1 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3 * common/config/arm/arm-common.c (compare_opt_names): Add function
4 comment. Use strcmp instead of manual loop.
6 2017-12-21 Martin Liska <mliska@suse.cz>
9 * gcov-dump.c (dump_gcov_file): Do not read info about
10 support_unexecuted_blocks for gcda files.
12 2017-12-21 Jakub Jelinek <jakub@redhat.com>
14 PR rtl-optimization/82973
15 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
16 (valid_for_const_vector_p): ... this.
17 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
18 (valid_for_const_vector_p): ... this. Adjust function comment.
19 (gen_vec_duplicate): Adjust caller.
20 * optabs.c (expand_vector_broadcast): Likewise.
21 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
22 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
24 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
25 instead of GET_CODE == CONST_FIXED.
26 (simplify_subreg): Use CONST_FIXED_P macro instead of
27 GET_CODE == CONST_FIXED.
30 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
31 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
32 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
33 (ix86_option_override_internal): Adjust for
34 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
35 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
36 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
37 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
38 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
39 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
40 def_builtin2 for CET builtins.
41 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
42 up formatting in IX86_BUILTIN_RDPID code.
43 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
44 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
45 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
46 ix86_isa_flags2 to ix86_isa_flags.
47 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
49 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
50 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
51 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
53 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
54 Or in OPTION_MASK_ISA_AVX512F_SET.
55 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
56 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
57 (ix86_handle_option): Adjust for
58 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
59 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
61 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
62 Alan Hayward <alan.hayward@arm.com>
63 David Sherwood <david.sherwood@arm.com>
65 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
66 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
67 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
68 Track polynomial offsets.
70 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
71 Alan Hayward <alan.hayward@arm.com>
72 David Sherwood <david.sherwood@arm.com>
74 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
75 Treat drb->init as a poly_int. Fail if its misalignment wrt
76 vector_alignment isn't known.
78 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
79 Alan Hayward <alan.hayward@arm.com>
80 David Sherwood <david.sherwood@arm.com>
82 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
84 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
85 (vect_loop_versioning): Take the loop versioning threshold as a
87 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
88 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
90 (vect_analyze_loop_2): Compute the loop versioning threshold
91 whenever loop versioning is needed, and store it in the new
92 field rather than combining it with the cost model threshold.
93 (vect_transform_loop): Update call to vect_loop_versioning.
94 Try to combine the loop versioning and cost thresholds here.
96 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
97 Alan Hayward <alan.hayward@arm.com>
98 David Sherwood <david.sherwood@arm.com>
100 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
101 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
102 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
103 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
104 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
105 (group_compare_offset): Update accordingly.
106 (split_small_address_groups_p): Likewise.
107 (record_use): Take addr_offset as a poly_uint64 rather than
108 an unsigned HOST_WIDE_INT.
109 (strip_offset): Return the offset as a poly_uint64 rather than
110 an unsigned HOST_WIDE_INT.
111 (record_group_use, split_address_groups): Track polynomial offsets.
112 (add_iv_candidate_for_use): Likewise.
113 (addr_offset_valid_p): Take the offset as a poly_int64 rather
114 than a HOST_WIDE_INT.
115 (strip_offset_1): Return the offset as a poly_int64 rather than
118 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
119 Alan Hayward <alan.hayward@arm.com>
120 David Sherwood <david.sherwood@arm.com>
122 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
123 rather than a HOST_WIDE_INT.
124 * tree.c (get_binfo_at_offset): Likewise.
126 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
127 Alan Hayward <alan.hayward@arm.com>
128 David Sherwood <david.sherwood@arm.com>
130 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
131 rather than a HOST_WIDE_INT.
132 * tree-sra.c (build_ref_for_offset): Likewise.
134 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
135 Alan Hayward <alan.hayward@arm.com>
136 David Sherwood <david.sherwood@arm.com>
138 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
140 * tree.c (mem_ref_offset): Likewise.
141 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
142 * builtins.c (get_object_alignment_2): Likewise.
143 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
144 * gimple-fold.c (get_base_constructor): Likewise.
145 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
146 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
148 * ipa-polymorphic-call.c
149 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
150 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
151 (get_ancestor_addr_info): Likewise.
152 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
153 * match.pd: Likewise.
154 * tree-data-ref.c (dr_analyze_innermost): Likewise.
155 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
156 * tree-eh.c (tree_could_trap_p): Likewise.
157 * tree-object-size.c (addr_object_size): Likewise.
158 * tree-ssa-address.c (copy_ref_info): Likewise.
159 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
160 (indirect_refs_may_alias_p): Likewise.
161 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
162 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
163 (non_rewritable_mem_ref_base): Likewise.
164 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
165 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
166 * varasm.c (decode_addr_const): Likewise.
168 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
169 Alan Hayward <alan.hayward@arm.com>
170 David Sherwood <david.sherwood@arm.com>
172 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
173 offsets for MEM_REFs.
175 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
176 Alan Hayward <alan.hayward@arm.com>
177 David Sherwood <david.sherwood@arm.com>
179 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
180 as a poly_uint64 rather than an unsigned int.
181 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
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 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
189 (decode_addr_const): Update accordingly.
191 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
192 Alan Hayward <alan.hayward@arm.com>
193 David Sherwood <david.sherwood@arm.com>
195 * tree.h (bit_field_size, bit_field_offset): New functions.
196 * hsa-gen.c (gen_hsa_addr): Use them.
197 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
198 (simplify_vector_constructor): Likewise.
199 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
200 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
201 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
202 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
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 * expr.h (emit_group_load, emit_group_load_into_temps)
210 (emit_group_store): Take the size as a poly_int64 rather than an int.
211 * expr.c (emit_group_load_1, emit_group_load): Likewise.
212 (emit_group_load_into_temp, emit_group_store): Likewise.
214 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
215 Alan Hayward <alan.hayward@arm.com>
216 David Sherwood <david.sherwood@arm.com>
218 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
219 an unsigned int to a poly_uint64.
220 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
221 sizes as poly_uint64s rather than unsigned ints.
222 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
225 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
226 Alan Hayward <alan.hayward@arm.com>
227 David Sherwood <david.sherwood@arm.com>
229 * lra-constraints.c (emit_inc): Change inc_amount from an int
232 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
233 Alan Hayward <alan.hayward@arm.com>
234 David Sherwood <david.sherwood@arm.com>
236 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
238 (add_stack_var, stack_var_cmp, partition_stack_vars)
239 (dump_stack_var_partition): Update accordingly.
240 (alloc_stack_frame_space): Take the size as a poly_int64 rather
241 than a HOST_WIDE_INT.
242 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
243 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
244 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
245 rather than a HOST_WIDE_INT.
247 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
248 Alan Hayward <alan.hayward@arm.com>
249 David Sherwood <david.sherwood@arm.com>
251 * target.def (return_pops_args): Treat both the input and output
252 sizes as poly_int64s rather than HOST_WIDE_INTS.
253 * targhooks.h (default_return_pops_args): Update accordingly.
254 * targhooks.c (default_return_pops_args): Likewise.
255 * doc/tm.texi: Regenerate.
256 * emit-rtl.h (incoming_args): Change pops_args, size and
257 outgoing_args_size from int to poly_int64_pod.
258 * function.h (expr_status): Change x_pending_stack_adjust and
259 x_stack_pointer_delta from int to poly_int64.
260 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
261 (ARGS_SIZE_RTX): Update accordingly.
262 * calls.c (highest_outgoing_arg_in_use): Change from int to
264 (stack_usage_watermark, stored_args_watermark): New variables.
265 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
266 (emit_call_1): Change the stack_size and rounded_stack_size
267 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
269 (save_fixed_argument_area): Check stack_usage_watermark.
270 (initialize_argument_information): Change old_pending_adj from
271 a HOST_WIDE_INT * to a poly_int64_pod *.
272 (compute_argument_block_size): Return the size as a poly_int64
274 (finalize_must_preallocate): Track polynomial argument sizes.
275 (compute_argument_addresses): Likewise.
276 (internal_arg_pointer_based_exp): Track polynomial offsets.
277 (mem_overlaps_already_clobbered_arg_p): Rename to...
278 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
279 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
280 Check stored_args_used_watermark.
281 (load_register_parameters): Update accordingly.
282 (check_sibcall_argument_overlap_1): Likewise.
283 (combine_pending_stack_adjustment_and_call): Take the unadjusted
284 args size as a poly_int64 rather than an int. Return a bool
285 indicating whether the optimization was possible and return
286 the new adjustment by reference.
287 (check_sibcall_argument_overlap): Track polynomail argument sizes.
288 Update stored_args_watermark.
289 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
290 (expand_call): Likewise. Maintain stack_usage_watermark and
291 stored_args_watermark. Update calls to
292 combine_pending_stack_adjustment_and_call.
293 (emit_library_call_value_1): Handle polynomial argument sizes.
294 Call stack_region_maybe_used_p and mark_stack_region_used.
295 Maintain stack_usage_watermark.
296 (store_one_arg): Likewise. Update call to
297 mem_overlaps_already_clobbered_arg_p.
298 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
300 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
301 * config/microblaze/microblaze.c (microblaze_function_prologue):
303 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
304 TARGET_RETURN_POPS_ARGS interface.
305 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
307 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
308 * config/i386/i386.c (ix86_return_pops_args): Update for new
309 TARGET_RETURN_POPS_ARGS interface.
310 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
311 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
312 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
313 TARGET_RETURN_POPS_ARGS interface.
314 * config/vax/vax.c (vax_return_pops_args): Likewise.
315 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
316 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
317 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
318 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
319 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
320 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
321 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
322 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
323 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
324 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
325 and x_stack_pointer_delta from int to poly_int64.
326 * dojump.c (do_pending_stack_adjust): Update accordingly.
327 * explow.c (allocate_dynamic_stack_space): Handle polynomial
328 stack_pointer_deltas.
329 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
330 (pad_to_arg_alignment): Track polynomial offsets.
331 (assign_parm_find_stack_rtl): Likewise.
332 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
333 * toplev.c (output_stack_usage): Update reference to
334 current_function_pushed_stack_size.
336 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
337 Alan Hayward <alan.hayward@arm.com>
338 David Sherwood <david.sherwood@arm.com>
340 * function.c (in_arg_offset, var_offset, dynamic_offset)
341 (out_arg_offset, cfa_offset): Change from int to poly_int64.
342 (instantiate_new_reg): Return the new offset as a poly_int64_pod
343 rather than a HOST_WIDE_INT.
344 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
345 (instantiate_virtual_regs_in_insn): Likewise.
347 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
348 Alan Hayward <alan.hayward@arm.com>
349 David Sherwood <david.sherwood@arm.com>
351 * rtl.h (get_args_size, add_args_size_note): New functions.
352 (find_args_size_adjust): Return a poly_int64 rather than a
354 (fixup_args_size_notes): Likewise. Make the same change to the
355 end_args_size parameter.
356 * rtlanal.c (get_args_size, add_args_size_note): New functions.
357 * builtins.c (expand_builtin_trap): Use add_args_size_note.
358 * calls.c (emit_call_1): Likewise.
359 * explow.c (adjust_stack_1): Likewise.
360 * cfgcleanup.c (old_insns_match_p): Update use of
361 find_args_size_adjust.
362 * combine.c (distribute_notes): Track polynomial arg sizes.
363 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
364 end_true_args_size, beg_delay_args_size and end_delay_args_size
365 from HOST_WIDE_INT to poly_int64.
