predict a cycle. Print about the overall increase.
35752 2017-03-09 Richard Biener <rguenther@suse.de>
35754 PR middle-end/79971
35755 * gimple-expr.c (useless_type_conversion_p): Preserve
35756 TYPE_SATURATING for fixed-point types.
35758 2017-03-09 Richard Biener <rguenther@suse.de>
35761 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35762 alignment of BLKmode params.
35764 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35767 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35768 (VALL_NO_V2Q): Likewise.
35770 * config/aarch64/aarch64-simd.md
35771 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35773 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35774 VALL_NO_V2Q mode iterator.
35775 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35777 2017-03-09 Martin Liska <mliska@suse.cz>
35779 PR tree-optimization/79631
35780 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35781 tree_int_cst_sign_bit just for INTEGER constants.
35783 2017-03-09 Martin Liska <mliska@suse.cz>
35787 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35790 2017-03-09 Marek Polacek <polacek@redhat.com>
35793 * tree.c (inchash::add_expr): Handle TREE_VEC.
35795 2017-03-09 Martin Liska <mliska@suse.cz>
35798 (chkp_narrow_size_and_offset): New function.
35799 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35800 (void chkp_parse_bit_field_ref): New function.
35801 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35802 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35804 2017-03-09 Martin Liska <mliska@suse.cz>
35807 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35808 (chkp_find_bounds_1): Remove gcc_unreachable.
35810 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35813 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35814 BUILT_IN_SYNC*, determine the access type from the size suffix and
35815 always build a MEM_REF with that type. Handle forgotten
35816 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35819 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35820 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35821 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35822 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35823 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35824 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35825 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35826 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35827 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35828 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35829 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35830 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35831 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35832 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35833 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35834 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35835 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35836 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35837 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35838 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35839 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35840 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35841 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35842 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35843 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35844 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35845 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35846 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35847 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35848 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35849 definitions outside of __OPTIMIZE__ guarded section.
35852 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35853 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35854 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35855 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35858 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35860 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35861 ("vfenez<mode>"): Add missing constraints.
35863 2017-03-08 Martin Sebor <msebor@redhat.com>
35866 * config/nds32/nds32.c (nds32_option_override):
35867 Fix misspelled diagnostic.
35869 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35872 * gimplify.c (gimplify_omp_for): Replace index var in outer
35873 taskloop statement with an artificial variable and add
35874 OMP_CLAUSE_PRIVATE clause for it.
35876 2017-03-08 Richard Biener <rguenther@suse.de>
35878 PR tree-optimization/79955
35879 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35880 for accesses that are completely outside of the variable.
35882 2017-03-08 Andrew Haley <aph@redhat.com>
35884 PR tree-optimization/79943
35885 * tree-ssa-loop-split.c (compute_new_first_bound): When
35886 calculating the new upper bound, (END-BEG) should be added, not
35889 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35891 * config/avr/avr.md (setmemhi): Make sure match_dup
35892 operand number comes before match_scratch.
35894 2017-03-08 Richard Biener <rguenther@suse.de>
35896 PR tree-optimization/79920
35897 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35898 with ncopies == 1 to ...
35899 (vect_transform_slp_perm_load): ... here. Properly compute
35900 all element loads by iterating VF times over the group. Do
35901 not handle ncopies (computed in a broken way) in
35902 vect_create_mask_and_perm.
35904 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35907 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35908 is a uniform vector, use uniform_vector_p return value instead of
35909 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35911 2017-03-07 Marek Polacek <polacek@redhat.com>
35913 PR middle-end/79809
35914 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35915 (alloca_call_type): Likewise.
35917 2017-03-07 Martin Liska <mliska@suse.cz>
35919 * gcov.c (process_args): Put comment to correct location.
35921 2017-03-07 Martin Liska <mliska@suse.cz>
35923 PR middle-end/68270
35924 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35925 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35926 (chkp_narrow_bounds_for_field): Likewise.
35927 (chkp_parse_array_and_component_ref): Pass one more argument to
35930 2017-03-07 Richard Biener <rguenther@suse.de>
35932 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35935 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35937 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35938 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35940 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35943 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35944 to end of description.
35945 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35947 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35949 PR rtl-optimization/79901
35950 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35952 (*avx512f_<code><mode>3<mask_name>): ... this.
35953 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35954 iterator instead of VI8_AVX2_AVX512BW.
35956 PR rtl-optimization/79901
35957 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35958 min/max expander, expand it using expand_vec_cond_expr.
35961 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35964 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35967 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35968 to void * for PCH reasons.
35969 * dwarf2out.c (output_loc_operands, output_die): Cast
35970 v.val_vec.array to unsigned char *.
35972 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35975 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35978 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35980 PR rtl-optimization/79571
35981 * lra-constraints.c (process_alt_operands): Calculate static
35982 reject and subtract it from overall when only addresses will be
35985 2017-03-06 Julia Koval <julia.koval@intel.com>
35988 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35989 incoming stack boundary to 128 for 64-bit targets.
35991 2017-03-06 Richard Biener <rguenther@suse.de>
35993 PR tree-optimization/79894
35994 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35995 to NULL after folding it.
35997 2017-03-06 Richard Biener <rguenther@suse.de>
35999 PR tree-optimization/79824
36000 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36001 check disabling peeling for gaps.
36003 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36005 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36006 attributes): Document gettimeofday.
36008 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36010 * config/s390/s390.c (s390_option_override_internal): Set
36011 PARAM_MIN_VECT_LOOP_BOUND
36013 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36015 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36016 * config/s390/s390.md: Likewise.
36018 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36021 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36022 (<avx2_avx512>_perm<mode>): Rename to ...
36023 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36025 (<avx512>_perm<mode>_mask): Rename to ...
36026 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36028 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36029 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36030 instead of VI8F_256_512.
36031 (avx512f_perm<mode>): New define_expand.
36032 (avx512f_perm<mode>_mask): Likewise.
36033 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36034 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36036 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36038 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36039 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36041 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36043 2017-03-06 Martin Liska <mliska@suse.cz>
36046 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36047 when having a SSA NAME w/o VAR_DECL assigned to it.
36049 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36051 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36052 msa_dpsub_<su>_d): Fix MODE for vec_select.
36054 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36056 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36058 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36060 2017-03-06 Richard Biener <rguenther@suse.de>
36062 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36063 * plugin.c (register_plugin_info): Likewise.
36064 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36066 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36068 * config/i386/sse.md (sse_storehps, sse_storelps,
36069 avx_<castmode><avxsizesuffix>_<castmode>,
36070 avx512f_<castmode><avxsizesuffix>_<castmode>,
36071 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36072 in condition that at least one operand is not a MEM.
36074 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36076 PR middle-end/79805
36077 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36078 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36080 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36081 gimple_call_nothrow_p flag based on whether original builtin can throw.
36082 If it can, emit following stmts on the fallthrough edge.
36083 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36084 don't create new bb if inserting just debug stmts on the edge, try to
36085 insert them on the fallthru bb or just reset debug stmts.
36087 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36090 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36091 restore recog_data (including the operand rtxes inside it) around
36092 the call to get_insn_template.
36094 2017-03-03 Martin Sebor <msebor@redhat.com>
36096 PR tree-optimization/79699
36097 * context.c (context::~context): Free MPFR caches to avoid
36098 a memory leak on program exit.
36100 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36102 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36103 Use wide_int::ulow () instead of .elt (0).
36105 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36107 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36108 (*pushxf): Limit oF constraint to 32bit targets and add oC
36109 constraint for 64bit targets.
36110 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36111 (*pushdf): Change rmF constraint to rmC.
36113 2017-03-03 Martin Liska <mliska@suse.cz>
36115 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36116 Remove unused variable.
36118 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36121 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36122 is a memory operand, increase num_memory.
36123 (ix86_expand_args_builtin): Likewise.
36125 2017-03-03 Jan Hubicka <jh@suse.cz>
36128 * ipa-devirt.c (maybe_record_node): Properly handle
36129 __cxa_pure_virtual visibility.
36131 2017-03-03 Martin Liska <mliska@suse.cz>
36133 PR tree-optimization/79803
36134 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36136 (pass_loop_prefetch::execute): Disabled optimization if an
36137 assumption about L1 cache size is not met.
36139 2017-03-03 Martin Liska <mliska@suse.cz>
36141 PR rtl-optimization/79574
36142 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36143 (hash_scan_set): Likewise.
36144 (dump_hash_table): Likewise.
36145 (hoist_code): Likewise.
36147 2017-03-03 Richard Biener <rguenther@suse.de>
36149 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36150 in place of uhigh ().
36151 (fixed_convert_from_real): Likewise.
36153 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36156 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36158 2017-03-03 Richard Biener <rguenther@suse.de>
36160 PR middle-end/79818
36161 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36162 TYPE_OVERFLOW_UNDEFINED check.
36164 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36166 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36168 (vector_ae_<mode>_p): Likewise.
36169 (vector_nez_<mode>_p): Likewise.
36170 (vector_ne_v2di_p): Likewise.
36171 (vector_ae_v2di_p): Likewise.
36172 (vector_ne_<mode>_p): Likewise.
36173 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36175 (vsx_tsqrt<mode>2_fe): Likewise.
36177 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36180 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36182 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36184 PR rtl-optimization/79780
36185 * cprop.c (one_cprop_pass): When second and further conditional trap
36186 in a single basic block is turned into an unconditional trap, turn it
36187 into a deleted note to avoid RTL verification failures.
36189 2017-03-02 Richard Biener <rguenther@suse.de>
36191 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36193 2017-03-02 Richard Biener <rguenther@suse.de>
36195 PR tree-optimization/79345
36197 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36198 param and abort the walk, returning -1 if it is hit.
36199 (walk_aliased_vdefs): Take a limit param and pass it on.
36200 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36201 defaulting to 0 and return a signed int.
36202 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36203 (check_defs): New helper.
36204 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36205 about uninitialized memory.
36206 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36207 bogus uninitialized warning.
36208 (fixed_convert_from_real): Likewise.
36210 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36212 PR tree-optimization/66768
36213 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36214 iv_use if base object can't be determined.
36216 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36218 PR tree-optimization/79345
36219 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36220 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36221 (get_pattern_stats): Initialize it.
36222 * genemit.c (gen_expand): Verify match_scratch numbers come after
36223 match_operand/match_dup numbers.
36224 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36225 match_scratch numbers.
36226 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36228 * config/s390/s390.md (trunctdsd2): Likewise.
36230 2017-03-02 Richard Biener <rguenther@suse.de>
36232 * wide-int.h (wide_int_storage::operator=): Implement in terms
36235 2017-03-02 Richard Biener <rguenther@suse.de>
36237 PR tree-optimization/79777
36238 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36239 the to insert expression to sth existing.
36241 2017-03-01 Martin Sebor <msebor@redhat.com>
36243 PR middle-end/79692
36244 * gimple-ssa-sprintf.c
36245 (directive::known_width_and_precision): New function.
36246 (format_integer): Use it.
36247 (get_mpfr_format_length): Consider the full range of precision
36248 when computing %g output with the # flag. Set the likely byte
36249 count to 3 rather than 1 when precision is indeterminate.
36250 (format_floating): Correct the lower bound of precision.
36252 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36254 * doc/invoke.texi: Document default code model for 64-bit Linux.
36256 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36259 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36260 udiv rather than div since input pattern is unsigned.
36262 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36264 * config/i386/i386.c (print_reg): Warn for values of
36265 unsupported size in integer register.
36267 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36270 * config/rs6000/predicates.md (current_file_function_operand): Do
36271 not allow self calls to be local if the function is replaceable.
36273 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36276 * config/rs6000/altivec.h (vec_ctz and others): Change the
36277 preprocessor macro that controls conditional compilation from
36278 _ARCH_PWR9 to __POWER9_VECTOR__.
36279 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36280 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36281 control (instead of _ARCH_PWR9 control) so that template
36282 definition uses power9-specific function.
36283 (vec_any_eq): Likewise.
36284 (vec_all_ne): Change macro definition to use a power9-specific
36285 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36286 _ARCH_PWR9 control).
36287 (vec_any_eq) Likewise.
36288 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36289 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36290 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36291 support for xvcmpnedp instruction.
36292 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36293 macro expansion so that Power9 implementation of vec_all_ne does
36294 not use the AltiVec predicate framework.
36295 (VCMPNEH_P): Likewise.
36296 (VCMPNEW_P): Likewise.
36297 (VCMPNED_P): Likewise.
36298 (VCMPNEFP_P): Likewise.
36299 (VCMPNEDP_P): Likewise.
36300 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36301 implementation of vec_any_eq to not use AltiVec predicate
36303 (VCMPAEH_P): Likewise.
36304 (VCMPAEW_P): Likewise.
36305 (VCMPAED_P): Likewise.
36306 (VCMPAEFP_P): Likewise.
36307 (VCMPAEDP_P): Likewise.
36308 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36309 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36310 not use the AltiVec predicate framework.
36311 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36312 of vec_any_eq to not use AltiVec predicate framework.
36313 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36314 support for predefined __POWER9_VECTOR__ macro to indicate that
36315 Power9 instruction selection is enabled.
36316 (altivec_overloaded_builtins): Remove extraneous
36317 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36318 function argument types RS6000_BTI_bool_V16QI and
36319 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36320 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36321 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36322 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36323 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36324 Power9 for implementations of vec_cmpne. Change the signature for
36325 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36326 (representing vec_all_ne) to remove the previously described first
36327 argument of type RS6000_BTI_INTSI, as this was an artifact of
36328 reliance on the AltiVec predicate framework, which is no longer
36329 used in the implementation of these functions. Add
36330 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36331 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36332 since, unlike the AltiVec predicate framework implementation, we
36333 do not share function descriptors between vec_alle and vec_anyeq.
36334 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36335 set of modes that receive special treatment even when
36336 TARGET_P9_VECTOR is true. The special treatment emits code that
36337 does not depend on Power9 instructions.
36338 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36339 define_expand to not rely on AltiVec predicate framework.
36340 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36342 (vector_ne_v2di_p): Change this define_expand to not rely on
36343 AltiVec predicate framework.
36344 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36346 (vector_ne_<mode>_p): Change this define_expand to not rely on
36347 AltiVec predicate framework.
36348 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36350 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36351 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36352 define_insn pattern.
36353 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36354 define_insn pattern because the xvcmpne<VSs>. instruction is not
36356 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36357 instruction is not supported.
36359 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36361 * config/nvptx/nvptx.c: Include intl.h.
36363 2017-03-01 Martin Jambor <mjambor@suse.cz>
36366 * ipa-prop.h (ipa_bits): Removed field known.
36367 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36368 to pointers. Adjusted their comments to warn about their sharing.
36369 (ipcp_transformation_summary): Change bits to a vector of pointers.
36370 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36371 (ipa_get_ipa_bits_for_value): Declare.
36372 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36373 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36374 (ipa_bits_hash_table): Likewise.
36375 (ipa_vr_ggc_hash_traits): Likewise.
36376 (ipa_vr_hash_table): Likewise.
36377 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36378 being pointers and vr_known being removed.
36379 (ipa_set_jf_unknown): Likewise.
36380 (ipa_get_ipa_bits_for_value): New function.
36381 (ipa_set_jfunc_bits): Likewise.
36382 (ipa_get_value_range): New overloaded functions.
36383 (ipa_set_jfunc_vr): Likewise.
36384 (ipa_compute_jump_functions_for_edge): Use the above functions to
36385 construct bits and vr parts of jump functions.
36386 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36387 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36389 (ipcp_grow_transformations_if_necessary): Also allocate
36390 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36392 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36393 them. Fix too long lines.
36394 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36395 vr_known being removed.
36396 (ipa_read_jump_function): Use new setter functions to construct bits
36397 and vr parts of jump functions or set them to NULL.
36398 (write_ipcp_transformation_info): Adjust for bits being pointers.
36399 (read_ipcp_transformation_info): Likewise.
36400 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36402 Include gt-ipa-prop.h.
36403 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36405 (ipcp_store_bits_results): Likewise.
36406 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36407 Do not write to existing jump functions but use a temporary instead.
36409 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36412 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36413 attempt to use its first operand as BIT_FIELD_REF base.
36415 2017-03-01 Richard Biener <rguenther@suse.de>
36417 PR middle-end/79721
36418 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36419 interpolating formula in wrapping arithmetic.
36420 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36422 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36424 PR tree-optimization/79734
36425 * tree-vect-generic.c (expand_vector_condition): Optimize
36426 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36427 Handle VEC_COND_EXPR where comparison has different inner width from
36428 type's inner width.
36430 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36432 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36433 markup, and similar issues. Remove @opindex entries for things
36434 that aren't options. Add missing -mmpy-option entries.
36436 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36438 PR tree-optimization/79737
36439 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36440 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36441 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36442 instead of byte_size. Formatting fix.
36443 (shift_bytes_in_array_right): Formatting fix.
36445 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36448 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36449 condition on optimize for the leaf function test.
36451 2017-02-28 Martin Liska <mliska@suse.cz>
36454 * read-rtl-function.c (function_reader::handle_unknown_directive):
36455 Bail out when one uses -flto.
36457 2017-02-28 Martin Liska <mliska@suse.cz>
36459 * common.opt: Replace space with tabular for options of <number>
36461 * config/i386/i386.opt: Show <number> value for
36462 -mlarge-data-threshold.
36463 * opts.c (print_filtered_help): Do not display number in hexadecimal
36466 2017-02-28 Martin Liska <mliska@suse.cz>
36468 * common.opt: Fix --help=option -Q for options which are of
36471 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36473 * config/i386/i386.c (print_reg): Error out for values
36474 of 8-bit size in invalid integer register.
36476 2017-02-28 Martin Sebor <msebor@redhat.com>
36478 PR tree-optimization/79691
36479 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36481 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36484 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36485 gcc_unreachable with output_operand_lossage.
36487 2017-02-28 Richard Biener <rguenther@suse.de>
36489 PR tree-optimization/79740
36490 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36492 (visit_nary_op): Insert the nary into the hashtable if we
36493 pattern-matched sth.
36494 * tree-ssa-pre.c (eliminate_insert): Robustify.
36496 2017-02-28 Richard Biener <rguenther@suse.de>
36498 PR middle-end/79731
36499 * fold-const.c (decode_field_reference): Reject out-of-bound
36502 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36504 * config/i386/i386.c: Include intl.h.
36505 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36506 instead of just cond ? "..." : "...".