366 (add_cfi_args_size): Take the args_size as a poly_int64 rather
367 than a HOST_WIDE_INT.
368 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
369 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
370 polynomial arg sizes.
371 * emit-rtl.c (try_split): Use get_args_size.
372 * recog.c (peep2_attempt): Likewise.
373 * reload1.c (reload_as_needed): Likewise.
374 * expr.c (find_args_size_adjust): Return the adjustment as a
375 poly_int64 rather than a HOST_WIDE_INT.
376 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
377 to a poly_int64 and change the return type in the same way.
378 (emit_single_push_insn): Track polynomial arg sizes.
380 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
381 Alan Hayward <alan.hayward@arm.com>
382 David Sherwood <david.sherwood@arm.com>
384 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
385 from HOST_WIDE_INT to poly_int64.
386 * expr.c (push_block, emit_push_insn): Likewise.
388 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
389 Alan Hayward <alan.hayward@arm.com>
390 David Sherwood <david.sherwood@arm.com>
392 * function.h (frame_space): Change start and length from HOST_WIDE_INT
394 (get_frame_size): Return the size as a poly_int64 rather than a
396 (frame_offset_overflow): Take the offset as a poly_int64 rather
397 than a HOST_WIDE_INT.
398 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
399 (assign_stack_temp): Likewise for the size.
400 * function.c (get_frame_size): Return a poly_int64 rather than
402 (frame_offset_overflow): Take the offset as a poly_int64 rather
403 than a HOST_WIDE_INT.
404 (try_fit_stack_local): Take the start, length and size as poly_int64s
405 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
406 rather than a HOST_WIDE_INT.
407 (add_frame_space): Take the start and end as poly_int64s rather than
409 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
410 (assign_stack_temp): Likewise for the size.
411 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
413 (find_temp_slot_from_address): Handle polynomial offsets.
414 (combine_temp_slots): Likewise.
415 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
417 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
418 poly_int64 rather than a HOST_WIDE_INT.
419 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
420 than a HOST_WIDE_INT.
421 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
422 polynomial frame offsets.
423 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
424 as a poly_int64 rather than an int.
425 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
426 * config/v850/v850-protos.h (compute_frame_size): Likewise.
427 * config/v850/v850.c (compute_frame_size): Likewise.
428 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
429 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
430 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
431 * config/pa/pa.c (pa_compute_frame_size): Likewise.
432 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
433 rather than a HOST_WIDE_INT.
434 * explow.c (get_dynamic_stack_base): Likewise.
435 * final.c (final_start_function): Use the constant lower bound
436 of the frame size for -Wframe-larger-than.
437 * ira.c (do_reload): Adjust for new get_frame_size return type.
438 * lra.c (lra): Likewise.
439 * reload1.c (reload): Likewise.
440 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
441 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
442 * rtlanal.c (get_initial_register_offset): Return the offset as
443 a poly_int64 rather than a HOST_WIDE_INT.
445 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
446 Alan Hayward <alan.hayward@arm.com>
447 David Sherwood <david.sherwood@arm.com>
449 * reload1.c (elim_table): Change initial_offset, offset and
450 previous_offset from HOST_WIDE_INT to poly_int64_pod.
451 (offsets_at): Change the target array's element type from
452 HOST_WIDE_INT to poly_int64_pod.
453 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
454 (elimination_costs_in_insn, update_eliminable_offsets)
455 (verify_initial_elim_offsets, set_offsets_for_label)
456 (init_eliminable_invariants): Update after above changes.
458 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
459 Alan Hayward <alan.hayward@arm.com>
460 David Sherwood <david.sherwood@arm.com>
462 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
463 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
464 (decomposition): Change start and end from HOST_WIDE_INT
466 (decompose, immune_p): Update accordingly.
467 (find_inc_amount): Return a poly_int64 rather than an int.
468 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
471 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
472 Alan Hayward <alan.hayward@arm.com>
473 David Sherwood <david.sherwood@arm.com>
475 * tree.h (get_inner_reference): Return the bitsize and bitpos
476 as poly_int64_pods rather than HOST_WIDE_INT.
477 * fold-const.h (ptr_difference_const): Return the pointer difference
478 as a poly_int64_pod rather than a HOST_WIDE_INT.
479 * expr.c (get_inner_reference): Return the bitsize and bitpos
480 as poly_int64_pods rather than HOST_WIDE_INT.
481 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
483 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
484 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
485 (optimize_bit_field_compare): Update call to get_inner_reference.
486 (decode_field_reference): Likewise.
487 (fold_unary_loc): Track polynomial offsets and sizes.
488 (split_address_to_core_and_offset): Return the bitpos as a
489 poly_int64_pod rather than a HOST_WIDE_INT.
490 (ptr_difference_const): Likewise for the pointer difference.
491 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
492 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
493 * dbxout.c (dbxout_expand_expr): Likewise.
494 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
495 (loc_list_from_tree_1, fortran_common): Likewise.
496 * gimple-laddress.c (pass_laddress::execute): Likewise.
497 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
498 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
499 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
500 * tree-affine.c (tree_to_aff_combination): Likewise.
501 (get_inner_reference_aff): Likewise.
502 * tree-data-ref.c (split_constant_offset_1): Likewise.
503 (dr_analyze_innermost): Likewise.
504 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
505 * tree-sra.c (ipa_sra_check_caller): Likewise.
506 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
507 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
508 (instrument_bool_enum_load, instrument_object_size): Likewise.
509 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
510 to get_inner_reference.
511 * hsa-gen.c (gen_hsa_addr): Likewise.
512 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
513 * tsan.c (instrument_expr): Likewise.
514 * match.pd: Update call to ptr_difference_const.
516 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
517 Alan Hayward <alan.hayward@arm.com>
518 David Sherwood <david.sherwood@arm.com>
520 * fold-const.c (fold_comparison): Track sizes and offsets as
521 poly_int64s rather than HOST_WIDE_INTs when folding address
524 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
525 Alan Hayward <alan.hayward@arm.com>
526 David Sherwood <david.sherwood@arm.com>
528 * expr.h (get_bit_range): Return the bitstart and bitend as
529 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
530 as a poly_int64 rather than a HOST_WIDE_INT.
531 * expr.c (get_bit_range): Likewise.
532 (expand_assignment): Update call accordingly.
533 * fold-const.c (optimize_bit_field_compare): Likewise.
535 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
536 Alan Hayward <alan.hayward@arm.com>
537 David Sherwood <david.sherwood@arm.com>
539 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
540 (aarch64_legitimate_address_p): Use it instead of an rtx code,
541 as an optional final parameter.
542 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
543 (aarch64_legitimate_address_p): Likewise.
544 (aarch64_print_address_internal): Take an aarch64_addr_query_type
545 instead of an rtx code.
546 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
547 (aarch64_legitimate_address_hook_p): Likewise.
548 (aarch64_print_ldpstp_address): Likewise.
549 (aarch64_print_operand_address): Likewise.
550 (aarch64_address_cost): Likewise.
551 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
552 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
553 (aarch64_mem_pair_lanes_operand): Likewise.
555 2017-12-20 Richard Biener <rguenther@suse.de>
557 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
558 update_stmt_if_modified.
560 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
562 PR tree-optimization/83491
563 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
564 before walking uses. Improve coding style and comments.
566 2017-12-20 Tom de Vries <tom@codesourcery.com>
568 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
570 2017-12-20 Jakub Jelinek <jakub@redhat.com>
573 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
575 (pass_ipa_free_fn_summary): Add small_p private data member,
576 initialize to false in the ctor.
577 (pass_ipa_free_fn_summary::clone,
578 pass_ipa_free_fn_summary::set_pass_param,
579 pass_ipa_free_fn_summary::gate): New methods.
580 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
581 | TODO_dump_symtab if small_p.
582 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
583 entry and add another instance of the pass with false parm after
585 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
586 ipa_free_fn_summary here.
588 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
590 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
592 2017-12-20 Martin Sebor <msebor@redhat.com>
595 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
596 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
597 (expand_builtin_strncmp): Same.
599 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
602 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
603 allowed debug stmts before labels.
604 (expand_gimple_basic_block): Likewise.
605 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
606 * gimple-iterator.h (gsi_after_labels): Likewise.
607 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
608 rename reused variable, and simplify using gsi_move_before.
609 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
610 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
611 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
612 (gimple_verify_flow_info, gimple_block_label): Likewise.
613 (make_blocks): Move debug markers after adjacent labels.
614 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
615 allowed debug insns outside blocks.
616 * df-scan.c (df_insn_delete): Likewise.
617 * lra-constraints.c (update_ebb_live_info): Likewise.
618 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
619 (vt_initialize, delete_vta_debug_insns): Likewise.
620 (reemit_marker_as_note): Drop BB parm. Adjust callers.
622 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
623 Alan Hayward <alan.hayward@arm.com>
624 David Sherwood <david.sherwood@arm.com>
626 * poly-int-types.h (round_down_to_byte_boundary): New macro.
627 (round_up_to_byte_boundary): Likewise.
628 * expr.h (get_bit_range): Add temporary shim.
629 * gimple-ssa-store-merging.c (store_operand_info): Change the
630 bitsize, bitpos, bitregion_start and bitregion_end fields from
631 unsigned HOST_WIDE_INT to poly_uint64.
632 (merged_store_group): Likewise load_align_base.
633 (compatible_load_p, compatible_load_p): Update accordingly.
634 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
635 (split_group, imm_store_chain_info::output_merged_store): Likewise.
636 (mem_valid_for_store_merging): Return the bitsize, bitpos,
637 bitregion_start and bitregion_end as poly_uint64s rather than
638 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
639 (handled_load): Take the bitsize, bitpos,
640 bitregion_start and bitregion_end as poly_uint64s rather than
641 unsigned HOST_WIDE_INTs.
642 (pass_store_merging::process_store): Update call to
643 mem_valid_for_store_merging.
645 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
646 Alan Hayward <alan.hayward@arm.com>
647 David Sherwood <david.sherwood@arm.com>
649 * builtins.c (get_object_alignment_2): Track polynomial offsets
650 and sizes. Update the alignment handling.
652 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
653 Alan Hayward <alan.hayward@arm.com>
654 David Sherwood <david.sherwood@arm.com>
656 * tree.h (get_inner_reference): Add a version that returns the
657 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
658 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
659 the case in which bitpos is not associated with the first byte.
661 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
662 Alan Hayward <alan.hayward@arm.com>
663 David Sherwood <david.sherwood@arm.com>
665 * tree-affine.h (get_inner_reference_aff): Return the size as a
667 * tree-affine.c (get_inner_reference_aff): Likewise.
668 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
669 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
671 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
672 Alan Hayward <alan.hayward@arm.com>
673 David Sherwood <david.sherwood@arm.com>
675 * fold-const.c (pointer_may_wrap_p): Take the offset as a
676 HOST_WIDE_INT rather than a poly_int64.
678 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
679 Alan Hayward <alan.hayward@arm.com>
680 David Sherwood <david.sherwood@arm.com>
682 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
683 HOST_WIDE_INT to poly_int64_pod.
684 (perform_symbolic_merge): Update accordingly.
685 (bswap_replace): Likewise.
687 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
688 Alan Hayward <alan.hayward@arm.com>
689 David Sherwood <david.sherwood@arm.com>
691 * tree-affine.h (aff_tree::offset): Change from widest_int
693 (wide_int_ext_for_comb): Delete.
694 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
695 constants as poly_widest_int rather than widest_int.
696 (aff_combination_constant_multiple_p): Return the multiplier
697 as a poly_widest_int.
698 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
700 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
701 version static and add an overload for poly_widest_int.
702 (aff_combination_const, aff_combination_add_cst)
703 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
704 the constants as poly_widest_int rather than widest_int.
705 (tree_to_aff_combination): Generalize INTEGER_CST case to
707 (aff_combination_to_tree): Track offsets as poly_widest_ints.
708 (aff_combination_add_product, aff_combination_mult): Handle
710 (aff_combination_constant_multiple_p): Return the multiplier
711 as a poly_widest_int.
712 * tree-predcom.c (determine_offset): Return the offset as a
714 (split_data_refs_to_components, suitable_component_p): Update
716 (valid_initializer_p): Update call to
717 aff_combination_constant_multiple_p.
718 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
719 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
720 as a poly_int64 rather than a HOST_WIDE_INT.
721 (get_address_cost): Handle polynomial offsets.
722 (iv_elimination_compare_lt): Likewise.
723 (rewrite_use_nonlinear_expr): Likewise.
725 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
726 Alan Hayward <alan.hayward@arm.com>
727 David Sherwood <david.sherwood@arm.com>
729 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
730 as a poly_int64_pod rather than a HOST_WIDE_INT.
731 (get_addr_base_and_unit_offset): Likewise.
732 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
733 (get_addr_base_and_unit_offset): Likewise.
734 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
735 to poly_int64 in example.
736 * fold-const.c (fold_binary_loc): Update call to
737 get_addr_base_and_unit_offset.
738 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
739 (maybe_canonicalize_mem_ref_addr): Likewise.
740 (gimple_fold_stmt_to_constant_1): Likewise.
741 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
743 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
744 * match.pd: Likewise.
745 * omp-low.c (lower_omp_target): Likewise.
746 * tree-sra.c (build_ref_for_offset): Likewise.
747 (build_debug_ref_for_model): Likewise.
748 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
749 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
750 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
751 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
752 (constant_pointer_difference): Likewise.
753 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
754 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
755 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
756 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
757 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
758 (set_ssa_val_to): Likewise.
759 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
760 (maybe_diag_stxncpy_trunc): Likewise.
761 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
762 * tree.c (build_simple_mem_ref_loc): Likewise.
763 (array_at_struct_end_p): Likewise.
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 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
770 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
771 (get_ref_base_and_extent_hwi): Declare.
772 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
773 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
774 (get_ref_base_and_extent_hwi): New function.
775 * cfgexpand.c (expand_debug_expr): Update call to
776 get_ref_base_and_extent.
777 * dwarf2out.c (add_var_loc_to_decl): Likewise.
778 * gimple-fold.c (get_base_constructor): Return the offset as a
779 poly_int64_pod rather than a HOST_WIDE_INT.
780 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
781 * ipa-polymorphic-call.c
782 (ipa_polymorphic_call_context::set_by_invariant)
783 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
784 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
785 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
786 rather than get_ref_base_and_extent.
787 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
788 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
789 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
791 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
792 call to get_ref_base_and_extent.
793 * tree-sra.c (create_access, get_access_for_expr): Likewise.
794 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
795 (stmt_kills_ref_p): Likewise.
796 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
797 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
799 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
800 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
801 when calling native_encode_expr.
802 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
803 call to get_ref_base_and_extent.
804 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
805 get_ref_base_and_extent.
806 * var-tracking.c (track_expr_p): Likewise.
808 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
809 Alan Hayward <alan.hayward@arm.com>
810 David Sherwood <david.sherwood@arm.com>
812 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
813 HOST_WIDE_INT to poly_int64_pod.
814 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
815 polynomail parameter offsets.
817 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
818 Alan Hayward <alan.hayward@arm.com>
819 David Sherwood <david.sherwood@arm.com>
821 * gengtype.c (main): Handle poly_int64_pod.
822 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
823 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
824 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
825 from HOST_WIDE_INT to poly_int64_pod.
826 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
827 (copy_cfa): New function.
828 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
829 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
830 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
831 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
832 (initial_return_save): Treat offsets as poly_ints.
833 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
834 offset is nonconstant.
835 (reg_save): Take the offset as a poly_int64. Fall back to
836 DW_CFA_expression for nonconstant offsets.
837 (queue_reg_save): Take the offset as a poly_int64.
838 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
840 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
841 Alan Hayward <alan.hayward@arm.com>
842 David Sherwood <david.sherwood@arm.com>
844 * rtl.h (operand_subword, operand_subword_force): Take the offset
845 as a poly_uint64 an unsigned int.
846 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
848 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
849 Alan Hayward <alan.hayward@arm.com>
850 David Sherwood <david.sherwood@arm.com>
852 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
853 'p' format code. Use INT_LIST rather than SUBREG as the example of
854 a code with an XINT and an XEXP. Remove the implication that
855 accessing an rtx field using XINT is expected to work.
856 * rtl.def (SUBREG): Change format from "ei" to "ep".
857 * rtl.h (rtunion::rt_subreg): New field.
858 (XCSUBREG): New macro.
859 (SUBREG_BYTE): Use it.
860 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
861 Update constructor accordingly.
862 (subreg_shape::operator ==): Update accordingly.
863 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
864 than an unsigned int.
865 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
866 a poly_uint64 rather than an unsigned int.
867 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
868 than an unsigned int.
869 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
870 (subreg_size_highpart_offset): Return a poly_uint64 rather than
871 an unsigned int. Take the sizes as poly_uint64s.
872 (subreg_offset_from_lsb): Return a poly_uint64 rather than
873 an unsigned int. Take the shift as a poly_uint64 rather than
875 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
876 as a poly_uint64 rather than an unsigned int.
877 (simplify_subreg_regno): Likewise.
878 (byte_lowpart_offset): Return the memory offset as a poly_int64
880 (subreg_memory_offset): Likewise. Take the subreg offset as a
881 poly_uint64 rather than an unsigned int.
882 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
883 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
884 poly_uint64 rather than an unsigned int.
885 * rtl.c (rtx_format): Describe 'p' in comment.
886 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
887 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
888 offset as a poly_uint64 rather than an unsigned int.
889 (byte_lowpart_offset): Return the memory offset as a poly_int64
891 (subreg_memory_offset): Likewise. Take the subreg offset as a
892 poly_uint64 rather than an unsigned int.
893 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
894 mode sizes as poly_uint64s rather than unsigned ints. Return a
895 poly_uint64 rather than an unsigned int.
896 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
897 (copy_insn_1): Handle 'p'.
898 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
899 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
900 an unsigned int. Return the shift in the same way.
901 (subreg_lsb): Return the shift as a poly_uint64 rather than an
903 (subreg_size_offset_from_lsb): Take the sizes and shift as
904 poly_uint64s rather than unsigned ints. Return the offset as
906 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
907 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
909 * rtlhash.c (add_rtx): Handle 'p'.
910 * genemit.c (gen_exp): Likewise.
911 * gengenrtl.c (type_from_format, gendef): Likewise.
912 * gensupport.c (subst_pattern_match, get_alternatives_number)
913 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
914 (subst_dup): Likewise.
915 * gengtype.c (adjust_field_rtx_def): Likewise.
916 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
917 (match_pattern_2): Likewise.
918 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
919 (rtx_test::subreg_field): New function.
920 (operator ==, safe_to_hoist_p, transition_parameter_type)
921 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
922 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
923 * genpeep.c (match_rtx): Likewise.
924 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
925 (rtx_writer::print_rtx_operand): Handle 'p'.
926 (print_value): Handle SUBREG.
927 * read-rtl.c (apply_int_iterator): Likewise.
928 (rtx_reader::read_rtx_operand): Handle 'p'.
929 * alias.c (rtx_equal_for_memref_p): Likewise.
930 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
931 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
933 * calls.c (expand_call): Likewise.
934 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
935 (make_extraction, gen_lowpart_for_combine): Likewise.
936 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
938 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
939 rather than an unsigned int. Treat subreg offsets as poly_ints.
940 (exp_equiv_p): Handle 'p'.
941 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
942 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
943 * dse.c (find_shift_sequence): Likewise.
944 * dwarf2out.c (rtl_for_decl_location): Likewise.
945 * expmed.c (extract_low_bits): Likewise.
946 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
947 (expand_expr_real_2): Likewise.
948 * final.c (alter_subreg): Likewise.
949 (leaf_renumber_regs_insn): Handle 'p'.
950 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
951 Treat subreg offsets as poly_ints.
952 * fwprop.c (forward_propagate_and_simplify): Likewise.
953 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
954 * ira.c (get_subreg_tracking_sizes): Likewise.
955 * ira-conflicts.c (go_through_subreg): Likewise.
956 * ira-lives.c (process_single_reg_class_operands): Likewise.
957 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
958 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
959 as a poly_uint64 rather than an unsigned int.
960 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
961 subreg offsets as poly_ints.
962 * lra-constraints.c (operands_match_p): Handle 'p'.
963 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
964 * lra-spills.c (assign_mem_slot): Likewise.
965 * postreload.c (move2add_valid_value_p): Likewise.
966 * recog.c (general_operand, indirect_operand): Likewise.
967 * regcprop.c (copy_value, maybe_mode_change): Likewise.
968 (copyprop_hardreg_forward_1): Likewise.
969 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
970 (record_subregs_of_mode): Likewise.
971 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
972 * reload.c (operands_match_p): Handle 'p'.
973 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
974 * reload1.c (alter_reg, choose_reload_regs): Likewise.
975 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
976 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
977 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
978 (simplify_const_poly_int_tests<N>::run): Likewise.
979 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
980 a poly_uint64 rather than an unsigned int.
981 * valtrack.c (debug_lowpart_subreg): Likewise.
982 * var-tracking.c (var_lowpart): Likewise.
983 (loc_cmp): Handle 'p'.
985 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
986 Alan Hayward <alan.hayward@arm.com>
987 David Sherwood <david.sherwood@arm.com>
989 * ira.c (get_subreg_tracking_sizes): New function.
990 (init_live_subregs): Take an integer size rather than a register.
991 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
992 to init_live_subregs.
994 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
995 Alan Hayward <alan.hayward@arm.com>
996 David Sherwood <david.sherwood@arm.com>
998 * expr.c (store_constructor_field): Change bitsize from a
999 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1000 HOST_WIDE_INT to a poly_int64.
1001 (store_constructor): Change size from a HOST_WIDE_INT to
1003 (store_field): Likewise bitsize and bitpos.
1005 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1006 Alan Hayward <alan.hayward@arm.com>
1007 David Sherwood <david.sherwood@arm.com>
1009 * expmed.h (store_bit_field): Change bitregion_start and
1010 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1011 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1012 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1013 (store_fixed_bit_field, store_split_bit_field): Likewise.
1014 * expr.c (store_constructor_field, store_field): Likewise.
1015 (optimize_bitfield_assignment_op): Likewise. Make the same change
1016 to bitsize and bitpos.
1017 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1018 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1019 same change in the constructor arguments.
1020 (get_best_mode): Change bitregion_start and bitregion_end from
1021 unsigned HOST_WIDE_INT to poly_uint64.
1022 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1023 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1025 (bit_field_mode_iterator::next_mode): Update for new types
1026 of m_bitregion_start and m_bitregion_end.