36507 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36508 * coverage.c (read_counts_file): Likewise.
36509 * omp-offload.c: Include intl.h.
36510 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36511 of just cond ? "..." : "...".
36512 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36513 of just cond ? "..." : "...".
36515 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36518 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36520 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36521 'tune for' CPU name.
36522 * config/arm/arm-cpu-data.h: Regenerated.
36524 2017-02-28 Richard Biener <rguenther@suse.de>
36526 PR tree-optimization/79732
36527 * tree-inline.c (expand_call_inline): Do not shadow var.
36529 2017-02-28 Richard Biener <rguenther@suse.de>
36531 PR tree-optimization/79723
36532 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36533 address-space properly.
36535 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36537 * doc/optinfo.texi (Optimization groups): Fix option used for
36539 * doc/invoke.texi (-fopt-info): Document "omp".
36540 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36541 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36542 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36543 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36544 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36546 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36548 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36549 OPTGROUP_OMP with "omp".
36551 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36554 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36555 for arithmetic shift of unsigned V2DI.
36557 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36559 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36560 arc/linux.h headers.
36561 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36562 (LINK_SPEC): Likewise.
36563 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36564 (EXTRA_SPECS): Likewise.
36565 (STARTFILE_SPEC): Likewise.
36566 (ENDFILE_SPEC): Likewise.
36567 (LIB_SPEC): Likewise.
36568 (TARGET_SDATA_DEFAULT): Likewise.
36569 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36570 (MULTILIB_DEFAULTS): Likewise.
36571 (DWARF2_UNWIND_INFO): Likewise.
36572 * config/arc/big.h: New file.
36573 * config/arc/elf.h: Likewise.
36574 * config/arc/linux.h: Likewise.
36575 * config/arc/t-uClibc: Remove.
36577 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36579 PR tree-optimization/77536
36580 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36581 (tree_transform_and_unroll_loop): Use above function to compute the
36582 estimated niter of unrolled loop and use it when scaling profile.
36583 Also use count info rather than frequency if it's non-zero.
36584 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36585 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36586 (vect_transform_loop): Call above function.
36588 2017-02-27 Richard Biener <rguenther@suse.de>
36590 PR tree-optimization/45397
36591 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36592 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36593 (visit_nary_op): Add pattern matching for CSEing sign-changed
36594 or truncated operations with wider ones.
36596 2017-02-27 Richard Biener <rguenther@suse.de>
36598 PR tree-optimization/79690
36599 * tree-vect-stmts.c (vectorizable_store): Use vector type
36600 built from the DR with address-space.
36602 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36604 * doc/invoke.texi (Optimize Options): Refine the description
36605 of asan-use-after-return.
36607 2017-02-25 Alan Modra <amodra@gmail.com>
36609 PR rtl-optimization/79584
36610 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36611 base, not ad->base_term, the reg within base. Remove assertion
36612 that ad->base == ad->base_term. Replace gen_int_mode using
36613 bogus mode with const0_rtx.
36615 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36617 PR middle-end/79396
36618 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36619 FMA_EXPR like tcc_binary or tcc_unary.
36621 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36624 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36626 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36627 (output_loc_operands): Handle DW_OP_call_ref and
36628 DW_OP_GNU_variable_value.
36629 (struct variable_value_struct): New type.
36630 (struct variable_value_hasher): Likewise.
36631 (variable_value_hash): New variable.
36632 (string_types): Remove.
36633 (copy_loc_descr): New function.
36634 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36635 (prepend_loc_descr_to_each): New function.
36636 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36637 instead of add_loc_descr_to_each if the first argument is single
36638 location list and the second has multiple.
36639 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36640 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36641 when looking for variable value which doesn't have other location info.
36642 (loc_list_from_tree): Formatting fix.
36643 (gen_array_type_die): Simplify DW_AT_string_length handling.
36644 (adjust_string_types): Remove.
36645 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36646 string_types. Call resolve_variable_values.
36647 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36648 (resolve_addr_in_expr): Likewise. Add A argument.
36649 (copy_deref_exprloc): Remove deref argument. Adjust for the
36650 original expression being DW_OP_GNU_variable_value with optionally
36651 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36652 optionally after it.
36653 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36654 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36655 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36656 (variable_value_hasher::hash, variable_value_hasher::equal): New
36658 (resolve_variable_value_in_expr, resolve_variable_value,
36659 resolve_variable_values, note_variable_value_in_expr,
36660 note_variable_value): New functions.
36661 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36664 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36667 * opts.h (handle_generated_option): Add GENERATED_P argument.
36668 * opts-common.c (handle_option): Adjust function comment.
36669 (handle_generated_option): Add GENERATED_P argument, pass it to
36671 (control_warning_option): Pass false to handle_generated_option
36673 * opts.c (maybe_default_option): Pass true to handle_generated_option
36675 * optc-gen.awk: Likewise.
36677 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36679 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36680 a REG, look at the REG it is a SUBREG of.
36681 (splitter for cmpeqsi_t): Ditto.
36683 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36685 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36686 the special USEs with the pattern of the insn, not the insn itself.
36688 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36691 * doc/invoke.texi: Document -mload-store-pairs.
36693 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36694 Sandra Loosemore <sandra@codesourcery.com>
36696 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36697 argument isn't a CONST_INT.
36698 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36699 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36700 (nios2_validate_compare): Bypass alternate compare logic if *op2
36701 is not a CONST_INT.
36702 (ldstwm_operation_p): Return false if first_base is not a REG or
36703 if first_offset is not a CONST_INT.
36705 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36707 * config/cris/cris.md: Use correct operand in a define_peephole2.
36709 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36711 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36713 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36715 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36716 this_insn if it is an INSN or JUMP_INSN.
36717 (force_offsettable): Look at base, not at addr.
36718 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36719 on things that aren't necessarily CONST_INTs.
36721 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36723 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36724 -mfpmath=sse is the default also for x86-32 targets with SSE2
36725 instruction set when @option{-ffast-math} is enabled
36727 2017-02-24 Jeff Law <law@redhat.com>
36729 PR rtl-optimizatoin/79286
36730 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36733 2017-02-24 Richard Biener <rguenther@suse.de>
36735 PR tree-optimization/79389
36736 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36739 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36741 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36742 function comment to reflect reality.
36743 (loop_exits_before_overflow): Fix typo in function description.
36745 2017-02-24 Richard Biener <rguenther@suse.de>
36747 PR tree-optimization/79389
36748 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36749 properly that a threading opportunity exists. Detect conditional
36750 copy/constant propagation opportunities.
36752 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36754 * config/visium/visium.md (type): Add trap.
36755 (b): New mode attribute.
36756 (*btst): Rename into...
36757 (*btst<mode>): ...this and adjust.
36758 (*cbranchsi4_btst_insn): Rename into...
36759 (*cbranch<mode>4_btst_insn): ...this and adjust.
36760 (trap): New define_insn.
36762 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36764 PR tree-optimization/79389
36765 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36766 (noce_reversed_cond_code): New function.
36767 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36768 reversed_comparison_code. Formatting fix.
36769 (noce_try_store_flag): Test rev_cond != NULL in addition to
36770 reversed_comparison_code.
36771 (noce_try_store_flag_constants): Likewise.
36772 (noce_try_store_flag_mask): Likewise.
36773 (noce_try_addcc): Use rev_cond if non-NULL instead of
36774 reversed_comparison_code.
36775 (noce_try_cmove_arith): Likewise. Formatting fixes.
36776 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36777 (noce_find_if_block): Initialize rev_cond.
36778 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36779 instead of false as last argument never attempt to reverse it
36782 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36784 PR tree-optimization/79663
36785 * tree-predcom.c (combine_chains): Process refs in reverse order
36786 only for ZERO length chains, and add explaining comment.
36788 2017-02-23 Jeff Law <law@redhat.com>
36790 PR tree-optimization/79578
36791 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36792 in call to operand_equal_p.
36794 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36797 * config/i386/cpuid.h: Fix another undefined behavior.
36799 2017-02-23 Richard Biener <rguenther@suse.de>
36801 PR tree-optimization/79683
36802 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36803 vector types for data-refs.
36805 2017-02-23 Martin Liska <mliska@suse.cz>
36807 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36809 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36811 PR middle-end/79665
36812 * internal-fn.c (get_range_pos_neg): Moved to ...
36813 * tree.c (get_range_pos_neg): ... here. No longer static.
36814 * tree.h (get_range_pos_neg): New prototype.
36815 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36816 are known to be in between 0 and signed maximum inclusive, try to
36817 expand both unsigned and signed divmod and use the cheaper one from
36820 2017-02-22 Jeff Law <law@redhat.com>
36822 PR tree-optimization/79578
36823 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36824 to compare base operands.
36826 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36829 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36830 gpc_reg_operand instead of fpr_reg_operand.
36832 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36834 * config/mips/mips.c (mips_return_in_memory): Force FP
36835 vector types to be returned in memory for o32 ABI.
36837 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36839 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36840 instead of DW_TAG_member for static data member declarations and don't
36841 set no_linkage_name for static inline data members.
36842 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36845 2017-02-22 Martin Liska <mliska@suse.cz>
36847 * doc/invoke.texi: Replace inequality signs with square brackets
36850 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36853 * lra-constraints.c (simplify_operand_subreg): Handle
36854 WORD_REGISTER_OPERATIONS targets.
36856 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36859 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36860 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36861 elimination by swapping fld*.
36863 2017-02-22 Richard Biener <rguenther@suse.de>
36865 PR tree-optimization/79673
36866 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36867 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36868 irrelevant address-space qualifiers and avoiding a
36869 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36871 2017-02-22 Richard Biener <rguenther@suse.de>
36873 PR tree-optimization/79666
36874 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36875 to not symbolically negate if that may introduce undefined
36878 2017-02-22 Martin Liska <mliska@suse.cz>
36881 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36882 * data-streamer-out.c (streamer_write_gcov_count_stream):
36884 * value-prof.c (stream_out_histogram_value): Make assert more
36885 precise based on type of counter.
36887 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36890 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36891 Use nonimmediate_operand instead of memory_operand for operand 1.
36892 (float-extend standard_x87sse_constant_load splitter): Ditto.
36894 2017-02-21 Jeff Law <law@redhat.com>
36896 PR tree-optimization/79621
36897 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36898 blocks with edges to themselves.
36900 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36903 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36904 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36905 Use gimple_call_builtin_p.
36908 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36909 on temporarily removed DEBUG_INSNs.
36911 PR tree-optimization/79649
36912 * tree-loop-distribution.c (classify_partition): Give up on
36913 non-generic address space loads/stores.
36915 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36917 * doc/loop.texi (Loop manipulation): Remove nonexistent
36918 tree_ssa_loop_version from the documentation.
36919 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36921 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36924 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36925 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36926 * config/rs6000/rs6000.c: Include except.h.
36927 (rs6000_expand_split_stack_prologue): Call
36928 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36930 2017-02-21 Martin Jambor <mjambor@suse.cz>
36933 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36934 have been analyzed.
36936 2017-02-21 Martin Jambor <mjambor@suse.cz>
36938 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36939 for backward compatibility only.
36940 * doc/invoke.texi (Option Summary): Remove all references to
36941 -fipa-cp-alignment.
36943 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36947 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36949 * lra-constraints.c (curr_insn_transform): Handle
36950 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36952 2017-02-21 Martin Liska <mliska@suse.cz>
36954 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36956 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36958 PR translation/79638
36959 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36961 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36964 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36965 (arm_function_ok_for_sibcall): Return false for an indirect call by
36966 descriptor if all the argument registers are used.
36967 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36968 alignment of the function.
36970 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36972 PR tree-optimization/61441
36973 * simplify-rtx.c (simplify_const_unary_operation): For
36974 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36975 the sNaN unmodified.
36977 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36979 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36980 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36981 instead of SYSTEM_HEADER_DIR.
36983 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36984 Martin Liška <mliska@suse.cz>
36986 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36987 Fix typos and grammar, use active voice, and clarify.
36989 2017-02-20 Marek Polacek <polacek@redhat.com>
36991 PR middle-end/79537
36992 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36995 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36997 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37000 * config/i386/i386.c (ix86_expand_builtin): Handle
37001 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37002 ix86_builtins_isa[fcode].isa as a requirement of those
37003 flags and any other flag in the bitmask.
37004 (ix86_init_mmx_sse_builtins): Use 0 instead of
37005 ~OPTION_MASK_ISA_64BIT as mask.
37006 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37007 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37008 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37009 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37011 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37014 * lra-constraints.c (split_reg): Check requested split mode
37015 is supported by the register.
37017 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37019 * lra-constraints.c (simplify_operand_subreg): Remove early
37022 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37025 * lra-constraints.c (curr_insn_transform): Tighten condition
37026 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37028 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37031 * lra-constraints.c (curr_insn_transform): Handle
37032 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37034 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37037 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37039 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37041 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37044 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37047 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37049 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37052 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37054 * final.c (last_columnnum, override_columnnum): New variables.
37055 (final_start_function): Set last_columnnum, pass it to begin_prologue
37056 hook and pass 0 to dwarf2out_begin_prologue.
37057 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37058 to source_line debug hook.
37059 (notice_source_line): Compute last_columnnum and for debug_column_info
37060 return true on column changes.
37061 * debug.h (struct gcc_debug_hooks): Add column argument to
37062 source_line and begin_prologue hooks.
37063 (debug_nothing_int_charstar_int_bool): Remove prototype.
37064 (debug_nothing_int_int_charstar,
37065 debug_nothing_int_int_charstar_int_bool): New prototypes.
37066 (dwarf2out_begin_prologue): Add column argument.
37067 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37068 begin_prologue hooks.
37069 (debug_nothing_int_charstar_int_bool): Remove.
37070 (debug_nothing_int_int_charstar,
37071 debug_nothing_int_int_charstar_int_bool): New functions.
37072 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37073 through to dwarf2out_source_line.
37074 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37075 (dwarf2out_source_line): Add column argument, emit it if requested.
37076 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37078 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37079 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37080 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37081 through to dwarf2out_begin_prologue.
37082 (vmsdbgout_source_line): Add column argument, pass it through to
37083 dwarf2out_source_line.
37084 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37085 dbxout_source_line caller.
37086 (dbxout_source_line): Add column argument.
37088 * common.opt (gno-column-info, gcolumn-info): New options.
37089 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37090 (check_die): Also test for multiple DW_AT_decl_column attributes.
37091 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37092 DW_AT_decl_column if requested.
37093 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37095 (gen_variable_die): Likewise.
37096 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37097 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37100 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37101 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37102 (ix86_handle_option): Handle OPT_m3dnowa.
37103 * doc/invoke.texi (-m3dnowa): Document.
37104 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37105 -m3dnowa instead of -m3dnow -march=athlon.
37108 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37109 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37111 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37114 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37115 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37116 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37117 generator for vsx_xxpermdi_<mode>_be.
37118 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37119 force big-endian semantics.
37120 (vsx_xxpermdi_<mode>_be): New define_expand with same
37121 implementation as previous version of vsx_xxpermdi_<mode>.
37123 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37125 PR tree-optimization/79327
37126 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37127 variable, its initialization and use.
37129 2017-02-17 Julia Koval <julia.koval@intel.com>
37131 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37132 (OPTION_MASK_ISA_PKU_UNSET): New.
37133 (ix86_handle_option): Handle -mrdpid.
37134 * config/i386/cpuid.h (bit_RDPID): New.
37135 * config/i386/driver-i386.c (host_detect_local_cpu):
37136 Detect RDPID feature.
37137 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37138 * config/i386/i386-c.c (ix86_target_macros_internal):
37140 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37141 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37142 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37143 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37144 * config/i386/i386.md (define_insn "rdpid"): New.
37145 * config/i386/i386.opt Add -mrdpid.
37146 * config/i386/immintrin.h (_rdpid_u32): New.
37148 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37150 PR rtl-optimization/79541
37151 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37152 instead of transforming it into USE.
37154 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37156 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37157 If HONOR_SNANS (SFmode) force the input to a register.
37158 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37159 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37160 an frsp or similar insn.
37162 2017-02-17 Martin Liska <mliska@suse.cz>
37164 PR rtl-optimization/79577
37165 * params.def (selsched-max-sched-times): Increase minimum to 1.
37167 2017-02-17 Martin Liska <mliska@suse.cz>
37169 PR rtl-optimization/79574
37170 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37172 2017-02-17 Martin Liska <mliska@suse.cz>
37174 PR tree-optimization/79529
37175 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37176 ssa_defined_default_def_p to handle cases which are implicitly
37178 * tree-ssa.c (ssa_defined_default_def_p): New function.
37179 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37180 which are implicitly defined.
37181 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37183 2017-02-17 Richard Biener <rguenther@suse.de>
37185 PR middle-end/79576
37186 * params.def (max-ssa-name-query-depth): Limit to 10.
37188 2017-02-17 Richard Biener <rguenther@suse.de>
37190 PR tree-optimization/79552
37191 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37194 2017-02-17 Richard Biener <rguenther@suse.de>
37197 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37199 2017-02-17 Marek Polacek <polacek@redhat.com>
37201 PR middle-end/79536
37202 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37203 (fold_negate_expr): New wrapper.
37205 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37207 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37208 Correct terminology and de-emphasize pre-standard behavior.
37210 2017-02-16 Alan Modra <amodra@gmail.com>
37212 PR rtl-optimization/79286
37213 * ira.c (def_dominates_uses): New function.
37214 (update_equiv_regs): Don't create an equivalence for insns that
37215 may trap where the register def does not dominate the use.
37217 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37219 PR rtl-optimization/78127
37220 * lra.c (lra): Call lra_eliminate before finish the loop after
37223 2017-02-16 Richard Biener <rguenther@suse.de>
37225 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37227 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37228 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37229 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37230 (isl_val_int_from_wi): New function.
37231 (extract_affine_gmp): Rename to ...
37232 (extract_affine_wi): ... this, take a widest_int.
37233 (extract_affine_int): Just wrap extract_affine_wi.
37234 (add_param_constraints): Use isl_val_int_from_wi.
37235 (add_loop_constraints): Likewise, and extract_affine_wi.
37237 2017-02-15 Jeff Law <law@redhat.com>
37239 PR middle-end/79521
37240 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37241 ira_init_register_move_cost_if_necessary.
37243 2017-02-15 Martin Sebor <msebor@redhat.com>
37245 PR middle-end/32003
37246 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37247 removed in a prior commit.