1027 (get_best_mode): Change bitregion_start and bitregion_end from
1028 unsigned HOST_WIDE_INT to poly_uint64.
1030 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1031 Alan Hayward <alan.hayward@arm.com>
1032 David Sherwood <david.sherwood@arm.com>
1034 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1035 accepts poly_uint64 offsets.
1036 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1037 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1038 * expmed.c (lowpart_bit_field_p): Likewise.
1039 (extract_bit_field_as_subreg): New function, split out from...
1040 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1041 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1042 extractions, check that BITSIZE matches the size of the extracted
1043 value and that BITNUM is an exact multiple of that size.
1044 If all else fails, try forcing the value into memory if
1045 BITNUM is variable, and adjusting the address so that the
1046 offset is constant. Split the part that can only handle constant
1047 bitsize and bitnum out into...
1048 (extract_integral_bit_field): ...this new function.
1049 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1050 rather than unsigned HOST_WIDE_INTs.
1052 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1053 Alan Hayward <alan.hayward@arm.com>
1054 David Sherwood <david.sherwood@arm.com>
1056 * expmed.h (store_bit_field): Take bitsize and bitnum as
1057 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1058 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1059 that returns the byte size.
1060 (store_bit_field_1): Take bitsize and bitnum as
1061 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1062 to simple_mem_bitfield_p. Split the part that can only handle
1063 constant bitsize and bitnum out into...
1064 (store_integral_bit_field): ...this new function.
1065 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1066 than unsigned HOST_WIDE_INTs.
1067 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1069 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1070 Alan Hayward <alan.hayward@arm.com>
1071 David Sherwood <david.sherwood@arm.com>
1073 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1074 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1076 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1077 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1078 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1079 offset as a poly_int64 rather than an int.
1080 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1081 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1082 * lra-constraints.c (equiv_address_substitution): Track offsets
1084 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1085 (curr_insn_transform): Handle the new form of sp_offset.
1086 * lra-eliminations.c (lra_elim_table): Change previous_offset
1087 and offset from HOST_WIDE_INT to poly_int64.
1088 (print_elim_table, update_reg_eliminate): Update accordingly.
1089 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1090 (get_elimination): Update accordingly.
1091 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1092 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1093 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1094 poly_int64 offsets generally.
1095 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1096 (mark_not_eliminable, init_elimination): Update accordingly.
1097 (remove_reg_equal_offset_note): Return a bool and pass the new
1098 offset back by pointer as a poly_int64.
1099 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1100 rather than a HOST_WIDE_INT.
1101 (do_remat): Track offsets poly_int64s.
1102 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1104 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1105 Alan Hayward <alan.hayward@arm.com>
1106 David Sherwood <david.sherwood@arm.com>
1108 * rtl.h (mem_attrs): Add a default constructor. Change size and
1109 offset from HOST_WIDE_INT to poly_int64.
1110 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1111 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1112 (widen_memory_access): Take the sizes and offsets as poly_int64s
1113 rather than HOST_WIDE_INTs.
1114 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1115 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1117 (adjust_offset_for_component_ref): Change the offset from a
1118 HOST_WIDE_INT to a poly_int64.
1119 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1120 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1121 * dce.c (find_call_stack_args): Likewise.
1122 * dse.c (record_store): Likewise.
1123 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1124 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1125 * read-rtl-function.c (test_loading_mem): Likewise.
1126 * rtlanal.c (may_trap_p_1): Likewise.
1127 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1128 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1129 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1130 (mem_attrs::mem_attrs): New function.
1131 (set_mem_attributes_minus_bitpos): Change bitpos from a
1132 HOST_WIDE_INT to poly_int64.
1133 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1134 (clear_mem_offset, clear_mem_size, change_address)
1135 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1136 initialize mem_attrs.
1137 (set_mem_offset, set_mem_size, adjust_address_1)
1138 (adjust_automodify_address_1, offset_address, widen_memory_access):
1139 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1141 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1142 Alan Hayward <alan.hayward@arm.com>
1143 David Sherwood <david.sherwood@arm.com>
1145 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1146 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1147 rather than 0 to represent an unknown size. Assert that the size
1148 is known when the mode isn't BLKmode.
1149 (may_trap_p_1): Use -1 for unknown sizes.
1150 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1152 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1153 Alan Hayward <alan.hayward@arm.com>
1154 David Sherwood <david.sherwood@arm.com>
1156 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1157 to poly_int64. Update commentary for positions_needed.large.
1158 (read_info_type): Change offset and width from HOST_WIDE_INT
1160 (set_usage_bits): Likewise.
1161 (canon_address): Return the offset as a poly_int64 rather than
1162 a HOST_WIDE_INT. Use strip_offset_and_add.
1163 (set_all_positions_unneeded, any_positions_needed_p): Use
1164 positions_needed.large to track stores with non-constant widths.
1165 (all_positions_needed_p): Likewise. Take the offset and width
1166 as poly_int64s rather than ints. Assert that rhs is nonnull.
1167 (record_store): Cope with non-constant offsets and widths.
1168 Nullify the rhs of an earlier store if we can't tell which bytes
1170 (find_shift_sequence): Take the access_size and shift as poly_int64s
1172 (get_stored_val): Take the read_offset and read_width as poly_int64s
1173 rather than HOST_WIDE_INTs.
1174 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1175 non-constant offsets and widths.
1177 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1178 Alan Hayward <alan.hayward@arm.com>
1179 David Sherwood <david.sherwood@arm.com>
1181 * inchash.h (inchash::hash::add_poly_int): New function.
1182 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1183 Use poly_int64 rather than HOST_WIDE_INT.
1184 (ao_ref::max_size_known_p): New function.
1185 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1186 rather than HOST_WIDE_INT.
1187 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1188 to temporaries until its interface is adjusted to match.
1189 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1190 (aliasing_component_refs_p, decl_refs_may_alias_p)
1191 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1192 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1193 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1195 * alias.c (ao_ref_from_mem): Likewise.
1196 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1197 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1198 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1199 (maybe_trim_complex_store, maybe_trim_constructor_store)
1200 (live_bytes_read, dse_classify_store): Likewise.
1201 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1202 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1203 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1204 (vn_reference_lookup_3): Likewise.
1205 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1207 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1208 Alan Hayward <alan.hayward@arm.com>
1209 David Sherwood <david.sherwood@arm.com>
1211 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1212 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1213 instead of ranges_overlap_p.
1215 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1216 Alan Hayward <alan.hayward@arm.com>
1217 David Sherwood <david.sherwood@arm.com>
1219 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1220 sizes as poly_int64s rather than HOST_WIDE_INTs.
1222 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1223 Alan Hayward <alan.hayward@arm.com>
1224 David Sherwood <david.sherwood@arm.com>
1226 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1227 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1228 * gimple-fold.c (fold_ctor_reference): Likewise.
1230 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1231 Alan Hayward <alan.hayward@arm.com>
1232 David Sherwood <david.sherwood@arm.com>
1234 * target.def (dwarf_poly_indeterminate_value): New hook.
1235 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1236 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1237 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1238 * doc/tm.texi: Regenerate.
1239 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1240 offset as a poly_int64.
1241 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1242 offset as a poly_int64.
1243 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1244 Take the offset as a poly_int64.
1245 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1246 (frame_pointer_fb_offset): Change to a poly_int64.
1247 (int_loc_descriptor): Take the offset as a poly_int64. Use
1248 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1249 (based_loc_descr): Take the offset as a poly_int64.
1250 Use strip_offset_and_add to handle (plus X (const)).
1251 Use new_reg_loc_descr instead of an open-coded version of the
1252 previous implementation.
1253 (mem_loc_descriptor): Handle CONST_POLY_INT.
1254 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1255 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1257 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1258 Alan Hayward <alan.hayward@arm.com>
1259 David Sherwood <david.sherwood@arm.com>
1261 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1262 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1263 * inchash.h (inchash::hash::add_poly_hwi): New function.
1264 * gengtype.c (main): Register poly_int64.
1265 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1266 offset as a poly_int.
1267 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1268 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1269 offset as a poly_int64.
1270 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1271 * print-rtl.c (print_poly_int): New function.
1272 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1274 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1276 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1277 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1278 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1279 Enforce track_offset_p for parts of a PARALLEL.
1280 (vt_add_function_parameter): Use const_offset for the final
1281 offset to track. Use get_tracked_reg_offset for the parts
1284 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1285 Alan Hayward <alan.hayward@arm.com>
1286 David Sherwood <david.sherwood@arm.com>
1288 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1289 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1290 * doc/tm.texi: Regenerate.
1291 * hooks.h (hook_bool_uint_uint_true): Delete.
1292 (hook_bool_puint64_puint64_true): Declare.
1293 * hooks.c (hook_bool_uint_uint_true): Delete.
1294 (hook_bool_puint64_puint64_true): New function.
1295 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1296 instead of unsigned ints.
1297 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1298 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1300 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1301 Alan Hayward <alan.hayward@arm.com>
1302 David Sherwood <david.sherwood@arm.com>
1304 * optabs.h (expand_operand): Add an int_value field.
1305 (create_expand_operand): Add an int_value parameter and use it
1306 to initialize the new expand_operand field.
1307 (create_integer_operand): Replace with a declaration of a function
1308 that accepts poly_int64s. Move the implementation to...
1309 * optabs.c (create_integer_operand): ...here.
1310 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1311 the mode preserves the value of int_value, instead of calling
1312 const_int_operand on the rtx. Use gen_int_mode to generate
1315 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1316 Alan Hayward <alan.hayward@arm.com>
1317 David Sherwood <david.sherwood@arm.com>
1319 * dumpfile.h (dump_dec): Declare.
1320 * dumpfile.c (dump_dec): New function.
1321 * pretty-print.h (pp_wide_integer): Turn into a function and
1322 declare a poly_int version.
1323 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1325 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1326 Alan Hayward <alan.hayward@arm.com>
1327 David Sherwood <david.sherwood@arm.com>
1329 * doc/generic.texi (POLY_INT_CST): Document.
1330 * tree.def (POLY_INT_CST): New tree code.
1331 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1332 * tree-core.h (tree_poly_int_cst): New struct.
1333 (tree_node): Add a poly_int_cst field.
1334 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1335 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1336 instead of a wide_int_ref.
1337 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1339 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1340 instead of an unsigned HOST_WIDE_INT.
1341 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1342 (ptrdiff_tree_p): Declare.
1343 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1344 extern inline implementations if the target doesn't use POLY_INT_CST.
1345 (poly_int_tree_p): New function.
1346 (wi::unextended_tree): New class.
1347 (wi::int_traits <unextended_tree>): New override.
1348 (wi::extended_tree): Add a default constructor.
1349 (wi::extended_tree::get_tree): New function.
1350 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1351 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1352 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1353 (wi::tree_to_poly_wide_ref): New typedefs.
1354 (wi::ints_for): Provide overloads for extended_tree and
1356 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1357 (wi::to_wide): New functions.
1358 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1359 * tree.c (poly_int_cst_hasher): New struct.
1360 (poly_int_cst_hash_table): New variable.
1361 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1362 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1364 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1365 (init_ttree): Initialize poly_int_cst_hash_table.
1366 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1367 a poly_int64 instead of a HOST_WIDE_INT.
1368 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1369 instead of an unsigned HOST_WIDE_INT.
1370 (wide_int_to_tree): Rename to...
1371 (wide_int_to_tree_1): ...this.