37249 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37251 PR tree-optimization/79347
37252 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37253 counters during peeling.
37255 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37257 * Makefile.in (site.exp): Remove "set ISLVER".
37259 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37262 * real.c (real_from_integer): Call real_convert even for decimal.
37264 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37267 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37269 2017-02-14 Andrew Pinski <apinski@cavium.com>
37271 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37272 cores and change the partno/implementer to be correct.
37273 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37274 the 'B" as the implementer.
37275 * config/aarch64/aarch64-tune.md: Regenerate.
37277 2017-02-14 Carl Love <cel@us.ibm.com>
37279 * config/rs6000/rs6000.c: Add case statement entry to make the
37280 xvcvuxdsp built-in argument unsigned.
37281 * config/rs6000/vsx.md: Fix the source and return operand types so they
37282 match the instruction definitions from the ISA document. Fix typo
37283 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37286 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37289 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37290 member early_clobber_alts.
37291 * lra-lives.c (reg_early_clobber_p): New.
37292 (process_bb_lives): Use it.
37293 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37294 (debug_operand_data): Initialize early_clobber_alts.
37295 (setup_operand_alternative): Set up early_clobber_alts.
37296 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37297 alternatives to new_insn_reg.
37298 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37300 (lra_update_insn_regno_info): Pass the new arg.
37302 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37304 PR middle-end/79505
37305 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37306 (new_oacc_loop_raw): Don't clear already cleared fields.
37309 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37310 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37311 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37313 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37316 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37318 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37321 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37322 the extra instruction to the right place to store 128-bit constant
37325 2017-02-14 Martin Sebor <msebor@redhat.com>
37327 PR middle-end/79448
37328 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37329 warning for strings of unknown length.
37331 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37333 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37335 2017-02-14 Jeff Law <law@redhat.com>
37338 * ira-costs.c (scan_one_insn): Initialize register move costs
37339 for pseudos seen in USE/CLOBBER insns.
37341 PR tree-optimization/79095
37342 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37343 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37344 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37345 if the operands are known to be not equal, then the resulting range
37347 (intersect_ranges): If the new range is ~[0,0] and the old range is
37348 wide, then prefer ~[0,0].
37349 * tree-vrp.c (overflow_comparison_p_1): New function.
37350 (overflow_comparison_p): New function.
37351 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37352 if NAME is used in an overflow test.
37353 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37354 overflow check that can be expressed as an equality test, then adjust
37355 ops to be that equality test.
37357 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37359 * config/s390/s390-builtin-types.def: Remove flags argument.
37360 * config/s390/s390.c (s390_init_builtins): Likewise.
37362 2017-02-14 Martin Liska <mliska@suse.cz>
37364 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37365 vector. Fix trailing white spaces.
37367 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37369 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37372 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37374 PR rtl-optimization/68664
37375 * config/arm/arm.c (arm_sched_can_speculate_insn):
37376 New function. Declare prototype.
37377 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37379 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37381 PR rtl-optimization/68664
37382 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37384 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37386 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37388 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37389 max skip bytes for function, loop and jump.
37391 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37393 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37394 ABS_EXPR for gimple dump.
37396 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37399 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37401 PR tree-optimization/79408
37402 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37403 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37404 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37405 also if rhs1 is INTEGER_CST.
37407 2017-02-14 Richard Biener <rguenther@suse.de>
37409 PR middle-end/79432
37410 * tree-into-ssa.c (insert_phi_nodes): When the function can
37411 have abnormal edges rewrite SSA names with broken use-def
37412 dominance out of SSA and register them for PHI insertion.
37414 2017-02-13 Martin Sebor <msebor@redhat.com>
37416 PR middle-end/79496
37417 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37418 clearing info.nowrite flag when snprintf size argument is a range.
37420 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37422 * cprop.c (cprop_jump): Add missing space in string literal.
37423 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37424 (get_constraint_for_component_ref): Likewise.
37425 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37426 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37427 * lra-constraints.c (process_alt_operands): Likewise.
37428 * ipa-inline.c (inline_small_functions): Likewise.
37429 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37430 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37431 * trans-mem.c (diagnose_tm_1_op): Likewise.
37432 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37433 (grid_parallel_clauses_gridifiable): Likewise.
37435 * config/nvptx/mkoffload.c (process): Add space in between
37438 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37439 "MOD4_SSE_REGS" and "ALL_REGS".
37441 * spellcheck.c (test_data): Add , in between "foo" and "food".
37443 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37446 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37447 boundary crossing check and subsequent code generation agree.
37449 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37451 * config/aarch64/aarch64.c (has_memory_op): Delete.
37452 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37455 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37457 PR rtl-optimization/79388
37458 PR rtl-optimization/79450
37459 * combine.c (distribute_notes): When removing TEM_INSN for which
37460 corresponding dest has last value recorded, invalidate that last
37463 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37465 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37466 of explicit '@'. Add missing assembly comment marker on branch costs
37469 2017-02-13 Nathan Sidwell <nathan@acm.org>
37471 * gengtype-lex.l (<in_struct>): Add '/'.
37473 2017-02-13 Martin Liska <mliska@suse.cz>
37476 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37478 2017-02-13 Richard Biener <rguenther@suse.de>
37480 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37482 * configure: Re-generate.
37483 * config.in: Likewise.
37484 * graphite-dependences.c: Simplify as if
37485 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37486 * graphite-isl-ast-to-gimple.c: Likewise.
37487 * graphite-optimize-isl.c: Likewise.
37488 * graphite-poly.c: Likewise.
37489 * graphite-sese-to-poly.c: Likewise.
37490 * graphite.h: Likewise.
37491 * toplev.c: Include isl/version.h and use isl_version () for
37492 printing the ISL version.
37493 * doc/install.texi: Update ISL requirement.
37495 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37497 * doc/standards.texi (Standards): Update reference to
37500 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37502 * doc/extend.texi (Named Address Spaces): sourceware.org now
37504 * doc/install.texi (Binaries): Ditto.
37507 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37509 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37510 standard terminology "stringize"/"stringizing" throughout.
37511 * doc/cppinternals.texi: Likewise.
37513 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37515 * doc/extend.texi: Fix some spelling mistakes and typos.
37516 * doc/invoke.texi: Likewise.
37518 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37521 * params.def (inline-min-speedup) Change from 10 to 8.
37523 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37525 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37528 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37531 * ipa-inline-analysis.c (get_minimal_bb): New function.
37532 (record_modified): Use it.
37533 (remap_edge_change_prob): Handle also ancestor functions.
37535 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37537 * doc/contrib.texi (Contributors): Remove broken link into
37538 the Mauve CVS repository.
37540 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37542 PR middle-end/79454
37543 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37544 result computation whenever lhs doesn't have vector mode, not
37545 just when it has BLKmode.
37547 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37549 * doc/makefile.texi (profiledbootstrap): Refer to the
37550 installation instructions only in textual form.
37552 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37555 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37557 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37559 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37560 (Specific): Update mingw-w64 reference.
37562 (Specific): Remove broken link to Renesas RX processor.
37564 2017-02-10 Richard Biener <rguenther@suse.de>
37566 * toplev.c (process_options): Do not mention obsolete graphite
37567 options when printing sorry message about missing graphite support.
37568 Mention -floop-nest-optimize.
37570 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37572 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37573 (vtst_p16): Likewise.
37574 (vtstq_p8): Likewise.
37575 (vtstq_p16): Likewise.
37577 (vtstq_p64): Likewise.
37578 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37579 (vset_lane_p64): New.
37580 (vsetq_lane_p64): New.
37582 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37584 PR tree-optimization/79411
37585 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37586 stmt operands are SSA_NAMEs used in abnormal phis.
37587 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37590 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37593 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37596 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37598 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37600 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37602 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37606 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37607 not arbitrary TREE_CONSTANT.
37610 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37611 "omp declare target link" attribute unless is_global_var.
37612 * omp-offload.c (find_link_var_op): Likewise.
37614 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37615 Chung-Lin Tang <cltang@codesourcery.com>
37617 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37619 (gimplify_adjust_omp_clauses): Don't delete TILE.
37620 (gimplify_omp_for): Deal with TILE.
37621 * internal-fn.c (expand_GOACC_TILE): New function.
37622 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37624 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37625 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37627 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37628 avoid DIV for outermost collapse var.
37629 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37630 Remove out of date comments, fix whitespace.
37631 * omp-general.c (omp_extract_for_data): Deal with tiling.
37632 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37633 adjust OLF_DIM_BASE value.
37634 (struct omp_for_data): Add tiling field.
37635 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37636 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37637 for auto loops. Remove default auto determining, moved to
37638 oacc_loop_fixed_partitions.
37639 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37640 stmts, add e_mask field.
37641 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37642 (oacc_thread_numbers): Use oacc_dim_call.
37643 (oacc_xform_tile): New.
37644 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37645 (finish_oacc_loop): Adjust for ifns vector.
37646 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37647 add case for GOACC_TILE fns.
37648 (oacc_loop_xform_loop): Delete.
37649 (oacc_loop_process): Iterate over call list directly, and add
37650 handling for GOACC_TILE fns.
37651 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37653 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37654 vector partitioning to outer loops. Assign 2 partitions to loops
37655 when available. Add TILE handling.
37656 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37657 (execite_oacc_device_lower): Process GOACC_TILE fns,
37658 ignore unknown specs.
37659 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37660 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37661 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37663 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37665 * configure.ac (ACX_BUGURL): Update.
37666 * configure: Regenerate.
37668 2017-02-09 Richard Biener <rguenther@suse.de>
37670 PR tree-optimization/69823
37671 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37672 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37674 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37676 * config/arc/arc-c.def: Add __NPS400__ definition.
37677 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37678 (TARGET_NPS400): Define.
37680 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37682 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37684 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37685 pointer, arch_info.
37686 (arc_cpu_types): Fill the arch_info field with a pointer into the
37687 arc_arch_types table.
37688 (arc_selected_cpu): Declare.
37689 * config/arc/arc.c (arc_selected_cpu): Make global.
37690 (arc_selected_arch): Delete.
37691 (arc_base_cpu): Delete.
37692 (arc_override_options): Remove references to deleted variables,
37693 update access to arch information.
37694 (ARC_OPT): Update access to arch information.
37695 (ARC_OPTX): Likewise.
37696 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37697 (TARGET_ARC600): Update access to arch information.
37698 (TARGET_ARC601): Likewise.
37699 (TARGET_ARC700): Likewise.
37700 (TARGET_EM): Likewise.
37701 (TARGET_HS): Likewise.
37702 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37705 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37708 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37709 condition/operands for integer GE/LE/GEU/LEU operations.
37711 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37713 PR translation/79397
37714 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37717 2017-02-08 Martin Jambor <mjambor@suse.cz>
37720 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37721 whether allocation happened.
37722 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37723 nothing was allocated.
37725 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37727 PR tree-optimization/79408
37728 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37729 constant, but SSA_NAME with a known integer range, use the minimum
37730 of that range instead of op1 to determine if modulo can be replaced
37731 with its first operand.
37733 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37735 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37737 2017-02-08 Richard Biener <rguenther@suse.de>
37739 PR tree-optimization/71824
37740 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37741 Check all loops contained in the merged region.
37743 2017-02-07 Andrew Pinski <apinski@cavium.com>
37745 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37747 2017-02-07 Andrew Pinski <apinski@cavium.com>
37749 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37750 (thunderxt88): Likewise.
37751 (thunderxt81): Disable LSE and change v8.1 to v8.
37752 (thunderxt83): Likewise.
37754 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37755 Richard Biener <rguenther@suse.de>
37757 PR middle-end/79399
37758 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37759 type from int to size_t.
37760 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37762 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37764 PR rtl-optimization/79386
37765 * cprop.c (bypass_conditional_jumps): Initialize
37766 bypass_last_basic_block already before splitting bbs after
37767 unconditional traps...
37768 (bypass_conditional_jumps): ... rather than here.
37771 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37772 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37773 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37774 fix -masm=intel patterns.
37776 2017-02-07 Richard Biener <rguenther@suse.de>
37778 PR tree-optimization/79256
37779 PR middle-end/79278
37780 * builtins.c (get_object_alignment_2): Use min_align_of_type
37781 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37782 and ADJUST_FIELD_ALIGN.
37784 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37786 * doc/tm.texi: Regenerate.
37787 * stor-layout.c (layout_decl): Adjust.
37788 (update_alignment_for_field): Likewise.
37789 (place_field): Likewise.
37790 (min_align_of_type): Likewise.
37791 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37792 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37793 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37794 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37795 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37796 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37797 * config/i386/i386.c (x86_field_alignment): Likewise.
37798 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37799 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37800 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37801 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37802 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37803 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37807 2017-01-30 Richard Biener <rguenther@suse.de>
37809 PR tree-optimization/79256
37810 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37811 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37814 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37816 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37817 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37818 builtins to SImode and emit a zero-extend, if necessary.
37820 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37822 * docs/invoke.texi (RISC-V Options): Alphabetize.
37824 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37826 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37829 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37831 * config/riscv/riscv.c: New file.
37832 * common/config/riscv/riscv-common.c: Likewise.
37833 * config.gcc: Likewise.
37834 * config/riscv/constraints.md: Likewise.
37835 * config/riscv/elf.h: Likewise.
37836 * config/riscv/generic.md: Likewise.
37837 * config/riscv/linux.h: Likewise.
37838 * config/riscv/multilib-generator: Likewise.
37839 * config/riscv/peephole.md: Likewise.
37840 * config/riscv/pic.md: Likewise.
37841 * config/riscv/predicates.md: Likewise.
37842 * config/riscv/riscv-builtins.c: Likewise.
37843 * config/riscv/riscv-c.c: Likewise.
37844 * config/riscv/riscv-ftypes.def: Likewise.
37845 * config/riscv/riscv-modes.def: Likewise.
37846 * config/riscv/riscv-opts.h: Likewise.
37847 * config/riscv/riscv-protos.h: Likewise.
37848 * config/riscv/riscv.h: Likewise.
37849 * config/riscv/riscv.md: Likewise.
37850 * config/riscv/riscv.opt: Likewise.
37851 * config/riscv/sync.md: Likewise.
37852 * config/riscv/t-elf-multilib: Likewise.
37853 * config/riscv/t-linux: Likewise.
37854 * config/riscv/t-linux-multilib: Likewise.
37855 * config/riscv/t-riscv: Likewise.
37856 * configure.ac: Likewise.
37857 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37858 Waterman as RISC-V maintainers.
37859 * doc/install.texi: Add RISC-V entries.
37860 * doc/invoke.texi: Add RISC-V options section.
37861 * doc/md.texi: Add RISC-V constraints section.
37862 * configure: Regenerated.
37864 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37867 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37868 false values to be constant vectors with all 0 or all 1 bits set.
37869 (vcondu<mode><mode>): Likewise.
37870 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37872 (fpmask_comparison_operator): Update comment.
37873 (vecint_comparison_operator): New predicate.
37874 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37875 vector conditionals when the true and false values are constant
37876 vectors with all 0 bits or all 1 bits set.
37878 2017-02-06 Martin Sebor <msebor@redhat.com>
37880 PR tree-optimization/79376
37881 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37883 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37885 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37886 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37887 to simplify split condition.
37889 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37891 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37892 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37895 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37897 PR rtl-optimization/68664
37898 * target.def (can_speculate_insn): New hook.
37899 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37900 * doc/tm.texi: Regenerate.
37901 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37902 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37903 (rs6000_sched_can_speculate_insn): New function.
37905 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37907 PR tree-optimization/79284
37908 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37909 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37910 vectorizable_mask_load_store, vectorizable_operation,
37911 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37912 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37913 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37914 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37915 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37916 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37917 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37918 is_gimple_assign (stmt). Replace another such test with
37919 is_gimple_assign (stmt).
37921 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37924 * config/avr/avr.c (rtl-iter.h): Include it.
37925 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37926 (avr_legitimate_combined_insn): ...and implementation.
37928 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37930 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37931 * config/s390/s390.c (s390_const_operand_ok)
37932 (s390_canonicalize_comparison, s390_extract_part)
37933 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37934 (s390_contiguous_bitmask_p, s390_rtx_costs)
37935 (legitimize_pic_address): Likewise.
37936 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37937 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37938 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37939 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37940 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37942 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37944 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37945 REGNO($0) == REGNO($1).
37947 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37949 * config/s390/linux.h(SIZE_TYPE): Add comment.
37951 2017-02-06 Julian Brown <julian@codesourcery.com>
37952 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37953 Virendra Pathak <virendra.pathak@broadcom.com>
37955 * config/aarch64/aarch64-cores.def: Change the scheduler
37957 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37958 * config/aarch64/thunderx2t99.md: New file.
37960 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37962 * doc/standards.texi (Go Language): Update link to language
37965 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37967 * tree-eh.c (lower_resx): Sanitize profile.
37968 (cleanup_empty_eh_move_lp): Likewise.
37970 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37973 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37975 * cfgloopmanip.h (loop_version): Update prototype.
37976 * modulo-sched.c (sms_schedule): Update call of loop_version.
37977 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37978 * tree-parloops.c (gen_parallel_loop): Likewise.
37979 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37980 * tree-ssa-loop-split.c (split_loop): Likewise.
37981 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37982 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37984 2017-02-05 Martin Liska <mliska@suse.cz>
37987 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37989 (print_operand_address): Initialize a struct to zero.
37991 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37993 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37994 garbage collector only in textual form.
37996 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37998 * doc/extend.texi (x86 specific memory model extensions for
37999 transactional memory): Simplify a phrase.
38001 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38004 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38005 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38006 (atomic_storedi_1): Likewise.
38008 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38010 PR tree-optimization/79338
38011 * tree-parloops.c (gather_scalar_reductions): Don't call
38012 vect_analyze_loop_form for loop->inner before destroying loop's
38015 2017-02-03 Martin Sebor <msebor@redhat.com>
38017 PR tree-optimization/79327
38018 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38019 when precision has resulted in leading zeros.
38020 (format_integer): Adjust the likely counter to assume an unknown
38021 argument that may be zero is non-zero.
38023 2017-02-03 Jason Merrill <jason@redhat.com>
38026 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38027 avoid copying non-taken branch.
38029 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38031 PR tree-optimization/79340
38032 * tree-vect-loop.c (vectorizable_reduction): Release
38033 vec_defs elements after safe_splicing them into other vectors.