1372 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1373 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1374 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1375 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1376 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1377 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1379 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1380 (streamer_read_tree_body): Likewise.
1381 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1382 (streamer_write_tree_body): Likewise.
1383 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1384 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1385 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1386 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1387 * gimple-expr.h (is_gimple_constant): Likewise.
1388 * gimplify.c (maybe_with_size_expr): Likewise.
1389 * print-tree.c (print_node): Likewise.
1390 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1391 * tree-pretty-print.c (dump_generic_node): Likewise.
1392 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1393 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1394 * tree-vrp.c (compare_values_warnv): Likewise.
1395 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1396 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1397 (force_expr_to_var_cost): Likewise.
1398 * tree-ssa-loop.c (for_each_index): Likewise.
1399 * fold-const.h (build_invariant_address, size_int_kind): Take a
1400 poly_int64 instead of a HOST_WIDE_INT.
1401 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1402 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1404 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1405 (int_const_binop_2): New function, split out from...
1406 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1407 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1408 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1409 * gimple-ssa-strength-reduction.c (slsr_process_add)
1410 (slsr_process_mul): Check for INTEGER_CSTs before using them
1412 * stor-layout.c (bits_from_bytes): New function.
1413 (bit_from_pos): Use it.
1414 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1415 by BITS_PER_UNIT to get the TYPE_SIZE.
1416 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1417 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1419 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1420 Alan Hayward <alan.hayward@arm.com>
1421 David Sherwood <david.sherwood@arm.com>
1423 * doc/rtl.texi (const_poly_int): Document. Also document the
1424 rtl sharing behavior.
1425 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1426 * rtl.h (const_poly_int_def): New struct.
1427 (rtx_def::u): Add a cpi field.
1428 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1429 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1430 (wi::rtx_to_poly_wide_ref): New typedef
1431 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1432 (poly_int_rtx_p): New functions.
1433 (trunc_int_for_mode): Declare a poly_int64 version.
1434 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1435 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1437 (strip_offset): Declare.
1438 (strip_offset_and_add): New function.
1439 * rtl.def (CONST_POLY_INT): New rtx code.
1440 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1441 (shared_const_p): Use poly_int_rtx_p.
1442 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1444 (gen_int_shift_amount): Likewise.
1445 * emit-rtl.c (const_poly_int_hasher): New class.
1446 (const_poly_int_htab): New variable.
1447 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1448 (const_poly_int_hasher::hash): New function.
1449 (const_poly_int_hasher::equal): Likewise.
1450 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1451 (immed_wide_int_const): Rename to...
1452 (immed_wide_int_const_1): ...this and make static.
1453 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1454 instead of a wide_int_ref.
1455 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1456 (gen_lowpart_common): Handle CONST_POLY_INT.
1457 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1458 * cselib.c (cselib_hash_rtx): Likewise.
1459 * dwarf2out.c (const_ok_for_output_1): Likewise.
1460 * expr.c (convert_modes): Likewise.
1461 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1462 * rtlhash.c (add_rtx): Likewise.
1463 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1464 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1465 Handle existing CONST_POLY_INT rtxes.
1466 * expmed.h (expand_shift): Take a poly_int64 instead of a
1468 * expmed.c (expand_shift): Likewise.
1469 * rtlanal.c (strip_offset): New function.
1470 (commutative_operand_precedence): Give CONST_POLY_INT the same
1471 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1473 * rtl-tests.c (const_poly_int_tests): New struct.
1474 (rtl_tests_c_tests): Use it.
1475 * simplify-rtx.c (simplify_const_unary_operation): Handle
1477 (simplify_const_binary_operation): Likewise.
1478 (simplify_binary_operation_1): Fold additions of symbolic constants
1479 and CONST_POLY_INTs.
1480 (simplify_subreg): Handle extensions and truncations of
1482 (simplify_const_poly_int_tests): New struct.
1483 (simplify_rtx_c_tests): Use it.
1484 * wide-int.h (storage_ref): Add default constructor.
1485 (wide_int_ref_storage): Likewise.
1486 (trailing_wide_ints): Use GTY((user)).
1487 (trailing_wide_ints::operator[]): Add a const version.
1488 (trailing_wide_ints::get_precision): New function.
1489 (trailing_wide_ints::extra_size): Likewise.
1491 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1492 Alan Hayward <alan.hayward@arm.com>
1493 David Sherwood <david.sherwood@arm.com>
1495 * emit-rtl.h (gen_int_shift_amount): Declare.
1496 * emit-rtl.c (gen_int_shift_amount): New function.
1497 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1499 * calls.c (shift_return_value): Likewise.
1500 * cse.c (fold_rtx): Likewise.
1501 * dse.c (find_shift_sequence): Likewise.
1502 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1503 (expand_shift, expand_smod_pow2): Likewise.
1504 * lower-subreg.c (shift_cost): Likewise.
1505 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1506 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1507 (expand_vec_perm_var): Likewise.
1508 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1509 (simplify_binary_operation_1): Likewise.
1510 * combine.c (try_combine, find_split_point, force_int_to_mode)
1511 (simplify_shift_const_1, simplify_shift_const): Likewise.
1512 (change_zero_ext): Likewise. Use simplify_gen_binary.
1514 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1516 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1518 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1520 * doc/invoke.texi (ARM Options): Document accepted extension options
1521 for -march=armv8.3-a.
1523 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1526 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1527 or --with-float=softfp, set the default CPU to arm10e.
1529 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1531 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1532 * config/visium/predicates.md (const_shift_operand): Likewise.
1533 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1534 (visium_legitimize_reload_address): Likewise.
1536 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1539 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1541 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1544 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1546 2017-12-20 Julia Koval <julia.koval@intel.com>
1548 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1549 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1550 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1551 * config.gcc: Include vpclmulqdqintrin.h.
1552 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1553 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1555 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1556 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1557 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1558 * config/i386/i386.c (isa2_opts): Add -mcx16.
1559 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1560 (ix86_option_override_internal): Move mcx16 to flags2.
1561 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1562 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1563 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1564 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1565 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1566 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1567 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1568 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1569 * doc/invoke.texi: Add -mvpclmulqdq.
1571 2017-12-20 Tom de Vries <tom@codesourcery.com>
1574 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1575 * calls.c (rtx_for_static_chain): ... here. New function.
1576 * calls.h (rtx_for_static_chain): Declare.
1577 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1578 instead of targetm.calls.static_chain.
1579 * df-scan.c (df_get_entry_block_def_set): Same.
1581 2017-12-19 Tom de Vries <tom@codesourcery.com>
1583 PR tree-optimization/83493
1584 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1587 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1589 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1590 inform with hardcoded english plural handling.
1592 2017-12-18 Jeff Law <law@redhat.com>
1594 PR tree-optimization/83477
1595 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1596 a non-virtual PHI, always push a new range.
1598 2017-12-19 Martin Sebor <msebor@redhat.com>
1601 * builtins.c (compute_objsize): Handle non-constant offsets.
1603 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1605 PR tree-optimization/83444
1606 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1607 character load case, if get_stridx on MEM_REF's operand doesn't
1608 look usable, retry with get_addr_stridx.
1610 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1613 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1614 (variable_tracking_main_1): Keep markers even when VTA fails.
1617 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1618 even if there are markers before it.
1619 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1621 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1623 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1624 typical order conditions.
1625 * sel-sched.c (extract_new_fences_from): Likewise.
1626 * config/visium/constraints.md (J, K, L): Likewise.
1627 * config/visium/predicates.md (const_shift_operand): Likewise.
1628 * config/visium/visium.c (visium_legitimize_address,
1629 visium_legitimize_reload_address): Likewise.
1630 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1631 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1632 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1633 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1634 SET_DUMP_DETAIL): Likewise.
1635 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1636 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1637 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1638 avr_set_core_architecture, avr_set_current_function,
1639 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1640 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1641 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1642 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1643 avr_map_decompose, avr_fold_builtin): Likewise.
1644 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1645 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1646 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1647 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1648 * config/m32c/m32c.c (m32c_conditional_register_usage,
1649 m32c_address_cost): Likewise.
1650 * config/m32c/predicates.md (shiftcount_operand,
1651 longshiftcount_operand): Likewise.
1652 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1653 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1654 can_use_cdx_ldstw): Likewise.
1655 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1656 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1658 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1659 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1660 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1661 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1662 * config/vax/vax.c (adjacent_operands_p): Likewise.
1663 * config/ft32/constraints.md (L, b, KA): Likewise.
1664 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1666 * cfgexpand.c (expand_stack_alignment): Likewise.
1667 * gcse.c (insert_expr_in_table): Likewise.
1668 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1669 * cgraphunit.c (cgraph_node::expand): Likewise.
1670 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1671 * emit-rtl.c (add_insn): Likewise.
1672 * input.c (dump_location_info): Likewise.
1673 * passes.c (NEXT_PASS): Likewise.
1674 * read-rtl-function.c (parse_note_insn_name,
1675 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1677 * sched-rgn.c (sched_rgn_init): Likewise.
1678 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1679 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1680 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1681 * explow.c (eliminate_constant_term): Likewise.
1682 * final.c (leaf_renumber_regs_insn): Likewise.
1683 * cfgrtl.c (print_rtl_with_bb): Likewise.
1684 * genhooks.c (emit_init_macros): Likewise.
1685 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1686 * tree-data-ref.c (conflict_fn): Likewise.
1687 * selftest.c (assert_streq): Likewise.
1688 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1689 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1690 fold_binary_loc, multiple_of_p): Likewise.
1691 * reload.c (push_reload, find_equiv_reg): Likewise.
1692 * et-forest.c (et_nca, et_below): Likewise.
1693 * dbxout.c (dbxout_symbol_location): Likewise.
1694 * reorg.c (relax_delay_slots): Likewise.
1695 * dojump.c (do_compare_rtx_and_jump): Likewise.
1696 * gengtype-parse.c (type): Likewise.
1697 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1698 simplify_const_relational_operation): Likewise.
1699 * reload1.c (do_output_reload): Likewise.
1700 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1701 * gengtype.c (type_for_name): Likewise.
1702 * gimple-ssa-sprintf.c (format_directive): Likewise.
1704 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1707 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1708 accessing it. Adjust comment.
1710 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1713 * predict.c (zero_one_minusone): New function.
1714 (apply_return_prediction): Avoid return prediction for functions
1715 returning only -1, 0 and 1 values, unless they only return -1 and 0
1718 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1720 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1721 fall-through warning.
1722 (arc_compute_frame_size): Remove unused variables.
1723 (arc_print_operand): Fix fprintif format.
1724 (arc_can_follow_jump): Clean fall-through warning.
1726 2017-12-19 Marek Polacek <polacek@redhat.com>
1729 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1732 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1734 * config/arc/arc.c (overriderregs): New variable.
1735 (arc_override_options): Track fixed/call saved/call options.
1736 (arc_conditional_register_usage): Check against overrideregs
1737 variable whenever we change register properties.
1739 2017-12-19 Nathan Sidwell <nathan@acm.org>
1741 * opts.c (finish_options): Don't prefix dump_base_name if it
1742 already contains directories.
1744 2017-12-19 Martin Liska <mliska@suse.cz>
1746 PR rtl-optimization/82675
1747 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1748 more element in sbitmap.
1750 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1752 * gimplify.c (gimplify_expr): Use error_operand_p.
1754 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1757 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1760 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1762 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1763 the jump is to a label.
1765 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1767 PR tree-optimization/83444
1768 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1773 * ipa-inline.c (flatten_remove_node_hook): New function.