38036 PR tree-optimization/79327
38037 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38038 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38040 (format_integer): Use wide_int_to_tree instead of build_int_cst
38041 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38042 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38043 of shortest and longest sequence.
38045 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38047 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38048 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38050 2017-02-03 Walter Lee <walt@tilera.com>
38053 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38054 after initial stackframe link reg save.
38055 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38057 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38060 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38061 wu for stxssp alternative.
38063 2017-02-03 Martin Sebor <msebor@redhat.com>
38065 PR tree-optimization/79352
38066 * gimple-fold.c (get_range_strlen): Add argument.
38067 (get_range_strlen): Change return type to bool.
38068 (get_maxval_strlen): Pass in a dummy argument.
38069 * gimple-fold.h (get_range_strlen): Change return type to bool.
38070 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38071 * tree.h (array_at_struct_end_p): Add argument.
38072 * tree.c (array_at_struct_end_p): Handle it.
38074 2017-02-03 Martin Liska <mliska@suse.cz>
38077 * multiple_target.c (create_dispatcher_calls): Redirect edge
38078 from a caller of a dispatcher.
38079 (expand_target_clones): Make the clones local.
38080 (ipa_target_clone): Do both target clones and resolvers.
38081 (ipa_dispatcher_calls): Remove the pass.
38082 (pass_dispatcher_calls::gate): Likewise.
38083 (make_pass_dispatcher_calls): Likewise.
38084 * passes.def (pass_target_clone): Put as very first IPA early
38087 2017-02-03 Martin Liska <mliska@suse.cz>
38089 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38090 in case of a function with ifunc attribute.
38092 2017-02-03 Martin Liska <mliska@suse.cz>
38094 * cgraph.c (cgraph_node::dump): Dump function version info.
38095 * symtab.c (symtab_node::dump_base): Add missing new line.
38097 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38099 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38100 (ifcombine_ifandif): Use it.
38102 2017-02-03 Martin Liska <mliska@suse.cz>
38104 * doc/invoke.texi: Document default value for
38105 use-after-scope-direct-emission-threshold.
38107 2017-02-03 Martin Liska <mliska@suse.cz>
38109 PR tree-optimization/79339
38110 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38111 (format_floating): Likewise.
38113 2017-02-03 Martin Liska <mliska@suse.cz>
38116 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38118 (ipa_node_params_t::remove): Likewise.
38119 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38120 initialization from removed ipa_node_params_t::insert.
38121 (ipa_node_params::~ipa_node_params): Move from removed
38122 ipa_node_params_t::release.
38123 * symbol-summary.h (symbol_summary::m_released): New member.
38124 Do not release a summary twice. Do not allow to call finalizer
38125 for types of a summary that live in GGC memory.
38127 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38129 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38132 2017-02-02 Martin Sebor <msebor@redhat.com>
38134 PR middle-end/79275
38135 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38136 (format_string): Tighten up the range of output for non-constant
38137 strings and correct the expected range for wide non-constant strings.
38139 2017-02-02 Martin Sebor <msebor@redhat.com>
38141 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38143 PR middle-end/32003
38144 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38146 (-fdump-tree-@var): Add to index and document how to come up
38147 with pass-specific option and dump file names.
38148 (-fdump-passes): Clarify where to look for output.
38150 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38152 PR middle-end/77445
38153 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38154 statistics of the analyzed path; allow threading for speed when
38155 any of BBs along the path are optimized for speed.
38157 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38159 PR middle-end/78468
38160 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38161 settings of the virtual registers.
38164 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38166 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38167 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38170 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38172 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38173 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38175 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38177 * config/s390/s390.md: Add missing comments with the expanded
38179 * config/s390/vector.md: Likewise.
38180 * config/s390/vx-builtins.md: Likewise.
38182 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38185 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38186 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38187 conditions on a single line.
38189 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38191 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38192 __S390_VX__ to __VX__.
38194 2017-02-01 Andrew Pinski <apinski@cavium.com>
38196 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38197 stmt_info to record_stmt_cost.
38198 (vect_get_known_peeling_cost): Pass stmt_info if known to
38200 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38201 cpu_vector_cost field into
38202 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38203 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38204 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38205 splitting of scalar_stmt_cost and vec_stmt_cost.
38206 (thunderx_vector_cost): Likewise.
38207 (cortexa57_vector_cost): LIkewise.
38208 (exynosm1_vector_cost): Likewise.
38209 (xgene1_vector_cost): Likewise.
38210 (thunderx2t99_vector_cost): Improve after the splitting of the two
38212 (aarch64_builtin_vectorization_cost): Update for the splitting of
38213 scalar_stmt_cost and vec_stmt_cost.
38215 2017-02-01 Torvald Riegel <triegel@redhat.com>
38216 Richard Henderson <rth@redhat.com>
38218 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38219 conditional on existance of a fast atomic load.
38220 * optabs-query.c (can_atomic_load_p): New function.
38221 * optabs-query.h (can_atomic_load_p): Declare it.
38222 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38223 no fast atomic load is available for the particular size of access.
38224 (expand_atomic_compare_and_swap): Likewise.
38225 (expand_atomic_load): Likewise.
38226 (expand_atomic_store): Likewise.
38227 (expand_atomic_fetch_op): Likewise.
38228 * testsuite/lib/target-supports.exp
38229 (check_effective_target_sync_int_128): Remove x86 because it provides
38230 no fast atomic load.
38231 (check_effective_target_sync_int_128_runtime): Likewise.
38233 2017-02-01 Richard Biener <rguenther@suse.de>
38235 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38236 (graphite_transform_loops): Provide opt-info for optimized nests.
38237 * tree-parloop.c (parallelize_loops): Provide opt-info for
38238 parallelized loops.
38240 2017-02-01 Richard Biener <rguenther@suse.de>
38242 PR middle-end/79315
38243 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38244 was not set before.
38246 2017-02-01 Richard Biener <rguenther@suse.de>
38248 PR tree-optimization/71824
38249 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38250 Verify the loops are valid in the merged SESE region.
38251 (scop_detection::can_represent_loop_1): Check analyzing the
38252 evolution of the number of iterations in the region succeeds.
38254 2017-01-31 Ian Lance Taylor <iant@golang.org>
38256 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38257 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38259 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38261 PR preprocessor/79210
38262 * input.c (get_substring_ranges_for_loc): Replace line_width
38263 assertion with error-handling.
38265 2017-01-31 Richard Biener <rguenther@suse.de>
38267 PR tree-optimization/77318
38268 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38269 (create_pw_aff_from_tree): Take loop parameter.
38270 (add_condition_to_pbb): Pass loop of the condition to
38271 create_pw_aff_from_tree.
38273 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38275 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38276 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38278 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38282 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38284 (convert_int_to_float128): Likewise.
38285 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38286 (convert_int_to_float128): Likewise.
38287 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38288 (UNSPEC_IEEE128_CONVERT): Likewise.
38289 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38290 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38291 Use local variables for IBM extended format.
38292 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38293 (fix_trunc<mode>si2_fprs): Likewise.
38294 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38295 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38296 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38297 to know that we can now have integers of all sizes in vector
38299 (fix<uns>_<mode>di2_hw): Likewise.
38300 (float<uns>_<mode>si2_hw): Likewise.
38301 (fix_<mode>si2_hw): Likewise.
38302 (fixuns_<mode>si2_hw): Likewise.
38303 (float<uns>_<mode>di2_hw): Likewise.
38304 (float_<mode>di2_hw): Likewise.
38305 (float_<mode>si2_hw): Likewise.
38306 (floatuns_<mode>di2_hw): Likewise.
38307 (floatuns_<mode>si2_hw): Likewise.
38308 (xscvqp<su>wz_<mode>): Delete, no longer used.
38309 (xscvqp<su>dz_<mode>): Likewise.
38310 (xscv<su>dqp_<mode>): Likewise.
38311 (ieee128_mfvsrd_64bit): Likewise.
38312 (ieee128_mfvsrd_32bit): Likewise.
38313 (ieee128_mfvsrwz): Likewise.
38314 (ieee128_mtvsrw): Likewise.
38315 (ieee128_mtvsrd_64bit): Likewise.
38316 (ieee128_mtvsrd_32bit): Likewise.
38318 2017-01-31 Martin Liska <mliska@suse.cz>
38321 * ipa-prop.c (ipa_free_all_node_params): Call release method
38322 instead of ~sumbol_summary to not to trigger double times
38325 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38327 PR tree-optimization/71691
38328 * bitmap.h (class auto_bitmap): New.
38329 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38330 is_maybe_undefined instead of ssa_undefined_value_p.
38332 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38334 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38335 __S390_ARCH_LEVEL__ to __ARCH__.
38337 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38339 PR tree-optimization/79267
38340 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38341 if should_remove_lhs_p is true.
38343 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38346 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38347 (add_alignment_attribute): New.
38348 (base_type_die): Add alignment attribute.
38349 (subrange_type_die): Likewise.
38350 (modified_type_die): Likewise.
38351 (gen_array_type_die): Likewise.
38352 (gen_descr_array_type_die: Likewise.
38353 (gen_enumeration_type_die): Likewise.
38354 (gen_subprogram_die): Likewise.
38355 (gen_variable_die): Likewise.
38356 (gen_field_die): Likewise.
38357 (gen_ptr_to_mbr_type_die): Likewise.
38358 (gen_struct_or_union_type_die): Likewise.
38359 (gen_subroutine_type_die): Likewise.
38360 (gen_typedef_die): Likewise.
38361 (base_type_cmp): Compare alignment attribute.
38363 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38366 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38367 (setb_unsigned) New pattern for setb with CCUNS.
38368 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38369 subfc./subfe sequence to avoid overflow problems. Generate a
38370 shorter sequence with cmpld/setb for power9.
38371 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38372 for generating subfc. instruction.
38373 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38374 now uses this instruction.
38376 2017-01-30 Ian Lance Taylor <iant@google.com>
38379 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38380 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38382 2017-01-30 Martin Sebor <msebor@redhat.com>
38384 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38385 Move constant to the right of a relational operator.
38386 (get_mpfr_format_length, format_character, format_string): Ditto.
38387 (should_warn_p, maybe_warn): Same.
38389 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38391 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38394 * asan.c (get_translation_unit_decl): Remove function.
38395 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38397 2017-01-30 Martin Liska <mliska@suse.cz>
38399 PR gcov-profile/79259
38400 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38401 -fprofile-generate.
38403 2017-01-30 Martin Liska <mliska@suse.cz>
38406 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38407 Initialize variables with NULL value.
38409 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38412 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38414 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38416 2017-01-30 Richard Biener <rguenther@suse.de>
38418 PR tree-optimization/79276
38419 * tree-vrp.c (process_assert_insertions): Properly adjust common
38420 when removing a duplicate.
38422 2017-01-30 Richard Biener <rguenther@suse.de>
38424 PR tree-optimization/79256
38425 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38426 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38428 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38430 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38433 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38434 ("*r<noxa>sbg_<mode>_sll_bitmask")
38435 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38436 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38437 Use contiguous_bitmask_nowrap_operand.
38439 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38442 * config/rs6000/altivec.h (vec_xl): Revise #define.
38443 (vec_xst): Likewise.
38445 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38447 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38449 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38451 PR rtl-optimization/79194
38452 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38453 traps before call to bypass_conditional_jumps.
38455 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38457 PR tree-optimization/71374
38458 * lra-constraints.c (check_conflict_input_operands): New.
38459 (match_reload): Use it.
38461 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38464 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38465 account to calculate conflict_set.
38467 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38469 PR rtl-optimization/78559
38470 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38471 other_insn in combine.
38473 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38475 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38476 uint16_type_node for BT_UINT16.
38478 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38480 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38481 "RTL Tests" to menu.
38482 (GIMPLE Tests): New node.
38483 (RTL Tests): New node.
38485 2017-01-27 Richard Biener <rguenther@suse.de>
38487 PR tree-optimization/79245
38488 * tree-loop-distribution.c (distribute_loop): Apply cost
38489 modeling also to detected patterns.
38491 2017-01-27 Richard Biener <rguenther@suse.de>
38493 PR tree-optimization/71433
38494 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38495 (compare_assert_loc): New function.
38496 (process_assert_insertions): Sort and optimize assert locations
38497 to remove duplicates and push down identical assertions on
38498 edges to their destination block.
38500 2017-01-27 Richard Biener <rguenther@suse.de>
38502 PR tree-optimization/79244
38503 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38504 out SSA names even if abnormal.
38506 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38508 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38509 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38510 instead of MPFR_RNDN.
38512 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38515 * arm.c (arm_option_override): Don't call build_target_option_node
38516 until after doing all option overrides.
38517 (arm_valid_target_attribute_tree): Likewise.
38519 2017-01-27 Martin Liska <mliska@suse.cz>
38521 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38522 for {cd}tors and C++ {cd}tors.
38524 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38526 * config/s390/s390.md ("*setmem_long_and")
38527 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38529 2017-01-26 Martin Sebor <msebor@redhat.com>
38531 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38534 2017-01-26 Martin Sebor <msebor@redhat.com>
38536 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38537 HAVE_DFmode before using XFmode or DFmode.
38538 (parse_directive): Avoid using the z length modifier to avoid
38539 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38541 PR middle-end/78703
38542 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38543 to accept adjustment as an array.
38544 (get_int_range): New function.
38545 (struct directive): Make width and prec arrays.
38546 (directive::set_width, directive::set_precision): Call get_int_range.
38547 (format_integer, format_floating): Handle width and precision ranges.
38548 (format_string, parse_directive): Same.
38550 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38553 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38554 comdat_type_p set, just clone them, but keep the children in the
38558 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38559 which have direct callers with -fvar-tracking-assignments enabled
38561 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38562 inside of type units.
38564 2017-01-26 Martin Sebor <msebor@redhat.com>
38566 PR middle-end/78703
38567 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38569 (struct format_result): Replace number_chars, number_chars_min,
38570 and number_chars_max with a single member of struct result_range.
38572 (format_result::operator+=): Adjust.
38573 (struct fmtresult): Remove bounded. Handle likely and unlikely
38575 (fmtresult::adjust_for_width_or_precision): New function.
38576 (fmtresult:type_max_digits): New function.
38577 (bytes_remaining): Handle likely and unlikely counters.
38578 (min_bytes_remaining): Remove.
38579 (format_percent): Simplify.
38580 (format_integer, format_floating): Set likely and unlikely counters.
38581 (get_string_length, format_character, format_string): Same.
38582 (format_plain, should_warn_p): New function.
38583 (maybe_warn): Call should_warn_p. Update diagnostic messages
38584 and handle those for all directives, including plain strings.
38585 (format_directive): Handle likely and unlikely counters.
38586 Remove unnecessary quoting from diagnostics. Add an informational
38588 (add_bytes): Remove.
38589 (pass_sprintf_length::compute_format_length): Simplify.
38590 (try_substitute_return_value): Handle likely and unlikely counters.
38592 2017-01-26 Carl Love <cel@us.ibm.com>
38594 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38595 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38597 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38600 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38601 endianess for subregs into account.
38602 * lra-constraints.c (lra_constraints): Do risky transformations
38603 always on the first iteration.
38604 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38605 last_call_used_reg_set.
38606 (process_bb_lives): Define and use last_call_used_reg_set.
38607 * lra.c (lra): Always continue after lra_constraints on the first
38610 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38612 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38614 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38616 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38618 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38619 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38620 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38621 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38622 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38623 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38624 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38625 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38626 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38628 2017-01-26 Marek Polacek <polacek@redhat.com>
38631 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38632 for the third operand.
38634 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38636 PR middle-end/79236
38637 * omp-low.c (struct omp_context): Add simt_stmt field.
38638 (scan_omp_for): Return omp_context *.
38639 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38640 context to the _simt_ SIMD stmt.
38641 (lower_omp_for): For combined SIMD with sibling _simt_
38642 SIMD, make sure to use the same decls in _looptemp_
38643 clauses as in the sibling.
38645 2017-01-26 David Sherwood <david.sherwood@arm.com>
38647 PR middle-end/79212
38648 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38651 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38654 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38655 emit fld b; fld a; if possible.
38657 * brig-builtins.def: Update copyright years.
38658 * config/arm/arm_acle_builtins.def: Update copyright years.
38660 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38663 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38664 constraint instead of o for the stxsd instruction.
38666 2017-01-25 Carl Love <cel@us.ibm.com>
38668 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38669 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38671 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38673 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38675 2017-01-25 Richard Biener <rguenther@suse.de>
38677 PR tree-optimization/69264
38678 * target.def (vector_alignment_reachable): Improve documentation.
38679 * doc/tm.texi: Regenerate.
38680 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38682 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38683 earlier changes with respect to TYPE_USER_ALIGN.
38684 (vector_alignment_reachable_p): Likewise. Improve dumping.
38686 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38689 * config/arm/arm.md (xordi3): Force constant operand into a register
38692 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38694 * doc/invoke.texi (-fstore-merging): Correct default optimization
38695 levels at which it is enabled.
38696 (-O): Move -fstore-merging from list to...
38699 2017-01-25 Richard Biener <rguenther@suse.de>
38702 * omp-expand.c: Include debug.h.
38703 (expand_omp_taskreg): Make sure to generate early debug before
38704 outlining anything from a function.
38705 (expand_omp_target): Likewise.
38706 (grid_expand_target_grid_body): Likewise.
38708 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38711 * asan.c (get_translation_unit_decl): New function.
38712 (asan_add_global): Extract modules file name from globals
38713 TRANSLATION_UNIT_DECL name.
38715 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38718 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38719 for long calls with APCS frame and VFP.
38721 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38723 * cfg.c (original_copy_tables_initialized_p): New function.
38724 * cfg.h (original_copy_tables_initialized_p): New decl.
38725 * cfgrtl.c (relink_block_chain): Guard the call to
38726 free_original_copy_tables with a call to
38727 original_copy_tables_initialized_p.
38728 * cgraph.h (symtab_node::native_rtl_p): New decl.
38729 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38730 (symtab_node::needed_p): Don't assert for early assembly output
38731 for __RTL functions.
38732 (cgraph_node::finalize_function): Set "force_output" for __RTL
38734 (cgraph_node::analyze): Bail out early for __RTL functions.
38735 (analyze_functions): Update assertion to support __RTL functions.
38736 (cgraph_node::expand): Bail out early for __RTL functions.
38737 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38739 * function.h (struct function): Update comment for field
38741 * gimple-expr.c: Include "tree-pass.h".
38742 (gimple_has_body_p): Return false for __RTL functions.
38743 * Makefile.in (OBJS): Add run-rtl-passes.o.
38744 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38746 (gcc::pass_manager::get_clean_slate): New accessor.
38747 * passes.c: Include "insn-addr.h".
38748 (should_skip_pass_p): Add logging. Update logic for running
38749 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38750 property-provider override so it is only done for gimple passes.
38752 (skip_pass): New function.
38753 (execute_one_pass): Call skip_pass when skipping passes.
38754 * read-md.c (md_reader::read_char): Support filtering
38755 the input to a subset of line numbers.
38756 (md_reader::md_reader): Initialize fields
38757 m_first_line and m_last_line.
38758 (md_reader::read_file_fragment): New function.
38759 * read-md.h (md_reader::read_file_fragment): New decl.
38760 (md_reader::m_first_line): New field.
38761 (md_reader::m_last_line): New field.
38762 * read-rtl-function.c (function_reader::create_function): Only
38763 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38764 curr_properties. Set DECL_INITIAL to a dummy block.
38765 (read_rtl_function_body_from_file_range): New function.
38766 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38768 * run-rtl-passes.c: New file.
38769 * run-rtl-passes.h: New file.
38771 2017-01-24 Jeff Law <law@redhat.com>
38773 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38776 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38778 PR tree-optimization/79159
38779 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38780 (record_nonwrapping_iv): Improve boundary using above function if no
38781 value range information.
38783 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38784 Martin Jambor <mjambor@suse.cz>
38786 * brig-builtins.def: New file.
38787 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38788 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38789 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38790 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38791 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38792 * builtin-types.def (BT_INT8): New.
38793 (BT_INT16): Likewise.
38794 (BT_UINT8): Likewise.
38795 (BT_UINT16): Likewise.
38796 (BT_FN_ULONG): Likewise.
38797 (BT_FN_UINT_INT): Likewise.
38798 (BT_FN_UINT_ULONG): Likewise.
38799 (BT_FN_UINT_LONG): Likewise.
38800 (BT_FN_UINT_PTR): Likewise.
38801 (BT_FN_ULONG_PTR): Likewise.
38802 (BT_FN_INT8_FLOAT): Likewise.
38803 (BT_FN_INT16_FLOAT): Likewise.
38804 (BT_FN_UINT32_FLOAT): Likewise.
38805 (BT_FN_UINT16_FLOAT): Likewise.
38806 (BT_FN_UINT8_FLOAT): Likewise.
38807 (BT_FN_UINT64_FLOAT): Likewise.
38808 (BT_FN_UINT16_UINT32): Likewise.
38809 (BT_FN_UINT32_UINT16): Likewise.
38810 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38811 (BT_FN_INT_PTR_INT): Likewise.
38812 (BT_FN_UINT_PTR_UINT): Likewise.
38813 (BT_FN_LONG_PTR_LONG): Likewise.
38814 (BT_FN_ULONG_PTR_ULONG): Likewise.
38815 (BT_FN_VOID_UINT64_UINT64): Likewise.
38816 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38817 (BT_FN_INT8_INT8_INT8): Likewise.
38818 (BT_FN_INT16_INT16_INT16): Likewise.
38819 (BT_FN_INT_INT_INT): Likewise.
38820 (BT_FN_UINT_FLOAT_UINT): Likewise.
38821 (BT_FN_FLOAT_UINT_UINT): Likewise.
38822 (BT_FN_ULONG_UINT_UINT): Likewise.
38823 (BT_FN_ULONG_UINT_PTR): Likewise.
38824 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38825 (BT_FN_UINT_UINT_UINT): Likewise.
38826 (BT_FN_VOID_UINT_PTR): Likewise.
38827 (BT_FN_UINT_UINT_PTR: Likewise.
38828 (BT_FN_UINT32_UINT64_PTR): Likewise.
38829 (BT_FN_INT_INT_UINT_UINT): Likewise.
38830 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38831 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38832 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38833 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38834 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38835 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38836 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38837 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38838 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38839 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38840 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38841 * doc/frontends.texi: List BRIG FE.
38842 * doc/install.texi (Testing): Add BRIG tesring requirements.
38843 * doc/invoke.texi (Overall Options): Mention BRIG.
38844 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38846 2017-01-24 Richard Biener <rguenther@suse.de>
38848 PR translation/79208
38849 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38851 2017-01-24 Martin Jambor <mjambor@suse.cz>
38854 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38855 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38856 and known_contexts.
38858 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38860 PR middle-end/79123
38861 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38862 casts from signed to unsigned really don't have a range.
38864 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38866 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38867 GMP_RNDx for compatiblity.
38869 2017-01-24 Martin Liska <mliska@suse.cz>
38872 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38873 that would prevent us to call alloca with -1 as argument.
38875 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38877 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38878 Avoid -Wformat-security warning.
38880 2017-01-23 Andrew Pinski <apinski@cavium.com>
38882 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38885 2017-01-23 Martin Sebor <msebor@redhat.com>
38887 PR middle-end/78703
38888 * gimple-ssa-sprintf.c (warn_level): New global.
38889 (format_integer): Use it here and throughout the rest of the file.
38890 Use the same switch to compute sign as base.
38891 (maybe_warn): New function.
38892 (format_directive): Factor out warnings into maybe_warn.
38893 Add debugging output. Use warn_level.
38894 (add_bytes): Use warn_level.
38895 (pass_sprintf_length::compute_format_length): Add debugging output.
38896 (try_substitute_return_value): Same.
38897 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38899 PR middle-end/78703
38900 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38901 (struct fmtresult, format_integer, format_floating): Adjust.
38902 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38903 (get_string_length, format_string,format_directive): Same.
38904 (pass_sprintf_length::compute_format_length): Same.
38905 (try_substitute_return_value): Simplify slightly.
38907 PR middle-end/78703
38908 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38909 (fmtresult::operator+=): Outlined.
38910 (struct fmtresult): Add ctors.
38911 (struct conversion_spec): Rename...
38912 (struct directive): ...to this. Add and remove data members.
38913 (directive::set_width, directive::set_precision): New functions.
38914 (format_percent): Use fmtresult ctor.
38915 (get_width_and_precision): Remove.
38916 (format_integer): Make naming changes. Avoid computing width and
38918 (format_floating): Same. Adjust indentation.
38919 (format_character, format_none): New functions.
38920 (format_string): Moved character handling to format_character.
38921 (format_directive): Remove arguments, change return type.
38922 (parse_directive): New function.
38923 (pass_sprintf_length::compute_format_length): Move directive
38924 parsing to parse_directive.
38926 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38928 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38929 (assign_assembler_name_if_needed): ... this.
38930 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38931 (assign_assembler_name_if_needed): ... this.
38932 (free_lang_data_in_cgraph): Adjust callers.
38933 * cgraphunit.c (cgraph_node::analyze): Likewise.
38934 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38937 2017-01-23 Richard Biener <rguenther@suse.de>
38939 PR tree-optimization/79088
38940 PR tree-optimization/79188
38941 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38942 resetting loop bounds after last path deletion. Reset loop
38943 bounds of the target loop, make code match the comments.
38944 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38945 Make sure loops need no fixups.
38947 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38949 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38950 exponent support with double type for first argument.
38951 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38952 type returned by __builtin_vec_extract_sig,
38953 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38954 functions from "vector int" to "vector unsigned int" or from
38955 "vector long long int" to "vector unsigned long long int".
38956 Changed type returned by __builtin_vec_extract_exp,
38957 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38958 functions from "vector int" to "vector unsigned int" or from
38959 "vector long long int" to "vector unsigned long long int".
38960 Changed return type of __builtin_vec_test_data_class,
38961 __builtin_vec_test_data_class_sp, and
38962 __builtin_vec_test_data_class_dp from "vector int" to
38963 "vector bool int" or from "vector long long int" to "vector bool
38964 long long int" and changed second argument type from "unsigned
38965 int" to "int". Added new overloaded function forms "vector float
38966 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38967 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38968 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38969 double, vector unsigned long long int)" and "vector double
38970 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38971 long int)". Changed return type of
38972 __builtin_scalar_test_data_class and
38973 __builtin_scalar_test_data_class_sp and
38974 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38975 int" and changed second argument from "unsigned int" to "int".
38976 Changed type returned by __builtin_scalar_test_neg,
38977 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38978 from "int" to "bool int". Added new overloaded function form
38979 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38980 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38981 exponent double-precision with floating point first argument.
38982 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38983 documentation of scalar_test_data_class, scalar_test_neg,
38984 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38985 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38986 vec_test_data_class built-in functions to reflect refinements in
38987 their type signatures.
38989 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38991 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38993 (aarch64_elf_asm_destructor): Likewise.
38995 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38997 PR rtl-optimization/78634
38998 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38999 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39000 * ifcvt.c (noce_try_cmove): Add missing cost check.
39002 PR rtl-optimization/71724
39003 * combine.c (if_then_else_cond): Look for situations where it is
39004 beneficial to undo the work of one of the recursive calls.
39006 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39008 PR tree-optimization/70754
39009 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39010 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39011 combined stmt before it if not NULL.
39012 (combine_chains): Process refs reversely and compute dominance point
39015 2017-01-23 Martin Liska <mliska@suse.cz>
39017 PR tree-optimization/79196
39018 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39019 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39021 (strlen_optimize_stmt): Call the renamed function.
39023 2017-01-23 Michael Matz <matz@suse.de>
39025 PR tree-optimization/78384
39026 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39028 2017-01-23 Richard Biener <rguenther@suse.de>
39030 PR tree-optimization/79186
39031 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39032 both incoming edges before moving an assert.
39034 2017-01-23 Martin Jambor <mjambor@suse.cz>
39036 * ipa-prop.c (load_from_param_1): Removed.
39037 (load_from_unmodified_param): Bits from load_from_param_1 put back
39039 (load_from_param): Removed.
39040 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39041 with stmt. Reverted back to use of load_from_unmodified_param.
39043 2017-01-23 Martin Jambor <mjambor@suse.cz>
39046 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39047 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39048 field a pointer to garbage collected vector, mark lattices and
39049 ipcp_orig_node with GTY((skip)).
39050 (ipa_get_param_count): Adjust to descriptors being a pointer.
39051 (ipa_get_param): Likewise.
39052 (ipa_get_type): Likewise.
39053 (ipa_get_param_move_cost): Likewise.
39054 (ipa_set_param_used): Likewise.
39055 (ipa_get_controlled_uses): Likewise.
39056 (ipa_set_controlled_uses): Likewise.
39057 (ipa_is_param_used): Likewise.
39058 (ipa_node_params_t): Move into garbage collector. New methods insert
39060 (ipa_node_params_sum): Annotate wth GTY(()).
39061 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39063 (ipa_load_from_parm_agg): Adjust declaration.
39064 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39065 * ipa-profile.c (ipa_profile): Likewise.
39066 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39067 (ipa_populate_param_decls): Make descriptors parameter garbage
39069 (ipa_dump_param): Adjust to descriptors being a pointer.
39070 (ipa_alloc_node_params): Likewise.
39071 (ipa_initialize_node_params): Likewise.
39072 (load_from_param_1): Make descriptors parameter garbage collected.
39073 (load_from_unmodified_param): Likewise.
39074 (load_from_param): Likewise.
39075 (ipa_load_from_parm_agg): Likewise.
39076 (ipa_node_params::~ipa_node_params): Removed.
39077 (ipa_free_all_node_params): Remove call to delete operator.
39078 (ipa_node_params_t::insert): New.
39079 (ipa_node_params_t::remove): Likewise.
39080 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39081 copy known_csts and known_contexts vectors.
39082 (ipa_read_node_info): Adjust to descriptors being a pointer.
39083 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39085 (ipcp_transform_function): Make descriptors variable garbage
39088 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39090 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39091 * config/i386/avx512dqintrin.h: Ditto.
39092 * config/i386/avx512fintrin.h: Ditto.
39093 * config/i386/i386.c: Handle new builtins.
39094 * config/i386/i386-builtin.def: Add new builtins.
39095 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39096 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39098 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39099 Martin Liska <mliska@suse.cz>
39101 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39102 * asan.c (asan_expand_poison_ifn): Support stores and use
39103 appropriate ASAN report function.
39104 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39105 * internal-fn.def (ASAN_POISON_USE): Declare.
39106 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39107 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39108 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39109 ASAN_POISON calls w/o LHS.
39110 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39111 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39112 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39113 * gimplify.c (asan_poison_variables): Add attribute
39114 use_after_scope_memory to variables that really needs to live
39116 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39117 having the attribute.
39119 2017-01-23 Martin Liska <mliska@suse.cz>
39121 * asan.c (create_asan_shadow_var): New function.
39122 (asan_expand_poison_ifn): Likewise.
39123 * asan.h (asan_expand_poison_ifn): New declaration.
39124 * internal-fn.c (expand_ASAN_POISON): Likewise.
39125 * internal-fn.def (ASAN_POISON): New builtin.
39126 * sanopt.c (pass_sanopt::execute): Expand
39127 asan_expand_poison_ifn.
39128 * tree-inline.c (copy_decl_for_dup_finish): Make function
39130 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39131 * tree-ssa.c (is_asan_mark_p): New function.
39132 (execute_update_addresses_taken): Rewrite local variables
39133 (identified just by use-after-scope as addressable) into SSA.
39135 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39137 * doc/install.texi (Specific): opensource.apple.com uses https
39138 now. Remove trailing slash.
39140 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39142 * README.Portability: Remove note on an Irix compatibility issue.
39144 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39146 * gcov.c (INCLUDE_ALGORITHM): Define.
39147 (INCLUDE_VECTOR): Define.
39148 No longer include <vector> and <algorithm> directly.
39150 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39152 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39154 * doc/invoke.texi (Code Gen Options): Ditto.
39156 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39159 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39161 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39163 rtl-optimization/79125
39164 * cprop.c (local_cprop_pass): Handle cases where we make an
39165 unconditional trap.
39167 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39171 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39172 read from, for big endian.
39174 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39176 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39177 register pauth builtins for LP64 only.
39179 2017-01-20 Marek Polacek <polacek@redhat.com>
39182 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39185 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39187 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39189 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39190 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39191 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39193 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39196 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39197 in big-endian mode when they are not a single duplicated value.
39199 2017-01-20 Richard Biener <rguenther@suse.de>
39201 * BASE-VER: Bump to 7.0.1.
39203 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39205 * omp-low.c (omplow_simd_context): New struct. Use it...
39206 (lower_rec_simd_input_clauses): ...here and...
39207 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39208 references to idx, lane, max_vf, is_simt.
39210 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39212 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39215 2017-01-20 Martin Jambor <mjambor@suse.cz>
39217 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39218 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39220 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39221 (GTFILES): Rename hsa.c to hsa-common.c.
39222 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39223 * hsa-dump.c: Likewise.
39224 * hsa-gen.c: Likewise.
39225 * hsa-regalloc.c: Likewise.
39226 * ipa-hsa.c: Likewise.
39227 * omp-expand.c: Likewise.
39228 * omp-low.c: Likewise.
39229 * toplev.c: Likewise.
39231 2017-01-20 Marek Polacek <polacek@redhat.com>
39234 * doc/invoke.texi: Document -Wduplicated-branches.
39235 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39236 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39237 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39238 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39239 return 0 only when not OEP_LEXICOGRAPHIC.
39240 (fold_build_cleanup_point_expr): Use the expression
39241 location when building CLEANUP_POINT_EXPR.
39242 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39243 * tree.c (add_expr): Handle error_mark_node.
39245 2017-01-20 Martin Liska <mliska@suse.cz>
39248 * tree-profile.c (init_ic_make_global_vars): Do not call
39250 (gimple_init_gcov_profiler): Likewise.
39252 2017-01-20 Martin Liska <mliska@suse.cz>
39255 * cgraph.h (maybe_create_reference): Remove argument and
39257 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39259 * ipa-cp.c (create_specialized_node): Likewise.
39260 * symtab.c (symtab_node::maybe_create_reference): Handle
39261 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39263 2017-01-20 Martin Liska <mliska@suse.cz>
39265 * read-rtl-function.c (function_reader::create_function): Use
39266 build_decl instread of build_decl_stat.
39268 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39270 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39271 * config/i386/avx512dqintrin.h: Ditto.
39272 * config/i386/avx512fintrin.h: Ditto.
39273 * config/i386/i386-builtin-types.def: Add new types.
39274 * config/i386/i386.c: Handle new types.
39275 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39276 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39277 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39278 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39279 (__builtin_ia32_kshiftridi): New.
39280 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39282 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39286 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39287 define to rs6000_init_stack_protect_guard.
39288 (rs6000_init_stack_protect_guard): New function.
39290 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39291 Yunqiang Su <yunqiang.su@imgtec.com>
39293 * config.gcc (supported_defaults): Add madd4.
39294 (with_madd4): Add validation.
39295 (all_defaults): Add madd4.
39296 * config/mips/mips.opt (mmadd4): New option.
39297 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39299 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39301 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39302 (ISA_HAS_FUSED_MADD4): Likewise.
39303 * doc/invoke.texi (-mmadd4): Document the new option.
39304 * doc/install.texi (--with-madd4): Document the new option.
39306 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39308 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39309 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39310 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39311 (aarch64_init_pauth_hint_builtins): New.
39312 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39313 (aarch64_expand_builtin): Expand new builtins.
39315 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39317 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39318 * combine-stack-adj.c (no_unhandled_cfa): Handle
39319 REG_CFA_TOGGLE_RA_MANGLE.
39320 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39321 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39322 info for return address signing.
39323 (aarch64_expand_epilogue): Likewise.
39325 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39327 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39328 * config/aarch64/aarch64-protos.h
39329 (aarch64_return_address_signing_enabled): New declaration.
39330 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39332 (aarch64_expand_prologue): Sign return address before it's pushed onto
39334 (aarch64_expand_epilogue): Authenticate return address fetched from
39336 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39337 (aarch64_attributes): New function attributes for "sign-return-address".
39338 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39339 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39340 ("*do_return"): Generate combined instructions according to key index.
39341 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39342 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39344 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39345 * config/aarch64/aarch64.opt (msign-return-address=): New.
39346 * doc/extend.texi (AArch64 Function Attributes): Documents
39347 "sign-return-address=".
39348 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39350 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39352 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39353 overall option summary.
39355 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39357 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39358 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39359 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39360 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39362 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39364 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39365 -mpower9-minmax by default for -mcpu=power9.