1774 (ipa_inline): Keep only nodes with flatten attribute at the end of
1775 the array in the order from ipa_reverse_postorder, only walk that
1776 portion of array for flattening, if there is more than one such
1777 node, temporarily register a removal hook and ignore removed nodes.
1779 PR tree-optimization/80631
1780 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1781 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1782 IFN_REDUC_MAX or IFN_REDUC_MIN.
1784 2017-12-18 Jeff Law <law@redhat.com>
1786 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1787 record symbolic equivalences from backedges in the CFG.
1790 2017-11-19 Jeff Law <law@redhat.com>
1792 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1793 of degenerates resulting from ignoring an edge.
1795 2017-12-18 Martin Sebor <msebor@redhat.com>
1798 PR tree-optimization/78450
1799 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1800 (handle_builtin_strlen): Call it.
1802 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1804 PR rtl-optimization/83424
1805 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1807 2017-12-18 Marek Polacek <polacek@redhat.com>
1810 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1811 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1814 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1816 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1817 (umaddsidi4, umaddsidi_split): Likewise.
1819 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1821 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1824 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1827 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1829 2017-12-18 Richard Biener <rguenther@suse.de>
1831 PR tree-optimization/81877
1832 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1833 (outermost_indep_loop): Adjust.
1834 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1835 (can_sm_ref_p): Adjust.
1837 2017-12-18 Richard Biener <rguenther@suse.de>
1840 * tree.c (array_at_struct_end_p): Return true if the underlying
1841 object has space for at least one element in excess of what
1842 the array domain specifies.
1844 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1846 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1847 example to avoid overfull hbox.
1848 * doc/invoke.texi (Option Summary): Add missing @gol.
1849 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1852 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1853 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1855 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1856 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1857 (AArch64 Options): Document -mverbose-cost-dump.
1858 (ARM Options): Likewise, plus -mflip-thumb.
1860 2017-12-17 Martin Sebor <msebor@redhat.com>
1863 * gimple-ssa-warn-restrict.c
1864 (builtin_memref::offset_out_of_bounds): Correct the handling of
1867 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1869 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1871 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1873 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1875 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1877 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1880 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1882 * doc/invoke.texi: Fix some typos.
1884 2017-12-16 Martin Sebor <msebor@redhat.com>
1886 PR tree-optimization/78918
1887 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1888 * builtins.c (check_sizes): Rename...
1889 (check_access): ...to this. Rename function arguments for clarity.
1890 (check_memop_sizes): Adjust names.
1891 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1892 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1893 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1894 (check_strncat_sizes, expand_builtin_strncat): Same.
1895 (expand_builtin_strncpy, expand_builtin_memset): Same.
1896 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1897 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1898 (maybe_emit_sprintf_chk_warning): Same.
1899 (expand_builtin_strcpy): Adjust.
1900 (expand_builtin_stpcpy): Same.
1901 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1902 in pointer-checking forms of memcpy, memmove, and mempcpy.
1903 (gcall_to_tree_minimal, max_object_size): Define new functions.
1904 * builtins.h (max_object_size): Declare.
1905 * calls.c (alloc_max_size): Call max_object_size instead of
1906 hardcoding ssizetype limit.
1907 (get_size_range): Handle new argument.
1908 * calls.h (get_size_range): Add a new argument.
1909 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1910 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1911 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1913 (gimple_fold_builtin_memory_chk): Same.
1914 (gimple_fold_builtin_stxcpy_chk): New function.
1915 * gimple-ssa-warn-restrict.c: New source.
1916 * gimple-ssa-warn-restrict.h: New header.
1917 * gimple.c (gimple_build_call_from_tree): Propagate location.
1918 * passes.def (pass_warn_restrict): Add new pass.
1919 * tree-pass.h (make_pass_warn_restrict): Declare.
1920 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1922 (handle_builtin_strcat): Same.
1923 (strlen_optimize_stmt): Rename...
1924 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1925 stpncpy, strncpy, and their checking forms.
1927 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1929 PR rtl-optimization/82849
1930 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1931 and get_max_loop_iterations_int.
1933 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1934 Alan Hayward <alan.hayward@arm.com>
1935 David Sherwood <david.sherwood@arm.com>
1937 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1938 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1939 size as a poly_uint64.
1940 (mode_for_vector, mode_for_int_vector): Take the number of vector
1941 elements as a poly_uint64.
1942 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1943 size as a poly_uint64.
1944 (mode_for_vector, mode_for_int_vector): Take the number of vector
1945 elements as a poly_uint64.
1947 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1948 Alan Hayward <alan.hayward@arm.com>
1949 David Sherwood <david.sherwood@arm.com>
1951 * machmode.h (MACRO_MODE): New macro.
1952 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1953 * targhooks.c (default_libcall_value, default_secondary_reload)
1954 (default_memory_move_cost, default_register_move_cost)
1955 (default_class_max_nregs): Likewise.
1957 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1958 Alan Hayward <alan.hayward@arm.com>
1959 David Sherwood <david.sherwood@arm.com>
1961 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1962 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1964 * genautomata.c (main): Likewise.
1965 * genconditions.c (write_header): Likewise.
1966 * genemit.c (main): Likewise.
1967 * genextract.c (print_header): Likewise.
1968 * genopinit.c (main): Likewise.
1969 * genoutput.c (output_prologue): Likewise.
1970 * genpeep.c (main): Likewise.
1971 * genpreds.c (write_insn_preds_c): Likewise.
1972 * genrecog.c (writer_header): Likewise.
1973 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1974 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1975 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1976 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1977 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1978 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1979 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1980 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1981 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1982 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1983 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1984 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1985 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1986 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1987 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1988 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1989 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1990 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1991 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1992 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1993 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1994 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1995 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1996 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1997 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1998 * config/darwin.c (IN_TARGET_CODE): Likewise.
1999 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2000 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2001 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2002 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2003 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2004 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2005 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2006 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2007 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2008 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2009 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2010 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2011 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2012 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2013 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2014 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2015 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2016 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2017 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2018 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2019 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2020 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2021 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2022 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2023 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2024 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2025 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2026 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2027 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2028 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2029 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2030 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2031 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2032 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2033 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2034 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2035 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2036 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2037 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2038 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2039 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2040 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2041 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2042 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2043 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2044 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2045 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2046 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2047 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2048 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2049 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2050 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2051 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2052 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2053 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2054 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2055 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2056 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2057 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2058 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2059 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2060 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2061 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2062 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2063 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2064 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2065 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2066 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2067 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2068 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2069 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2070 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2071 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2072 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2073 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2074 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2075 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2076 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2077 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2078 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2079 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2080 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2081 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2082 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2083 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2084 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2085 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2086 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2087 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2088 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2089 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2090 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2091 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2092 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2093 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2094 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2095 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2096 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2097 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2098 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2099 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2100 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2101 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2102 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2103 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2104 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2106 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2107 Alan Hayward <alan.hayward@arm.com>
2108 David Sherwood <david.sherwood@arm.com>
2110 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2113 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2114 Alan Hayward <alan.hayward@arm.com>
2115 David Sherwood <david.sherwood@arm.com>
2117 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2118 * doc/md.texi (vec_series@var{m}): Document.
2119 * tree.def (VEC_SERIES_EXPR): New tree code.
2120 * tree.h (build_vec_series): Declare.
2121 * tree.c (build_vec_series): New function.
2122 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2123 * tree-pretty-print.c (dump_generic_node): Likewise.
2124 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2125 * tree-inline.c (estimate_operator_cost): Likewise.
2126 * expr.c (expand_expr_real_2): Likewise.
2127 * optabs-tree.c (optab_for_tree_code): Likewise.
2128 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2129 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2130 * expmed.c (make_tree): Handle VEC_SERIES.
2131 * optabs.def (vec_series_optab): New optab.
2132 * optabs.h (expand_vec_series_expr): Declare.
2133 * optabs.c (expand_vec_series_expr): New function.
2134 * tree-vect-generic.c (expand_vector_operations_1): Check that
2135 the operands also have vector type.
2137 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2138 Alan Hayward <alan.hawyard@arm.com>
2139 David Sherwood <david.sherwood@arm.com>
2141 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2142 (VEC_COND_EXPR): Add missing @tindex.
2143 * doc/md.texi (vec_duplicate@var{m}): Document.
2144 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2145 * tree.c (build_vector_from_val): Add stubbed-out handling of
2146 variable-length vectors, using VEC_DUPLICATE_EXPR.
2147 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2148 * cfgexpand.c (expand_debug_expr): Likewise.
2149 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2150 * tree-inline.c (estimate_operator_cost): Likewise.
2151 * tree-pretty-print.c (dump_generic_node): Likewise.
2152 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2153 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2154 (test_vec_duplicate_folding): New function.
2155 (fold_const_c_tests): Call it.
2156 * optabs.def (vec_duplicate_optab): New optab.
2157 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2158 * optabs.h (expand_vector_broadcast): Declare.
2159 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2160 vec_duplicate_optab.
2161 * expr.c (store_constructor): Try using vec_duplicate_optab for
2163 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2165 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2168 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2169 div/mod latencies a bit.
2171 2017-12-15 Jeff Law <law@redhat.com>
2173 PR tree-optimization/36550
2174 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2175 (mark_threaded_blocks): Rewrite code to avoid block copying when
2176 optimizing for size. Don't pessimize blocks which will be
2177 copied, but all the statements will be dead.
2179 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2181 PR tree-optimization/81165
2182 * tree-ssa-threadupdate.c (uses_in_bb): New.
2183 (estimate_threading_killed_stmts): New.
2184 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2185 * tree-ssa-threadedge.c
2186 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2189 2017-12-15 Jeff Law <law@redhat.com>
2191 PR tree-optimization/83410
2192 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2193 threads when parallelizing loops.
2195 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2197 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2199 * config/alpha/alpha.c (vms_attribute_table): Swap
2200 affects_type_identity and handler fields, adjust comments.
2201 * config/mips/mips.c (mips_attribute_table): Likewise.
2202 * config/visium/visium.c (visium_attribute_table): Likewise.
2203 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2204 * config/microblaze/microblaze.c (microblaze_attribute_table):
2206 * config/spu/spu.c (spu_attribute_table): Likewise.
2207 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2208 * config/arc/arc.c (arc_attribute_table): Likewise.
2209 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2210 * config/v850/v850.c (v850_handle_interrupt_attribute,
2211 v850_handle_data_area_attribute): Formatting fixes.
2212 (v850_attribute_table): Swap affects_type_identity and handler
2213 fields, adjust comments.
2214 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2215 * config/arm/arm.c (arm_attribute_table): Likewise.
2216 * config/avr/avr.c (avr_attribute_table): Likewise.
2217 * config/s390/s390.c (s390_attribute_table): Likewise.
2218 * config/sh/sh.c (sh_attribute_table): Likewise.
2219 * config/i386/i386.c (ix86_handle_cconv_attribute,
2220 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2221 (ix86_attribute_table): Swap affects_type_identity and handler
2222 fields, adjust comments.
2223 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2224 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2225 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2226 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2227 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2228 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2229 * config/rx/rx.c (rx_attribute_table): Likewise.
2230 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2231 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2232 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2233 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2234 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2235 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2236 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2237 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2238 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2239 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2240 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2242 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2244 (register_scoped_attributes, decl_attributes): Formatting fixes.
2246 PR tree-optimization/83269
2247 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2248 subtraction in arg0's type if type is signed and arg0 is unsigned.
2252 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2253 convert? on @0 instead of convert. Check type of @1, not @0.
2254 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2255 convert? on @0 instead of convert. Check type of @1, not @0.
2256 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2257 only optimize if either both @1 and @2 types are narrower
2258 precision, or both are wider or equal precision, and in the former
2259 case only if both have undefined overflow.
2261 2017-12-15 Richard Biener <rguenther@suse.de>
2264 * internal-fn.def (IFN_NOP): Add.
2265 * internal-fn.c (expand_NOP): Do nothing.
2266 * lto-streamer-in.c (input_function): Instead of removing
2267 sanitizer calls replace them with IFN_NOP calls.
2269 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2270 Alan Hayward <alan.hayward@arm.com>
2271 David Sherwood <david.sherwood@arm.com>
2273 * dse.c (store_info, read_info_type): Replace begin and end with
2275 (print_range): New function.
2276 (set_all_positions_unneeded, any_positions_needed_p)
2277 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2279 (record_store): Likewise. Optimize the case in which all positions
2281 (get_stored_val): Replace read_begin and read_end with read_offset
2283 (replace_read): Update call accordingly.
2285 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2287 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2288 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2289 (loop_cand::loop_cand): Initialize above members.
2290 (loop_cand::supported_operations): Delete.
2291 (loop_cand::can_interchange_p): Inline above function.
2292 (loop_cand::classify_simple_reduction): Record number of constant
2293 initialized simple reductions.
2294 (should_interchange_loops): New parameters. Check stmt cost of loops
2296 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2297 Update call to should_interchange_loops.
2298 (should_interchange_loop_nest): Update call to
2299 should_interchange_loops.
2301 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2304 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2305 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2306 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2308 2017-12-15 Julia Koval <julia.koval@intel.com>
2310 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2311 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2312 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2313 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2314 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2316 2017-12-15 Julia Koval <julia.koval@intel.com>
2318 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2319 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2320 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2321 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2322 _mm_aesenc_epi128): New intrinsics.
2324 2017-12-15 Julia Koval <julia.koval@intel.com>
2326 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2327 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2328 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2329 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2330 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2332 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2334 * gimple-ssa-strength-reduction.c (analyze_increments):
2335 Distinguish replacement costs for constant strides from those for
2338 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2340 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2342 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2344 * doc/invoke.texi: Document -Wcast-function-type.
2345 * recog.h (stored_funcptr): Change signature.
2346 * tree-dump.c (dump_node): Avoid warning.
2347 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2349 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2352 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2355 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2357 PR tree-optimization/83312
2358 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2359 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2360 "val" param, and to cope with arbitrary basic blocks.
2361 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2362 handle NULL_TREE for "val", dropping "bb" param.
2363 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2364 drop "bb" param. Handle NULL_TREE for "val".
2365 (find_case_label_for_value): Make "switch_stmt" param const.
2366 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2368 (vrp_prop::check_all_array_refs): Reimplement as...
2369 (check_array_bounds_dom_walker::before_dom_children): ...this new
2370 vfunc. Replace linear search through BB block list, excluding
2371 those with non-executable in-edges via dominator walk.
2373 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2375 * config/arm/arm.opt (mverbose-cost-dump): New option.
2376 * config/arm/arm.c (arm_rtx_costs): Use it.
2378 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2381 * reload1.c (emit_input_reload_insns): Skip debug markers.
2383 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2385 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2389 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2390 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2391 * config/sh/sh.c: Skip debug insns besides notes.
2392 * config/sh/sh.md: Likewise.
2393 * config/sh/sh_treg_combine.cc: Likewise.
2394 * config/sh/sync.md: Likewise.
2396 2017-12-14 Tom de Vries <tom@codesourcery.com>
2398 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2399 Add item for weak_undefined.
2401 2017-12-14 Richard Biener <rguenther@suse.de>
2403 PR tree-optimization/67842
2404 * sese.h (bb_in_region): Remove #if 0'ed code.
2406 2017-12-14 Richard Biener <rguenther@suse.de>
2408 PR tree-optimization/83326
2409 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2410 may_be_zero parameter and handle it by not marking the first
2411 peeled copy as not exiting the loop.
2412 (try_peel_loop): Likewise.
2413 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2414 to handle the case of constant or zero iterations and perform
2415 loop header copying on-the-fly.
2417 2017-12-14 Richard Biener <rguenther@suse.de>
2419 PR tree-optimization/83418
2420 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2421 Instead of asserting we don't get unfolded comparisons deal with
2424 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2427 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2428 basic blocks. Assert debug bind insns don't appear outside of bbs,
2429 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2432 PR tree-optimization/83198
2433 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2434 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2435 value if arg is a REAL_CST with incompatible type.
2437 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2438 Bin Cheng <bin.cheng@arm.com>
2441 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2443 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2446 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2448 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2449 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2451 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2452 Alan Hayward <alan.hayward@arm.com>
2453 David Sherwood <david.sherwood@arm.com>
2455 * poly-int.h: New file.
2456 * poly-int-types.h: Likewise.
2457 * coretypes.h: Include them.
2458 (POLY_INT_CONVERSION): Define.
2459 * target.def (estimated_poly_value): New hook.
2460 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2461 * doc/tm.texi: Regenerate.
2462 * doc/poly-int.texi: New file.
2463 * doc/gccint.texi: Include it.
2464 * doc/rtl.texi: Describe restrictions on subreg modes.
2465 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2466 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2467 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2468 * targhooks.h (default_estimated_poly_value): Declare.
2469 * targhooks.c (default_estimated_poly_value): New function.
2470 * target.h (estimated_poly_value): Likewise.
2471 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2472 (wi::unary_traits): Delete.
2473 (wi::binary_traits::signed_shift_result_type): Define for
2474 offset_int << HOST_WIDE_INT, etc.
2475 (generic_wide_int::operator <<=): Define for all types and use
2476 wi::lshift instead of <<.
2477 (wi::hwi_with_prec): Add a default constructor.
2478 (wi::ints_for): New class.
2479 (operator <<): Define for all wide-int types.
2480 (operator /): New function.
2481 (operator %): Likewise.
2482 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2483 (ASSERT_MAYBE_NE_AT): New macros.
2485 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2486 Dominik Vogt <vogt@linux.vnet.ibm.com>
2489 * emit-rtl.c (init_emit): Remove ??? comment.
2490 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2491 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2492 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2493 alignment of 3 virtual registers to BITS_PER_WORD.
2495 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2497 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2499 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2500 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2502 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2505 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2507 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2509 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2510 with the exception of debug begin stmt markers appear before
2514 * final.c (rest_of_handle_final): Call variable_tracking_main only
2515 if !flag_var_tracking.
2517 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2518 Jakub Jelinek <jakub@redhat.com>
2522 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2523 labels debug stmts that can only appear after labels.
2525 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2527 PR rtl-optimization/82398
2528 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2529 EXPR_USEFULNESS in priority comparison.
2531 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2533 PR rtl-optimization/83393
2534 * combine.c (move_deaths): If reg_stat points to a too new insn in
2535 last_death, do not use it: find the proper insn instead.
2537 2017-12-12 Jeff Law <law@redhat.com>
2539 PR tree-optimization/83298
2540 PR tree-optimization/83362
2541 PR tree-optimization/83383
2542 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2543 push_value_range a public interface. Add new argument to
2544 record_ranges_from_stmt.
2545 * gimple-ssa-evrp-analyze.c
2546 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2547 Update comments. Handle recording temporary equivalences.
2548 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2549 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2550 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2551 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2552 gimple-ssa-evrp-analyze.h.
2553 (record_temporary_equivalences_from_phis): Add new argument. When
2554 the PHI arg is an SSA_NAME, set the result's range to the range
2556 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2557 from statements too.
2558 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2559 Pass it down to children as needed.
2560 (thread_outgoing_edges): Likewise.
2561 (thread_across_edge): Likewise. Push/pop range state as needed.
2562 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2564 2017-12-12 Julia Koval <julia.koval@intel.com>
2566 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2567 (PTA_CANNONLAKE): Remove PTA_CLWB.
2569 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2572 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2573 (vcondv2div2df): Likewise.
2574 (vconduv2dfv2di): Likewise.
2576 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2578 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2579 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2580 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2581 ATTR_MATHFN_FPROUNDING_ERRNO.
2583 2017-12-12 Richard Biener <rguenther@suse.de>
2585 PR tree-optimization/83385
2586 * graphite-scop-detection.c (get_order, order): Remove.
2587 (bb_to_rpo): New global.
2589 (build_scops): Sort pbbs in RPO order.
2591 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2593 * combine.c (simplify_set): Do not transform subregs to zero_extends
2594 if the destination mode is a vector mode.
2596 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2598 PR tree-optimization/83359
2599 * tree-cfg.h (fold_loop_internal_call): Declare.
2600 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2601 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2602 (find_loop_dist_alias): New function.
2603 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2604 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2607 PR tree-optimization/80631
2608 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2609 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2610 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2611 hardcoding zero as the value if COND_EXPR is never true. For
2612 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2613 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2614 hardcoding MAX_EXPR as the reduction operation.
2615 (is_nonwrapping_integer_induction): Allow negative step.
2616 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2617 vect_create_epilog_for_reduction, if no value is suitable, don't
2618 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2620 2017-12-12 Richard Biener <rguenther@suse.de>
2622 PR tree-optimization/81889
2623 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2624 range info from the non-wrapping IV instead of just the range
2627 2017-12-12 Julia Koval <julia.koval@intel.com>
2629 * config.gcc: Add vaesintrin.h.
2630 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2632 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2633 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2635 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2636 * config/i386/immintrin.h: Include vaesintrin.h.
2637 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2638 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2639 _mm_aesdec_epi128): New intrinsics.
2641 2017-12-12 Julia Koval <julia.koval@intel.com>
2643 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2644 OPTION_MASK_ISA_VAES_UNSET): New.
2645 (ix86_handle_option): Handle -mvaes.
2646 * config/i386/cpuid.h: Define bit_VAES.
2647 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2648 * config/i386/i386-c.c (__VAES__): New.
2649 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2650 (ix86_valid_target_attribute_inner_p): Ditto.
2651 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2652 * config/i386/i386.opt: Add -mvaes.
2653 * doc/invoke.texi: Ditto.
2655 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2657 * debug.h (gcc_debug_hooks): Add inline_entry.
2658 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2659 * debug.c (do_nothing_debug_hooks): Likewise.
2660 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2661 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2662 (dwarf2_lineno_debug_hooks): Likewise.
2664 * common.opt (gstatement-frontiers): New, setting
2665 debug_nonbind_markers_p.
2666 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2667 * toplev.c (process_options): Autodetect value for debug statement
2669 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2670 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2672 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2673 markers. Integrate source bind into debug stmt expand loop.
2674 (pass_expand::execute): Check debug marker limit. Avoid deep
2675 TER and expand debug locations for debug bind insns only.
2676 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2678 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2679 * final.c (reemit_insn_block_notes): Take current block from
2680 nonbind markers. Declare note where it's first set.
2681 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2682 begin stmt markers if enabled.
2683 (notice_source_line): Handle nonbind markers. Fail if their
2684 location is unknown or that of builtins.
2685 (rest_of_handle_final): Convert begin stmt markers to notes if
2686 var-tracking didn't run.
2687 (rest_of_clean_state): Skip begin stmt markers.