39366 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39367 128-bit floating point.
39369 2017-01-20 Alan Modra <amodra@gmail.com>
39371 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39372 optimizing for size.
39374 2017-01-20 Alan Modra <amodra@gmail.com>
39377 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39378 for strcmp and strncmp from corresponding builtin decl.
39380 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39382 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39383 instead of i386/rtems-64.h.
39384 * config/i386/rtems-64.h: Remove.
39386 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39390 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39392 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39394 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39396 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39397 Change int to HOST_WIDE_INT.
39398 * config/aarch64/aarch64-protos.h
39399 (aarch64_simd_gen_const_vector_dup): Likewise.
39400 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39402 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39404 * langhooks-def.h (lhd_type_for_size): New decl.
39405 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39406 * langhooks.c (lhd_type_for_size): New function, taken from
39409 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39411 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39412 define_bypass for CR latency.
39413 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39414 (power9-alu2): Add define_bypass for CR latency.
39416 (power9-mul): Update insn latency.
39417 (power9-mul-compare): Update insn latency, bypass latency and remove
39420 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39422 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39424 * config/aarch64/aarch64.md
39425 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39426 aarch64_nopcrelative_literal_loads.
39427 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39429 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39431 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39432 TARGET_LOONGSON_3A.
39433 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39435 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39438 * config.gcc (supported_defaults): Add lxc1-sxc1.
39439 (with_lxc1_sxc1): Add validation.
39440 (all_defaults): Add lxc1-sxc1.
39441 * config/mips/mips.opt (mlxc1-sxc1): New option.
39442 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39444 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39445 __mips_no_lxc1_sxc1.
39446 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39447 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39448 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39450 2017-01-19 Richard Biener <rguenther@suse.de>
39452 PR tree-optimization/72488
39453 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39454 sure to restore SSA info.
39455 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39457 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39459 PR rtl-optimization/79121
39460 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39461 of the inner type when shifting an extended value.
39463 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39466 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39467 interposable aliases.
39469 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39472 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39473 Use the evmergelohi instruction.
39474 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39475 (mov_sitf_e500_subreg8_2_be): Likewise.
39476 (mov_sitf_e500_subreg12_2_le): Likewise.
39477 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39478 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39479 (mov_sitf_e500_subreg8_2_le): Likewise.
39480 (mov_sitf_e500_subreg12_2_be): Likewise.
39482 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39484 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39485 attribute from vecsimple to vecperm.
39486 (altivec_vbpermq2): Likewise.
39488 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39491 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39493 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39494 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39495 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39496 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39497 case where N arg is SIZE_MAX.
39498 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39499 (cmpstrsi): Add pattern.
39501 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39503 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39504 __builtin_vec_revb builtins.
39505 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39506 built-in functions to support generation of the ISA 3.0 XXBR<x>
39507 vector byte reverse instructions.
39508 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39509 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39510 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39511 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39512 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39513 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39514 (P9V_BUILTIN_VEC_REVB): Likewise.
39515 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39516 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39517 (p9_xxbrq_v16qi): Likewise.
39518 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39519 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39520 (p9_xxbrh_v8hi): Likewise.
39521 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39522 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39523 vec_revb built-in functions.
39525 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39527 PR rtl-optimization/78952
39528 * config/i386/i386.md (any_extract): New code iterator.
39529 (*insvqi_2): Use any_extract for source operand.
39530 (*insvqi_3): Use any_shiftrt for source operand.
39532 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39534 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39536 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39538 2017-01-18 Matthias Klose <doko@ubuntu.com>
39540 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39542 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39544 * config/rs6000/altivec.h (vec_bperm): Change #define.
39545 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39546 (altivec_vbpermq2): New define_insn.
39547 (altivec_vbpermd): Likewise.
39548 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39549 function interface.
39550 (VBPERMD): Likewise.
39551 (VBPERM): New polymorphic function interface.
39552 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39553 Add entries for P9V_BUILTIN_VEC_VBPERM.
39554 * doc/extend.texi: Add interfaces for vec_bperm.
39556 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39558 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39559 first letter of error messages.
39560 (s390_resolve_overloaded_builtin): Likewise.
39561 * config/s390/s390.c (s390_expand_builtin): Likewise.
39562 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39563 (s390_valid_target_attribute_inner_p): Likewise.
39564 * config/s390/s390.md ("tabort"): Likewise.
39566 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39568 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39569 (ISA_AVOID_DIV_HILO): New macro.
39570 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39571 (ISA_HAS_DDIV): Likewise.
39573 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39575 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39577 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39579 * doc/invoke.texi (fabi-version): Spelling fix.
39581 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39584 * doc/invoke.texi (fabi-version): Mention mangling fix for
39587 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39590 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39592 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39595 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39596 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39598 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39600 (stack_protect_test): Ditto.
39601 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39602 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39604 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39605 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39606 -mstack-protector-guard-offset=.
39607 (RS/6000 and PowerPC Options): Ditto.
39609 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39611 * config/i386/i386.h (MASK_CLASS_P): New define.
39612 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39613 there are no registers from different register sets also when
39614 mask registers are used. Update function comment.
39615 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39616 to (*k/*r) and (*k/*km) alternatives.
39618 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39620 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39621 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39622 (EH_RETURN_HANDLER_RTX): New define.
39623 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39624 Force frame pointer in EH return functions.
39625 (aarch64_expand_epilogue): Add barrier for eh_return.
39626 (aarch64_final_eh_return_addr): Remove.
39627 (aarch64_eh_return_handler_rtx): New function.
39628 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39630 (aarch64_eh_return_handler_rtx): New prototype.
39632 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39634 * config/rs6000/altivec.h (vec_rlmi): New #define.
39635 (vec_vrlnm): Likewise.
39636 (vec_rlnm): Likewise.
39637 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39638 (UNSPEC_VRLNM): Likewise.
39639 (VIlong): New mode iterator.
39640 (altivec_vrl<VI_char>mi): New define_insn.
39641 (altivec_vrl<VI_char>nm): Likewise.
39642 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39644 (VRLDNM): Likewise.
39645 (RLNM): New polymorphic function entry.
39646 (VRLWMI): New monomorphic function entry.
39647 (VRLDMI): Likewise.
39648 (RLMI): New polymorphic function entry.
39649 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39650 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39651 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39654 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39657 * dwarf2out.c (field_byte_offset): Restore the
39658 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39659 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39660 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39663 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39666 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39668 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39671 * dwarf2out.c (add_data_member_location_attribute): For constant
39672 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39673 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39674 DW_AT_byte_size attributes.
39676 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39678 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39679 after forcing to constant memory when the code model is medium.
39681 2017-01-17 Julia Koval <julia.koval@intel.com>
39684 * config/i386/avx512fintrin.h
39685 (_mm512_i32gather_ps): Change __addr type to void const*.
39686 (_mm512_mask_i32gather_ps): Ditto.
39687 (_mm512_i32gather_pd): Ditto.
39688 (_mm512_mask_i32gather_pd): Ditto.
39689 (_mm512_i64gather_ps): Ditto.
39690 (_mm512_mask_i64gather_ps): Ditto.
39691 (_mm512_i64gather_pd): Ditto.
39692 (_mm512_mask_i64gather_pd): Ditto.
39693 (_mm512_i32gather_epi32): Ditto.
39694 (_mm512_mask_i32gather_epi32): Ditto.
39695 (_mm512_i32gather_epi64): Ditto.
39696 (_mm512_mask_i32gather_epi64): Ditto.
39697 (_mm512_i64gather_epi32): Ditto.
39698 (_mm512_mask_i64gather_epi32): Ditto.
39699 (_mm512_i64gather_epi64): Ditto.
39700 (_mm512_mask_i64gather_epi64): Ditto.
39701 (_mm512_i32scatter_ps): Change __addr type to void*.
39702 (_mm512_mask_i32scatter_ps): Ditto.
39703 (_mm512_i32scatter_pd): Ditto.
39704 (_mm512_mask_i32scatter_pd): Ditto.
39705 (_mm512_i64scatter_ps): Ditto.
39706 (_mm512_mask_i64scatter_ps): Ditto.
39707 (_mm512_i64scatter_pd): Ditto.
39708 (_mm512_mask_i64scatter_pd): Ditto.
39709 (_mm512_i32scatter_epi32): Ditto.
39710 (_mm512_mask_i32scatter_epi32): Ditto.
39711 (_mm512_i32scatter_epi64): Ditto.
39712 (_mm512_mask_i32scatter_epi64): Ditto.
39713 (_mm512_i64scatter_epi32): Ditto.
39714 (_mm512_mask_i64scatter_epi32): Ditto.
39715 (_mm512_i64scatter_epi64): Ditto.
39716 (_mm512_mask_i64scatter_epi64): Ditto.
39717 * config/i386/avx512pfintrin.h
39718 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39719 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39720 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39721 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39722 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39723 (_mm512_prefetch_i32scatter_ps): Ditto.
39724 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39725 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39726 (_mm512_prefetch_i64scatter_pd): Ditto.
39727 (_mm512_prefetch_i64scatter_ps): Ditto.
39728 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39729 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39730 * config/i386/avx512vlintrin.h
39731 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39732 (_mm_mmask_i32gather_ps): Ditto.
39733 (_mm256_mmask_i32gather_pd): Ditto.
39734 (_mm_mmask_i32gather_pd): Ditto.
39735 (_mm256_mmask_i64gather_ps): Ditto.
39736 (_mm_mmask_i64gather_ps): Ditto.
39737 (_mm256_mmask_i64gather_pd): Ditto.
39738 (_mm_mmask_i64gather_pd): Ditto.
39739 (_mm256_mmask_i32gather_epi32): Ditto.
39740 (_mm_mmask_i32gather_epi32): Ditto.
39741 (_mm256_mmask_i32gather_epi64): Ditto.
39742 (_mm_mmask_i32gather_epi64): Ditto.
39743 (_mm256_mmask_i64gather_epi32): Ditto.
39744 (_mm_mmask_i64gather_epi32): Ditto.
39745 (_mm256_mmask_i64gather_epi64): Ditto.
39746 (_mm_mmask_i64gather_epi64): Ditto.
39747 (_mm256_i32scatter_ps): Change __addr type to void*.
39748 (_mm256_mask_i32scatter_ps): Ditto.
39749 (_mm_i32scatter_ps): Ditto.
39750 (_mm_mask_i32scatter_ps): Ditto.
39751 (_mm256_i32scatter_pd): Ditto.
39752 (_mm256_mask_i32scatter_pd): Ditto.
39753 (_mm_i32scatter_pd): Ditto.
39754 (_mm_mask_i32scatter_pd): Ditto.
39755 (_mm256_i64scatter_ps): Ditto.
39756 (_mm256_mask_i64scatter_ps): Ditto.
39757 (_mm_i64scatter_ps): Ditto.
39758 (_mm_mask_i64scatter_ps): Ditto.
39759 (_mm256_i64scatter_pd): Ditto.
39760 (_mm256_mask_i64scatter_pd): Ditto.
39761 (_mm_i64scatter_pd): Ditto.
39762 (_mm_mask_i64scatter_pd): Ditto.
39763 (_mm256_i32scatter_epi32): Ditto.
39764 (_mm256_mask_i32scatter_epi32): Ditto.
39765 (_mm_i32scatter_epi32): Ditto.
39766 (_mm_mask_i32scatter_epi32): Ditto.
39767 (_mm256_i32scatter_epi64): Ditto.
39768 (_mm256_mask_i32scatter_epi64): Ditto.
39769 (_mm_i32scatter_epi64): Ditto.
39770 (_mm_mask_i32scatter_epi64): Ditto.
39771 (_mm256_i64scatter_epi32): Ditto.
39772 (_mm256_mask_i64scatter_epi32): Ditto.
39773 (_mm_i64scatter_epi32): Ditto.
39774 (_mm_mask_i64scatter_epi32): Ditto.
39775 (_mm256_i64scatter_epi64): Ditto.
39776 (_mm256_mask_i64scatter_epi64): Ditto.
39777 (_mm_i64scatter_epi64): Ditto.
39778 (_mm_mask_i64scatter_epi64): Ditto.
39779 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39780 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39781 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39782 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39783 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39784 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39785 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39786 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39787 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39788 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39789 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39790 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39791 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39792 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39793 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39794 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39795 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39796 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39797 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39798 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39799 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39800 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39801 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39802 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39803 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39804 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39805 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39806 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39807 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39808 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39809 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39810 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39811 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39812 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39813 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39814 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39815 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39816 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39817 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39818 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39819 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39820 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39821 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39822 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39823 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39824 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39825 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39826 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39827 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39828 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39829 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39830 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39831 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39832 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39833 definitions accordingly.
39835 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39836 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39839 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39842 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39845 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39846 hard regs for inner regno.
39848 2017-01-17 Martin Liska <mliska@suse.cz>
39851 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39852 assumption and add comment.
39854 2017-01-17 Nathan Sidwell <nathan@acm.org>
39856 * ipa-visibility.c (localize_node): New function, broken out of ...
39857 (function_and_variable_visibility): ... here. Call it.
39859 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39861 PR middle-end/77445
39862 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39863 correctly set frequency of oudgoing edge.
39864 (duplicate_thread_path): Fix profile updating.
39866 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39869 * configure.ac: Add GCC_BASE_VER.
39870 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39871 version from BASE-VER file.
39872 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39873 (gcc.o): Depend on $(BASEVER).
39874 * common.opt (dumpfullversion): New option.
39875 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39876 * doc/invoke.texi: Document -dumpfullversion.
39877 * doc/install.texi: Document --with-gcc-major-version-only.
39878 * configure: Regenerated.
39880 2017-01-17 Richard Biener <rguenther@suse.de>
39882 PR tree-optimization/71433
39883 * tree-vrp.c (register_new_assert_for): Merge same asserts
39884 on all incoming edges.
39885 (process_assert_insertions_for): Handle insertions at the
39888 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39890 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39891 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39893 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39896 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39899 2017-01-17 Alan Modra <amodra@gmail.com>
39902 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39903 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39904 symbolic stack limit when pic.
39906 2017-01-16 Martin Sebor <msebor@redhat.com>
39908 PR tree-optimization/78608
39909 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39911 2017-01-16 Jeff Law <law@redhat.com>
39914 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39915 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39916 for several include directories that may be relative to sysroot.
39917 * config/i386/x-mingw32 (gplus_includedir): Define.
39918 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39919 (native_system_includedir): Likewise.
39920 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39921 override if TARGET_SYSTEM_ROOT is defined.
39922 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39924 PR tree-optimization/79090
39925 PR tree-optimization/33562
39926 PR tree-optimization/61912
39927 PR tree-optimization/77485
39928 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39929 and computed trims into the dump file.
39931 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39933 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39935 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39938 * gimplify.c (gimplify_init_constructor): If want_value and
39939 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39943 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39944 sequence. Formatting fixes.
39945 (doloop_optimize): Formatting fixes.
39948 * gcc.c (debug_level_greater_than_spec_func): New function.
39949 (static_spec_functions): Add debug-level-gt spec function.
39950 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39952 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39953 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39954 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39955 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39956 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39957 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39959 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39961 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39962 QImode fixups to general and mask registers only.
39964 2017-01-16 Carl Love <cel@us.ibm.com>
39966 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39967 for built-in functions
39968 vector signed char vec_nabs (vector signed char)
39969 vector signed short vec_nabs (vector signed short)
39970 vector signed int vec_nabs (vector signed int)
39971 vector signed long long vec_nabs (vector signed long long)
39972 vector float vec_nabs (vector float)
39973 vector double vec_nabs (vector double)
39974 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39976 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39977 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39978 * doc/extend.texi: Update the documentation file for the new built-in
39981 2017-01-16 Martin Sebor <msebor@redhat.com>
39983 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39986 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39988 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39989 UNSPEC_VSX__XXSPLTD to require special splat handling.
39991 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39994 * system.h: Poison strndup.
39996 2017-01-16 Alan Modra <amodra@gmail.com>
39999 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40002 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40004 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40006 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40008 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40009 call recog here. Assert that INSN_CODE (insn) is non-negative.
40011 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40014 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40016 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40017 in the currently scheduled RTL fragment.
40019 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40021 PR rtl-optimization/78751
40022 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40025 2017-01-14 Jeff Law <law@redhat.com>
40027 PR tree-optimization/79090
40028 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40029 variable length stores.
40030 (compute_trims): Delete dead assignment to *trim_tail.
40031 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40034 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40036 PR rtl-optimization/78626
40037 PR rtl-optimization/78727
40038 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40039 of a block, and split such blocks after everything else is finished.
40041 2017-01-14 Alan Modra <amodra@gmail.com>
40044 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40045 target legitimate_combined_insn.
40046 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40047 (rs6000_legitimate_combined_insn): New function.
40048 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40050 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40051 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40052 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40054 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40056 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40058 2017-01-13 Jeff Law <law@redhat.com>
40060 PR tree-optimization/33562
40061 PR tree-optimization/61912
40062 PR tree-optimization/77485
40063 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40065 (delete_dead_assignment): Likewise.
40066 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40067 statement to delete_dead_call and delete_dead_assignment.
40069 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40072 * substring-locations.c (format_warning_va): Strengthen case 1 so
40073 that both endpoints of the substring must be within the format
40074 range for just the substring to be printed.
40076 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40078 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40079 * config/i386/i386.c (ix86_target_string): Add missing options
40080 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40081 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40082 flags_other and ix86_target_other to flags2_other. Display unknown
40084 (ix86_valid_target_attribute_inner_p): Add missing options and
40085 reorder options by implied ISAs, as in ix86_target_string.
40087 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40089 * hash-table.h (hash_table::too_empty_p): New function.
40090 (hash_table::expand): Use it.
40091 (hash_table::traverse): Likewise.
40092 (hash_table::empty_slot): Use sizeof (value_type) instead of
40093 sizeof (PTR) to convert bytes to elements. Shrink the table
40094 if the current size is excessive for the current number of
40097 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40099 * ira-costs.c (record_reg_classes): Break from the inner loop
40100 early once alt_fail is known to be true. Update outer loop
40101 handling accordingly.
40103 2017-01-13 Jeff Law <law@redhat.com>
40105 * tree-ssa-dse.c (decrement_count): New function.
40106 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40107 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40108 when we know the partially dead statement is a mem* function.
40110 PR tree-optimization/61912
40111 PR tree-optimization/77485
40112 * tree-ssa-dse.c: Include expr.h.
40113 (maybe_trim_constructor_store): New function.
40114 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40116 PR tree-optimization/33562
40117 PR tree-optimization/61912
40118 PR tree-optimization/77485
40119 * doc/invoke.texi: Document new dse-max-object-size param.
40120 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40121 * tree-ssa-dse.c: Include params.h.
40122 (dse_store_status): New enum.
40123 (initialize_ao_ref_for_dse): New, partially extracted from
40125 (valid_ao_ref_for_dse, normalize_ref): New.
40126 (setup_live_bytes_from_ref, compute_trims): Likewise.
40127 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40128 (maybe_trim_partially_dead_store): Likewise.
40129 (maybe_trim_complex_store): Likewise.
40130 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40131 Track what bytes live from the original store. Return tri-state
40132 for dead, partially dead or live.
40133 (dse_dom_walker): Add constructor, destructor and new private members.
40134 (delete_dead_call, delete_dead_assignment): New extracted from
40136 (dse_optimize_stmt): Make a member of dse_dom_walker.
40137 Use initialize_ao_ref_for_dse.
40139 PR tree-optimization/33562
40140 PR tree-optimization/61912
40141 PR tree-optimization/77485
40142 * sbitmap.h (bitmap_count_bits): Prototype.
40143 (bitmap_clear_range, bitmap_set_range): Likewise.
40144 * sbitmap.c (bitmap_clear_range): New function.
40145 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40147 2017-01-13 Martin Liska <mliska@suse.cz>
40150 * function.c (set_cfun): Add new argument force.
40151 * function.h (set_cfun): Likewise.
40152 * ipa-inline-transform.c (inline_call): Use the function when
40153 strict alising from is dropped for function we inline to.
40155 2017-01-13 Richard Biener <rguenther@suse.de>
40157 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40158 for dumping GIMPLE INTEGER_CSTs.
40160 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40162 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40163 to 201112L since C++17.
40165 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40168 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40169 if -fsanitize=kernel-address is present.
40171 2017-01-13 Richard Biener <rguenther@suse.de>
40173 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40174 as _Literal ( type ) number in case usual suffixes do not
40175 preserve all information.
40177 2017-01-13 Richard Biener <rguenther@suse.de>
40179 PR tree-optimization/77283
40180 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40181 and ssa-iterators.h.
40182 (is_feasible_trace): Implement a cost model based on joiner
40185 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40188 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40189 char or short to __float128/_Float128 directly.
40191 2017-01-12 Martin Sebor <msebor@redhat.com>
40193 to -Wformat-overflow.
40194 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40195 (min_bytes_remaining): Same.
40196 (get_string_length): Same.
40197 (format_string): Same.
40198 (format_directive): Same.
40200 (pass_sprintf_length::handle_gimple_call): Same.
40202 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40204 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40205 info.nowrite calls with no lhs that can't throw. Return bool
40206 whether gsi_remove has been called or not.
40207 (pass_sprintf_length::handle_gimple_call): Return bool whether
40208 try_substitute_return_value called gsi_remove. Formatting fix.
40209 (pass_sprintf_length::execute): Don't use gsi_remove if
40210 handle_gimple_call returned true.
40213 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40214 be removed due to side-effects, don't remove following barrier nor
40215 turn the successor edge into fallthru edge.
40217 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40220 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40221 element-reversing loads and stores as not swappable.
40223 2017-01-12 Nathan Sidwell <nathan@acm.org>
40224 Nicolai Stange <nicstange@gmail.com>
40226 * combine.c (try_combine): Don't ignore result of overlap checking
40227 loop. Combine overlap & asm check into single loop.
40229 2017-01-12 Richard Biener <rguenther@suse.de>
40231 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40232 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40234 2017-01-12 Richard Biener <rguenther@suse.de>
40236 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40237 and TS_TARGET_OPTION directly derive from TS_BASE.
40238 * tree-core.h (tree_optimization_option): Derive from tree_base.
40239 (tree_target_option): Likewise.
40241 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40243 * config/i386/i386.c (memory_address_length): Increase len
40244 only when rip_relative_addr_p returns false.
40246 2017-01-11 Julia Koval <julia.koval@intel.com>
40248 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40249 (OPTION_MASK_ISA_SGX_SET): New.
40250 (ix86_handle_option): Handle OPT_msgx.
40251 * config.gcc: Added sgxintrin.h.
40252 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40253 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40254 * config/i386/i386.c (ix86_target_string): Add -msgx.
40256 (ix86_option_override_internal): Handle new options.
40257 (ix86_valid_target_attribute_inner_p): Add sgx.
40258 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40259 * config/i386/i386.opt: Add msgx.
40260 * config/i386/sgxintrin.h: New file.
40261 * config/i386/x86intrin.h: Add sgxintrin.h.
40263 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40266 * fold-const.c (maybe_nonzero_address): Return 1 for function
40268 (tree_single_nonzero_warnv_p): Don't handle function local objects
40272 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40275 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40278 * opts.c: Include "spellcheck.h"
40279 (struct string_fragment): New struct.
40280 (struct edit_distance_traits<const string_fragment &>): New
40282 (get_closest_sanitizer_option): New function.
40283 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40285 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40287 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40289 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40290 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40291 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40292 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40293 for initial die_offset if dwarf_split_debug_info.
40294 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40295 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40297 (output_skeleton_debug_sections): Formatting fix. Use
40298 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40299 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40301 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40303 * config/arm/cortex-a53.md: Add bypasses for
40304 cortex_a53_r2f_cvt.
40305 (cortex_a53_r2f): Only use for transfers.
40306 (cortex_a53_f2r): Likewise.
40307 (cortex_a53_r2f_cvt): Add reservation for conversions.
40308 (cortex_a53_f2r_cvt): Likewise.
40310 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40312 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40313 to all inlined functions, change static to extern.
40315 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40318 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40320 (arm_assemble_integer): Likewise.
40322 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40324 * config.gcc: Use new awk script to check CPU, FPU and architecture
40325 parameters for --with-... options.
40326 * config/arm/parsecpu.awk: New file
40327 * config/arm/arm-cpus.in: New file.
40328 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40330 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40332 * config/arm/t-arm: Update dependency rules.
40333 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40334 of processing .def files.
40335 * config/arm/genopt.sh: Deleted.
40336 * config/arm/gentune.sh: Deleted.
40337 * config/arm/arm-cores.def: Deleted.
40338 * config/arm/arm-arches.def: Deleted.
40339 * config/arm/arm-fpus.def: Deleted.
40340 * config/arm/arm-tune.md: Regenerated.
40341 * config/arm/arm-tables.opt: Regenerated.
40342 * config/arm/arm-cpu.h: New generated file.
40343 * config/arm/arm-cpu-data.h: New generated file.
40344 * config/arm/arm-cpu-cdata.h: New generated file.
40346 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40349 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40351 (input_varpool_node): Unpack dynamically_initialized bit.
40353 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40355 PR rtl-optimization/79032
40356 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40357 the alignment of the adjusted memory reference against that of MODE,
40358 instead of the alignment of the original memory reference.
40360 2017-01-11 Martin Jambor <mjambor@suse.cz>
40362 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40364 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40365 decorated functions.
40367 2017-01-11 Richard Biener <rguenther@suse.de>
40369 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40370 set range/nonnull info for PHI results. Do not set it on
40371 stmts marked for removal.
40373 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40375 * expr.c (store_field): In the bitfield case, fetch the return value
40376 from the registers before applying a single big-endian adjustment.
40377 Always do a final load for a BLKmode value not larger than a word.
40379 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40382 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40383 that we correctly handle column numbers greater than
40384 LINE_MAP_MAX_COLUMN_NUMBER.
40386 2017-01-10 Martin Sebor <msebor@redhat.com>
40388 PR middle-end/78245
40389 * gimple-ssa-sprintf.c (get_destination_size): Call
40390 {init,fini}object_sizes.
40391 * tree-object-size.c (addr_object_size): Adjust.
40392 (pass_through_call): Adjust.
40393 (pass_object_sizes::execute): Adjust.
40394 * tree-object-size.h (fini_object_sizes): Declare.
40396 2017-01-10 Martin Sebor <msebor@redhat.com>
40398 PR tree-optimization/78775
40399 * builtins.c (get_size_range): Move...
40400 * calls.c: ...to here.
40401 (alloc_max_size): Accept zero argument.
40402 (operand_signed_p): Remove.
40403 (maybe_warn_alloc_args_overflow): Call get_size_range.
40404 * calls.h (get_size_range): Declare.
40406 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40408 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40409 from TI's devices.csv file as of September 2016.
40410 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40412 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40414 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40415 * doc/invoke.texi: Likewise.
40416 * doc/md.texi: Likewise.
40417 * doc/objc.texi: Likewise.
40419 2017-01-10 Joshua Conner <joshconner@google.com>
40421 * config/arm/fuchsia-elf.h: New file.
40422 * config/fuchsia.h: New file.
40423 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40424 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40426 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40428 2016-01-10 Richard Biener <rguenther@suse.de>
40430 PR tree-optimization/79034
40431 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40432 Propagate out degenerate PHIs in the joiner.
40434 2017-01-10 Martin Liska <mliska@suse.cz>
40436 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40437 (sort_congruence_classes_by_decl_uid): Likewise.
40438 (sort_congruence_class_groups_by_decl_uid): Likewise.
40439 (sem_item_optimizer::merge_classes): Sort class, groups in these
40440 classes and members in the groups by DECL_UID of declarations.
40441 This would make merge operations stable.
40443 2017-01-10 Martin Liska <mliska@suse.cz>
40445 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40446 usage of m_classes_vec.
40447 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40448 (sem_item_optimizer::get_group_by_hash): Likewise.
40449 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40450 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40451 (sem_item_optimizer::verify_classes): Likewise.
40452 (sem_item_optimizer::process_cong_reduction): Likewise.
40453 (sem_item_optimizer::dump_cong_classes): Likewise.
40454 (sem_item_optimizer::merge_classes): Likewise.
40455 * ipa-icf.h (congruence_class_hash): Rename from
40456 congruence_class_group_hash. Remove declaration of m_classes_vec.
40458 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40460 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40461 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40462 * config.gcc: Add avx512vpopcntdqintrin.h.
40463 * config/i386/avx512vpopcntdqintrin.h: New.
40464 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40465 * config/i386/i386-builtin-types.def: Add new types.
40466 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40467 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40468 __builtin_ia32_vpopcountq_v8di_mask): New.
40469 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40470 __AVX512VPOPCNTDQ__.
40471 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40472 (PTA_AVX512VPOPCNTDQ): Define.
40473 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40474 TARGET_AVX512VPOPCNTDQ_P): Define.
40475 * config/i386/i386.opt: Add mavx512vpopcntdq.
40476 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40477 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40479 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40481 PR middle-end/77484
40482 * predict.def (PRED_CALL): Set to 67.
40484 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40486 * expr.c (store_field): In the bitfield case, if the value comes from
40487 a function call and is of an aggregate type returned in registers, do
40488 not modify the field mode; extract the value in all cases if the mode
40489 is BLKmode and the size is not larger than a word.
40491 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40494 * config/i386/cpuid.h: Fix undefined behavior.
40496 2017-01-04 Jeff Law <law@redhat.com>
40498 PR tree-optimization/79007
40499 PR tree-optimization/67955
40500 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40501 conservative for pt.null when flag_non_call_exceptions is on.
40503 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40505 PR translation/79019
40506 PR translation/79020
40507 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40508 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40509 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40511 * config/avr/avr.opt (maccumulate-args): Likewise.
40512 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40513 * common.opt (freport-bug): Likewise.
40514 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40515 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40516 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40517 translatable string.
40518 * config/i386/i386.c (function_value_32): Likewise.
40519 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40520 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40522 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40523 * common/config/msp430/msp430-common.c (msp430_handle_option):
40525 * symtab.c (symtab_node::verify_base): Likewise.
40526 * opts.c (set_debug_level): Likewise.
40527 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40528 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40529 missing whitespace to translatable strings.
40530 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40531 * config/sh/superh.h: Likewise.
40532 * config/i386/xopintrin.h: Likewise.
40533 * config/i386/znver1.md: Likewise.
40534 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40535 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40536 * double-int.h (struct double_int): Likewise.
40537 * double-int.c (div_and_round_double): Likewise.
40538 * wide-int.cc: Likewise.
40539 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40540 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40541 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40542 (crossjumps_occurred): ... this.
40543 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40546 PR tree-optimization/78899
40547 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40548 returning bool return struct loop *, NULL for failure and the new
40550 (versionable_outer_loop_p): Don't version outer loop if it has
40551 dont_vectorized bit set.
40552 (tree_if_conversion): When versioning outer loop, ensure
40553 tree_if_conversion is performed also on the inner loop of the
40554 non-vectorizable outer loop copy.
40555 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40556 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40557 prevent vectorization of it.
40558 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40559 the outer loop vectorization of the non-scalar version is attempted
40560 before vectorization of the inner loop in scalar version. If
40561 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40562 vectorization of its inner loop.
40563 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40564 has 2 inner loops, rename also on edges from bb whose single pred
40565 is outer_loop->header. Fix typo in function comment.
40567 2017-01-09 Martin Sebor <msebor@redhat.com>
40570 * asan.c (asan_emit_stack_protection): Increase local buffer size
40571 to avoid snprintf truncation warning.
40573 2017-01-09 Andrew Pinski <apinski@cavium.com>
40575 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40576 to reference thunderx2t99 for the tuning structure
40577 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40579 (thunderx2t99_extra_costs): This.
40580 * config/aarch64/aarch64-tune.md: Regenerate.
40581 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40582 (vulcan_addrcost_table): This.
40583 (vulcan_regmove_cost): Rename to ...
40584 (thunderx2t99_regmove_cost): This.
40585 (vulcan_vector_cost): Rename to ...
40586 (thunderx2t99_vector_cost): this.
40587 (vulcan_branch_cost): Rename to ...
40588 (thunderx2t99_branch_cost): This.
40589 (vulcan_tunings): Rename to ...
40590 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40591 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40593 2017-01-09 Martin Jambor <mjambor@suse.cz>
40597 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40598 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40599 (propagate_vr_accross_jump_function): Use the above function for all
40600 value range computations for pass-through jump functions and type
40601 converasion from explicit value range values.
40602 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40603 parameters from TYPE_ARG_TYPES.
40604 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40605 (ipa_write_node_info): Stream type of the actual argument.
40606 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40608 2017-01-09 Martin Liska <mliska@suse.cz>
40611 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40612 (lookup_compiler): Do not show error message with have_E.
40614 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40616 PR tree-optimization/78938
40617 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40618 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40619 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40620 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40623 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40625 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40628 2017-01-09 Richard Biener <rguenther@suse.de>
40630 PR tree-optimization/78997
40631 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40632 name condition properly.
40634 2017-01-09 Richard Biener <rguenther@suse.de>
40637 * dwarf2out.c (is_cxx): New overload with context.
40638 (is_naming_typedef_decl): Use it.
40640 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40642 * invoke.texi (Option Summary): Correct spacing in option lists
40643 and add line breaks to fix over-long lines.
40645 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40647 PR middle-end/17660
40649 * extend.texi (Common Variable Attributes): Add xref to GCC
40650 Internals manual to explain mode attribute keywords.
40652 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40655 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40656 and Preprocessor Options.
40657 (Options for Linking): Document -pthread here....
40658 (RS/6000 and PowerPC Options): ...not here.
40659 (Solaris 2 Options): ...or here.
40660 * doc/cppopts.texi: Document -pthread.
40662 2017-01-08 Martin Sebor <msebor@redhat.com>
40664 PR middle-end/77708
40665 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40666 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40667 New member functions.
40668 (format_directive): Used them.
40670 (pass_sprintf_length::handle_gimple_call): Same.
40671 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40672 to avoid truncation for any argument.
40673 (extract_affine_mul): Same.
40674 * tree.c (get_file_function_name): Same.
40676 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40678 PR middle-end/77484
40679 * predict.def (PRED_INDIR_CALL): Set to 86.
40681 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40683 PR preprocessor/54124
40684 * doc/cppopts.texi: Reformat -d subtable to list the full name
40685 of the options. Add cross-reference to the docs for the general
40686 compiler -d options.
40687 * doc/invoke.texi (Developer Options): Add cross-reference to the
40688 preprocessor-specific -d option documentation.
40690 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40692 PR preprocessor/13498
40693 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40694 redudant material, and reflect new command-line options.
40695 (System Headers): Likewise.
40697 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40699 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40700 -isystem, and -idirafter. Copy-edit.
40701 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40702 default for -ftrack-macro-expansion. Delete obsolete and
40703 badly-formatted implementation details about -fdebug-cpp output.
40704 * doc/cppwarnopts.texi: Copy-edit.
40706 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40709 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40710 that the transition from a max line width >= 1<<10 to narrower
40711 lines works correctly.
40713 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40715 * doc/options.texi (PerFunction): New.
40716 * opt-functions.awk (switch_flags): Map both Optimization and
40717 PerFunction to CL_OPTIMIZATION.
40718 * opth-gen.awk: Test for PerFunction flag along with
40720 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40721 it only when the latter is present. Skip those that don't in
40722 the hash function generator.
40723 * common.opt (fvar-tracking): Mark as PerFunction instead of
40725 (fvar-tracking-assignments): Likewise.
40726 (fvar-tracking-assignments-toggle): Likewise.
40727 (fvar-tracking-uninit): Likewise.
40729 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40731 PR translation/79018
40732 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40735 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40738 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40739 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40740 TARGET_LONG_JUMP_TABLE_OFFSETS.
40741 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40742 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40743 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40744 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40745 * config/m68k/m68k.md (tablejump expander): Likewise.
40746 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40747 TARGET_LONG_JUMP_TABLE_OFFSETS.
40748 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40749 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40751 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40752 David Holsgrove <david.holsgrove@xilinx.com>
40754 * common/config/microblaze/microblaze-common.c
40755 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40756 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40758 * config/microblaze/microblaze.c (microblaze_must_save_register)
40759 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40761 (microblaze_eh_return): New function.
40762 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40763 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40764 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40765 * config/microblaze/microblaze.md (eh_return): New pattern.
40767 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40769 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40770 GCC_DIAGNOSTIC_STRINGIFY): Define.
40772 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40774 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40776 * config/arm/arm.md (<mcrr>): New.
40778 * config/arm/arm.c (arm_arch5te): New.
40779 (arm_option_override): Set arm_arch5te.
40780 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40782 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40783 (arm_mcrr_qualifiers): ... this. New.
40784 (MRRC_QUALIFIERS): Define to...
40785 (arm_mrrc_qualifiers): ... this. New.
40786 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40788 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40789 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40790 (MRRCI, mrrc, MRRC): New.
40791 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40792 VUNSPEC_MRRC2): New.
40794 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40796 * config/arm/arm.md (<mcr>): New.
40798 * config/arm/arm.c (arm_coproc_builtin_available): Add
40799 support for mcr, mrc, mcr2 and mrc2.
40800 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40801 (arm_mcr_qualifiers): ... this. New.
40802 (MRC_QUALIFIERS): Define to ...
40803 (arm_mrc_qualifiers): ... this. New.
40804 (MCR_QUALIFIERS): Define to ...
40805 (arm_mcr_qualifiers): ... this. New.
40806 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40808 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40809 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40810 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40811 VUNSPEC_MRC2): New.
40813 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40815 * config/arm/arm.md (*ldc): New.
40819 * config/arm/arm.c (arm_coproc_builtin_available): Add
40820 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40821 (arm_coproc_ldc_stc_legitimate_address): New.
40822 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40823 'qualifier_const_pointer'.
40824 (LDC_QUALIFIERS): Define to...
40825 (arm_ldc_qualifiers): ... this. New.
40826 (STC_QUALIFIERS): Define to...
40827 (arm_stc_qualifiers): ... this. New.
40828 * config/arm/arm-protos.h
40829 (arm_coproc_ldc_stc_legitimate_address): New.
40830 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40831 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40832 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40833 stc2, stcl, stc2l): New.
40834 * config/arm/constraints.md (Uz): New.
40835 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40836 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40837 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40838 VUNSPEC_STC2L): New.
40840 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40842 * config/arm/arm.md (<cdp>): New.
40843 * config/arm/arm.c (neon_const_bounds): Rename this ...
40844 (arm_const_bounds): ... this.
40845 (arm_coproc_builtin_available): New.
40846 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40847 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40848 (CDP_QUALIFIERS): Define to...
40849 (arm_cdp_qualifiers): ... this. New.
40851 (arm_expand_builtin_args): Add case for 6 arguments.
40852 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40853 (arm_const_bounds): ... this.
40854 (arm_coproc_builtin_available): New.
40855 * config/arm/arm_acle.h (__arm_cdp): New.
40857 * config/arm/arm_acle_builtins.def (cdp): New.
40859 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40860 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40861 'arm_const_bounds'.
40862 * config/arm/types.md (coproc): New.
40863 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40864 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40865 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40866 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40868 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40870 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40871 (UBINOP_QUALIFIERS): New.
40873 (acle_builtin_data): New. Change comment.
40874 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40875 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40876 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40877 arm_acle_builtins.def.
40878 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40879 (arm_init_acle_builtins): New.
40880 (CRC32_BUILTIN): Remove.
40881 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40882 crc32cb, crc32ch and crc32cw.
40883 (arm_init_crc32_builtins): Remove.
40884 (arm_init_builtins): Use arm_init_acle_builtins rather
40885 than arm_init_crc32_builtins.
40886 (arm_expand_acle_builtin): New.
40887 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40888 * config/arm/arm_acle_builtins.def: New.
40890 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40892 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40893 (arm_builtin_datum): ... this.
40894 (arm_init_neon_builtin): Rename to ...
40895 (arm_init_builtin): ... this. Add a new parameters PREFIX
40896 and USE_SIG_IN_NAME.
40897 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40898 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40899 'arm_builtin_datum'.
40900 (arm_init_vfp_builtins): Likewise.
40901 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40902 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40903 (arm_expand_neon_args): Rename to ...
40904 (arm_expand_builtin_args): ... this. Rename builtin_arg
40905 enum values and differentiate between ARG_BUILTIN_MEMORY
40906 and ARG_BUILTIN_NEON_MEMORY.
40907 (arm_expand_neon_builtin_1): Rename to ...
40908 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40909 values, arm_expand_builtin_args and add bool parameter NEON.
40910 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40911 (arm_expand_vfp_builtin): Likewise.
40912 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40914 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40916 PR middle-end/77484
40917 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40918 * predict.c (tree_estimate_probability_bb): Reverse direction of
40919 polymorphic call predictor.
40921 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40923 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40924 (determine_pass_name_match): ...this new function and...
40925 (should_skip_pass_p): ...this new function.
40927 2017-01-06 Nathan Sidwell <nathan@acm.org>
40929 * ipa-visibility.c (function_and_variable_visibility): Reformat
40930 comments and long lines. Remove extrneous if.
40931 * symtab.c (symtab_node::make_decl_local): Fix code format.
40932 (symtab_node::set_section_for_node): Fix comment typo.
40934 2017-01-06 Martin Liska <mliska@suse.cz>
40937 * lra-constraints.c: Rename invariant to lra_invariant.
40938 * predict.c (set_even_probabilities): Initialize e to NULL.
40940 2017-01-05 Martin Sebor <msebor@redhat.com>
40942 PR tree-optimization/78910
40943 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40944 (format_integer): Correct off-by-one error in the handling
40945 of precision with negative numbers in signed conversions..
40947 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40949 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40951 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40953 PR tree-optimization/71016
40954 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40955 factor_out_conditional_conversion. Formatting fix.
40956 (factor_out_conditional_conversion): Add cond_stmt argument.
40957 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40958 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40961 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40963 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40964 read-rtl-function.o, and selftest-rtl.o.
40965 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40966 (selftest::aarch64_test_loading_full_dump): New function.
40967 (selftest::aarch64_run_selftests): New function.
40968 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40969 selftest::aarch64_run_selftests.
40970 * config/i386/i386.c
40971 (selftest::ix86_test_loading_dump_fragment_1): New function.
40972 (selftest::ix86_test_loading_call_insn): New function.
40973 (selftest::ix86_test_loading_full_dump): New function.
40974 (selftest::ix86_test_loading_unspec): New function.
40975 (selftest::ix86_run_selftests): Call the new functions.
40976 * emit-rtl.c (maybe_set_max_label_num): New function.
40977 * emit-rtl.h (maybe_set_max_label_num): New decl.
40978 * function.c (instantiate_decls): Guard call to
40979 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40980 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40982 * gensupport.c (gen_reader::gen_reader): Pass "false"
40983 for new "compact" param of rtx_reader.
40984 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40985 rather than an empty string for NULL strings.
40986 * read-md.c: Potentially include config.h rather than bconfig.h.
40987 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40988 (have_error): New global, copied from errors.c.
40989 (md_reader::read_name): Rename to...
40990 (md_reader::read_name_1): ...this, adding "out_loc" param,
40991 and converting "missing name or number" to returning false, rather
40993 (md_reader::read_name): Reimplement in terms of read_name_1.
40994 (md_reader::read_name_or_nil): New function.
40995 (md_reader::read_string): Handle "(nil)" by returning NULL.
40996 (md_reader::md_reader): Add new param "compact".
40997 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40998 (md_reader::read_file): New method.
40999 * read-md.h (md_reader::md_reader): Add new param "compact".
41000 (md_reader::read_file): New method.
41001 (md_reader::is_compact): New accessor.
41002 (md_reader::read_name): Convert return type from void to file_location.
41003 (md_reader::read_name_or_nil): New decl.
41004 (md_reader::read_name_1): New decl.
41005 (md_reader::m_compact): New field.
41006 (noop_reader::noop_reader): Pass "false" for new "compact" param
41008 (rtx_reader::rtx_reader): Add new "compact" param.
41009 (rtx_reader::read_rtx_operand): Make virtual and convert return
41010 type from void to rtx.
41011 (rtx_reader::read_until): New decl.
41012 (rtx_reader::handle_any_trailing_information): New virtual function.
41013 (rtx_reader::postprocess): New virtual function.
41014 (rtx_reader::finalize_string): New virtual function.
41015 (rtx_reader::m_in_call_function_usage): New field.
41016 (rtx_reader::m_reuse_rtx_by_id): New field.
41017 * read-rtl-function.c: New file.
41018 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41019 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41020 (selftest::verify_three_block_rtl_cfg): New decl.
41021 * read-rtl-function.h: New file.
41022 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41023 For host, include function.h, memmodel.h, and emit-rtl.h.
41024 (one_time_initialization): New function.
41025 (struct compact_insn_name): New struct.
41026 (compact_insn_names): New array.
41027 (find_code): Handle insn codes in compact dumps.
41028 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41029 (bind_subst_iter_and_attr): Likewise.
41030 (add_condition_to_string): Likewise.
41031 (add_condition_to_rtx): Likewise.
41032 (apply_attribute_uses): Likewise.
41033 (add_current_iterators): Likewise.
41034 (apply_iterators): Likewise.
41035 (initialize_iterators): Guard usage of apply_subst_iterator with
41036 #ifdef GENERATOR_FILE.
41037 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41038 (md_reader::read_mapping): Likewise.
41039 (add_define_attr_for_define_subst): Likewise.
41040 (add_define_subst_attr): Likewise.
41041 (read_subst_mapping): Likewise.
41042 (check_code_iterator): Likewise.
41043 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41045 (one_time_initialization): New function.
41046 (rtx_reader::read_until): New method.
41047 (read_flags): New function.
41048 (parse_reg_note_name): New function.
41049 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41050 Handle reuse_rtx ids.
41051 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41052 Add parsing support for RTL dumps, mirroring the special-cases in
41053 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41054 values, and calling handle_any_trailing_information.
41055 (rtx_reader::read_rtx_operand): Convert return type from void
41056 to rtx, returning return_rtx. Handle case 'e'. Call
41057 finalize_string on XSTR and XTMPL fields.
41058 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41059 "(nil)" values were omitted. Call the postprocess vfunc on the
41061 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41062 class ctor. Initialize m_in_call_function_usage. Call
41063 one_time_initialization.
41064 * rtl-tests.c (selftest::test_uncond_jump): Call
41065 set_new_first_and_last_insn.
41066 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41067 * selftest-rtl.c: New file.
41068 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41069 (selftest::get_insn_by_uid): New decl.
41070 * selftest-run-tests.c (selftest::run_tests): Call
41071 read_rtl_function_c_tests.
41072 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41073 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41076 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41078 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41079 operands in a special way. Assert that pos+len <= mode precision.
41081 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41083 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41084 3 argument Alias with unlimited for the negative form.
41085 (fno-vect-cost-model): Removed.
41087 2017-01-05 Martin Liska <mliska@suse.cz>
41089 * hsa-gen.c (gen_hsa_divmod): New function.
41090 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41092 2017-01-05 Martin Liska <mliska@suse.cz>
41095 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41098 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41100 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41101 small constant length operands.
41103 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41105 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41106 between loop iterations.
41108 2017-01-05 Martin Liska <mliska@suse.cz>
41111 * gimplify.c (gimplify_decl_expr): Compare to
41112 asan_poisoned_variables instread of checking flags.
41113 (gimplify_target_expr): Likewise.
41114 (gimplify_expr): Likewise.
41115 (gimplify_function_tree): Conditionally initialize
41116 asan_poisoned_variables.
41118 2017-01-04 Jeff Law <law@redhat.com>
41120 PR tree-optimizatin/78812
41121 * rtl.h (contains_mem_rtx_p): Prototype.
41122 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41123 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41124 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41125 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41126 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41128 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41130 * input.c (assert_char_at_range): Default-initialize actual_range.
41132 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41134 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41135 to match the caller.
41137 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41139 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41140 insns after final jump in test to emit dummy move.
41142 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41144 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41145 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41147 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41149 * multiple_target.c (create_dispatcher_calls): Init e_next.
41150 * tree-ssa-loop-split.c (split_loop): Init border.
41151 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41154 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41159 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41160 (altivec_register_operand): Do not return true if the operand
41161 contains a SUBREG mixing SImode and SFmode.
41162 (vsx_register_operand): Likewise.
41163 (vsx_reg_sfsubreg_ok): New predicate.
41164 (vfloat_operand): Do not return true if the operand contains a
41165 SUBREG mixing SImode and SFmode.
41166 (vint_operand): Likewise.
41167 (vlogical_operand): Likewise.
41168 (gpc_reg_operand): Likewise.
41169 (int_reg_operand): Likewise.
41170 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41171 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41172 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41174 (rs6000_emit_move_si_sf_subreg): New helper function.
41175 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41176 fixup SUBREGs involving SImode and SFmode.
41177 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41178 numbers for the new peephole2 optimization.
41179 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41180 the GLIBC math library that do AND/IOR/XOR operations on single
41181 precision floating point.
41182 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41183 target macros to say whether we need to avoid SUBREGs mixing
41185 (TARGET_ALLOW_SF_SUBREG): Likewise.
41186 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41187 (UNSPEC_SI_FROM_SF): Likewise.
41188 (iorxor): Change spacing.
41189 (and_ior_xor): New iterator for AND, IOR, and XOR.
41190 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41191 (movdi_from_sf_zero_ext): Likewise.
41192 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41193 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41194 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41195 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41196 (fms<mode>4): Likewise.
41197 (fnma<mode>4): Likewise.
41198 (fnms<mode>4): Likewise.
41199 (nfma<mode>4): Likewise.
41200 (nfms<mode>4): Likewise.
41202 2017-01-04 Marek Polacek <polacek@redhat.com>
41205 * doc/invoke.texi: Document -Wpointer-compare.
41207 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41209 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41212 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41213 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41216 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41219 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41220 documentation of the powerpc_popcntb_ok attribute.
41221 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41222 code to issue warning messages if a requested CPU configuration is
41223 not supported by the binary (assembler and loader) toolchain.
41224 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41225 made to define a built-in function that has been disabled.
41226 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41227 made to define a built-in function that has been disabled.
41228 (altivec_init_builtins): Add comment explaining why definition
41229 of the DST built-in functions is not preceded by an assertion
41230 check. Add assertions to prevent ICE if attempts are made to
41231 define an altivec predicate or an abs* built-in function that has
41233 (htm_init_builtins): Add comment explaining why definition of the
41234 htm built-in functions is not preceded by an assertion check.
41236 2017-01-04 Jeff Law <law@redhat.com>
41238 PR tree-optimizatin/67955
41239 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41240 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41241 the points-to solution does not include pt_null. Use DECL_PT_UID
41244 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41246 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41247 Use gen_int_mode instead of gen_lopwart for const_int operands.
41249 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41251 PR tree-optimization/71563
41252 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41253 out of range value - has low bits known to be zero.
41255 2017-01-04 Alan Modra <amodra@gmail.com>
41257 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41258 * configure: Regenerate.
41259 * config.in: Regenerate.
41261 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41264 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41265 a substring of the message, but strcmp with the whole message. Ifdef
41266 ENABLE_NLS, translate the message first using dgettext.
41268 2017-01-03 Jeff Law <law@redhat.com>
41270 PR tree-optimizatin/78856
41271 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41272 (mark_threaded_blocks): Remove code to truncate thread paths that
41273 cross multiple loop headers. Instead invalidate the cached loop
41274 iteration information and handle case of a thread path walking
41275 into an irreducible region.
41277 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41280 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41281 assertions. Add support for doing the signbit if the IEEE 128-bit
41282 floating point value is in a GPR.
41283 * config/rs6000/rs6000.md (Fsignbit): Delete.
41284 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41285 Update the length attribute if the value is in a GPR.
41286 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41287 the sign or zero extension instruction, since the value is always 0/1.
41288 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41291 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41292 extracting SImode to a GPR register so that we can generate a
41293 store, limit the vector to be in a traditional Altivec register
41294 for the vextuwrx instruction.
41296 2017-01-03 Ian Lance Taylor <iant@google.com>
41298 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41300 2017-01-03 Martin Sebor <msebor@redhat.com>
41302 PR tree-optimization/78696
41303 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41304 precision. Use MPFR for %f for greater fidelity. Correct handling
41306 (pass_sprintf_length::compute_format_length): Set width and precision
41307 specified by asrerisk to void_node for vararg functions.
41308 (try_substitute_return_value): Adjust dump output.
41310 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41312 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41314 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41316 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41317 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41318 -mlra/-mno-lra was passed to the compiler.
41320 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41322 PR rtl-optimization/65618
41323 * emit-rtl.c (try_split): Move initialization of "before" and
41324 "after" to just before the call to emit_insn_after_setloc.
41326 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41328 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41330 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41332 * dwarf2out.c (gen_enumeration_type_die): When
41333 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41335 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41337 PR tree-optimization/78965
41338 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41339 Change first argument from const call_info & to call_info &. For %n
41340 set info.nowrite to false.
41342 PR middle-end/78901
41343 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41344 possibly throwing calls.
41346 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41347 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41348 and fns handling, rather than in a separate case SSA_NAME.
41350 2017-01-02 Jeff Law <law@redhat.com>
41352 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41353 fixes for first_period and second_period variables.
41355 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41358 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41359 (*insvqi_1): New insn pattern.
41360 (*insvqi_1_mem_rex64): Ditto.
41361 (*insvqi_2): Ditto.
41362 (*insvqi_3): Rename from *insvqi.
41364 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41366 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41368 * doc/cfg.texi (Edges): Remove reference to Java.
41369 (Maintaining the CFG): Ditto.
41371 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41373 PR middle-end/77674
41374 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41375 transparent aliases.
41377 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41379 PR middle-end/77484
41380 * predict.def (PRED_CALL): Update hitrate.
41381 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41382 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41383 into direct/indirect/polymorphic variants.
41385 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41387 Update copyright years.
41389 * gcc.c (process_command): Update copyright notice dates.
41390 * gcov-dump.c (print_version): Ditto.
41391 * gcov.c (print_version): Ditto.
41392 * gcov-tool.c (print_version): Ditto.
41393 * gengtype.c (create_file): Ditto.
41394 * doc/cpp.texi: Bump @copying's copyright year.
41395 * doc/cppinternals.texi: Ditto.
41396 * doc/gcc.texi: Ditto.
41397 * doc/gccint.texi: Ditto.
41398 * doc/gcov.texi: Ditto.
41399 * doc/install.texi: Ditto.
41400 * doc/invoke.texi: Ditto.
41402 Copyright (C) 2017 Free Software Foundation, Inc.
41404 Copying and distribution of this file, with or without modification,
41405 are permitted in any medium without royalty provided the copyright
41406 notice and this notice are preserved.