2688 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2690 * function.c (allocate_struct_function): Set begin_stmt_markers.
2691 * function.h (struct function): Add debug_marker_count counter
2692 and debug_nonbind_markers flag.
2693 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2694 * gimple-low.c (lower_function_body): Adjust
2695 debug_nonbind_markers.
2696 (lower_stmt): Drop or skip gimple debug stmts.
2697 (lower_try_catch): Skip debug stmts.
2698 * gimple.c (gimple_build_debug_begin_stmt): New.
2699 (gimple_copy): Increment debug_marker_count if copying one.
2700 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2701 * gimplify.c (rexpr_location): New.
2702 (rexpr_has_location): New.
2703 (warn_switch_unreachable_r): Handle gimple debug stmts.
2704 (shortcut_cond_r): Call expr_location.
2706 (find_goto_label): New.
2707 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2709 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2711 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2712 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2713 (LANG_HOOKS_INITIALIZER): ... this.
2714 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2715 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2717 (update_ebb_live_info): Skip debug insn markers.
2718 * lra.c (debug_insn_static_data): Rename to...
2719 (debug_bind_static_data): ... this.
2720 (debug_marker_static_data): New.
2721 (lra_set_insn_recog_data): Select one of the above depending
2723 (lra_update_isn_regno_info): Don't assume debug insns have
2725 (push_insns): Skip debug insns.
2726 * lto-streamer-in.c (input_function): Drop debug stmts
2727 depending on active options. Adjust debug_nonbind_markers.
2728 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2729 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2730 begin stmt marker notes.
2731 (print_insn): Likewise.
2732 * recog.c (extract_insn): Recognize rtl for debug markers.
2733 * rtl.def (DEBUG_MARKER): New.
2734 * tree-inline.c: Include params.h.
2735 (remap_gimple_stmt): Handle nonbind markers.
2736 (maybe_move_debug_stmts_to_successors): Likewise.
2737 (copy_debug_stmt): Likewise.
2738 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2739 markers regardless of no side effects.
2740 (tsi_link_before): Don't update container's side effects when adding
2741 a begin stmt marker.
2742 (tsi_link_after): Likewise.
2743 (expr_first): Skip begin stmt markers.
2744 (expr_last): Likewise.
2745 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2746 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2747 Disregard nonbind markers.
2748 * tree.c (make_node_stat): Don't set side effects for begin stmt
2750 (build1_stat): Likewise.
2751 * tree.def (DEBUG_BEGIN_STMT): New.
2752 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2753 * var-tracking.c (delete_debug_insns): Renamed to...
2754 (delete_vta_debug_insns): ... this.
2755 (reemit_marker_as_note): New.
2756 (vt_initialize): Reemit markers.
2757 (delete_vta_debug_insns): Likewise.
2758 (vt_debug_insns_local): Reemit or delete markers.
2759 (variable_tracking_main_1): Likewise.
2760 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2761 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2762 (gimple_debug_nonbind_marker_p): New.
2763 (gimple_build_debug_bind): Adjust.
2764 (gimple_build_debug_begin_stmt): New.
2765 * doc/invoke.texi (max-debug-marker-count): New param.
2766 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2767 (debug_parameter_ref, debug_marker): New.
2768 (NOTE_INSN_BEGIN_STMT): New.
2769 (DEBUG_INSN): Describe begin stmt markers.
2771 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2772 without debug insns, we wouldn't, but clean up debug insns
2773 after a control flow insn nevertheless.
2775 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2776 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2777 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2778 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2779 (rtl_tidy_fallthru_edge): Likewise.
2780 (rtl_verify_fallthru): Likewise.
2781 (rtl_verify_bb_layout): Likewise.
2782 (skip_insns_after_block): Likewise.
2783 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2784 * dwarf2out.c: Include print-rtl.h.
2785 (dwarf2out_next_real_insn): New.
2786 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2787 Dump debug binds in asm comments.
2788 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2789 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2790 callers to use gsi_start_nondebug_bb instead.
2791 (gsi_after_labels): Skip gimple debug stmts.
2792 (gsi_start_nondebug): New.
2793 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2794 (proper_loop_form_for_interchange): Adjust.
2795 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2796 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2797 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2798 (collect_fallthrough_labels): Likewise.
2799 (should_warn_for_implicit_fallthrough): Likewise.
2800 (warn_implicit_fallthrough_r): Likewise.
2801 (expand_FALLTHROUGH_r): Likewise.
2802 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2803 (graphite_copy_stmts_from_block): Skip nonbind markers.
2804 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2805 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2806 * jump.c (clean_barriers): Skip debug insns.
2807 * omp-expand.c (expand_parallel_call): Skip debug insns.
2808 (expand_task_call): Likewise.
2809 (remove_exit_barrier): Likewise.
2810 (expand_omp_taskreg): Likewise.
2811 (expand_omp_for_init_counts): Likewise.
2812 (expand_omp_for_generic): Likewise.
2813 (expand_omp_for_static_nochunk): Likewise.
2814 (expand_omp_for_static_chunk): Likewise.
2815 (expand_omp_simd): Likewise.
2816 (expand_omp_taskloop_for_outer): Likewise.
2817 (expand_omp_taskloop_for_inner): Likewise.
2818 (expand_oacc_for): Likewise.
2819 (expand_omp_sections): Likewise.
2820 (expand_omp_single): Likewise.
2821 (expand_omp_synch): Likewise.
2822 (expand_omp_atomic_load): Likewise.
2823 (expand_omp_atomic_store): Likewise.
2824 (expand_omp_atomic_fetch_op): Likewise.
2825 (expand_omp_atomic_pipeline): Likewise.
2826 (expand_omp_atomic_mutex): Likewise.
2827 (expand_omp_target): Likewise.
2828 (grid_expand_omp_for_loop): Likewise.
2829 (grid_expand_target_grid_body): Likewise.
2830 (build_omp_regions_1): Likewise.
2831 * omp-low.c (check_combined_parallel): Skip debug stmts.
2832 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2833 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2834 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2836 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2837 (make_edges): Likewise.
2838 (cleanup_dead_labels): Likewise.
2839 (gimple_can_merge_blocks_p): Likewise.
2840 (stmt_starts_bb_p): Likewise.
2841 (gimple_block_label): Likewise.
2842 (gimple_redirect_edge_and_branch): Likewise.
2843 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2845 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2847 * tree-pretty-print (print_declaration): Omit initializer in slim
2849 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2851 (eliminate_unnecessary_stmts): Stabilize block removal order.
2852 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2853 * var-tracking.c (get_first_insn): New.
2854 (vt_emit_notes): Call it.
2855 (vt_initialize): Walk any insns before the first BB.
2856 (delete_debug_insns): Likewise.
2858 * gimple.h (enum gimple_debug_subcode): Add
2859 GIMPLE_DEBUG_BEGIN_STMT.
2860 (gimple_debug_begin_stmt_p): New.
2861 (gimple_debug_nonbind_marker_p): New.
2862 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2863 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2864 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2865 * insn-notes.def (BEGIN_STMT): New.
2866 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2867 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2868 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2869 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2870 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2871 (INSN_DEBUG_MARKER_KIND): New.
2872 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2873 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2874 (INSN_VAR_LOCATION_PTR): New.
2875 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2877 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2878 deps only if debug bind insns are enabled.
2879 (pass_expand::execute): Avoid deep TER and expand
2880 debug locations for debug bind insns only.
2881 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2882 debug stmts special handling down to debug bind stmts.
2883 * combine.c (try_combine): Narrow debug insns special handling
2884 down to debug bind insns.
2885 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2886 Narrow debug insns preexisting special handling down to debug
2888 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2889 handling down to debug bind insns.
2890 * function.c (instantiate_virtual_regs): Skip debug markers,
2891 adjust handling of debug binds.
2892 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2893 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2894 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2895 insns to debug bind insns.
2896 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2897 special handling of debug stmts to debug bind stmts.
2898 * ipa-split.c (split_function): Likewise.
2899 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2900 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2902 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2903 * regrename.c (build_def_use): Likewise.
2904 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2905 (pass_cprop_hardreg): Narrow special casing of debug insns to
2907 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2908 * reload1.c (reload): Likewise.
2909 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2910 casing of debug stmts to debug bind stmts.
2911 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2912 * ssa-iterators.h (num_imm_uses): Likewise.
2913 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2914 debug stmts to debug bind stmts.
2915 * tree-inline.c (tree_function_versioning): Narrow special casing
2916 of debug stmts to debug bind stmts.
2917 * tree-loop-distribution.c (generate_loops_for_partition):
2918 Narrow special casing of debug stmts to debug bind stmts.
2919 * tree-sra.c (analyze_access_subtree): Narrow special casing
2920 of debug stmts to debug bind stmts.
2921 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2922 stmts to debug bind stmts.
2923 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2924 casing of debug stmts to debug bind stmts.
2925 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2926 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2927 casing of debug stmts to debug bind stmts.
2928 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2930 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2931 debug stmts to debug bind stmts.
2932 (gimple_replace_ssa_lhs): Likewise.
2933 (insert_debug_temp_for_var_def): Likewise.
2934 (insert_debug_temps_for_defs): Likewise.
2935 (reset_debug_uses): Likewise.
2936 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2937 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2938 (adjust_debug_stmts): Likewise.
2939 (adjust_phi_and_debug_stmts): Likewise.
2940 (vect_do_peeling): Likewise.
2941 * tree-vect-loop.c (vect_transform_loop): Likewise.
2942 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2943 * var-tracking.c (adjust_mems): Narrow special casing of debug
2944 insns to debug bind insns.
2945 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2946 (compute_bb_dataflow, vt_find_locations): Likewise.
2947 (vt_expand_loc, emit_notes_for_changes): Likewise.
2948 (vt_init_cfa_base): Likewise.
2949 (vt_emit_notes): Likewise.
2950 (vt_initialize): Likewise.
2951 (vt_finalize): Likewise.
2953 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2954 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2955 (next_nonnote_nondebug_insn_bb): New.
2956 (prev_nonnote_nondebug_insn_bb): New.
2957 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2958 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2959 (prev_nonnote_nondebug_insn_bb): Declare.
2960 (next_nonnote_nondebug_insn_bb): Declare.
2961 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2962 * cfgrtl.c (get_last_bb_insn): Likewise.
2963 * lra.c (push_insns): Likewise.
2965 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2968 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2970 * selftest.c (selftest::test_runner::test_runner): New ctor.
2971 (selftest::test_runner::~test_runner): New dtor.
2972 * selftest.h (class selftest::test_runner): New class.
2974 2017-12-11 Carl Love <cel@us.ibm.com>
2976 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2977 vec_extract_fp32_from_shortl]): Add #defines.
2978 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2979 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2980 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2981 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2982 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2983 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2984 * doc/extend.texi: Add documentation for the added builtins.
2986 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2988 PR rtl-optimization/80693
2989 PR rtl-optimization/81019
2990 PR rtl-optimization/81020
2991 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2992 are not mentioned in i3. Place the REG_UNUSED note on i2,
2993 possibly modified to REG_DEAD, if it did not originate in i3.
2995 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2997 * recog.c (store_data_bypass_p_1): New function.
2998 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2999 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3001 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3003 PR rtl-optimization/83361
3004 * ifcvt.c (if_convert): Call fixup_partitions.
3006 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3008 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3009 early folding of splat_u{8,16,32}.
3011 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3013 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3014 output_operand_lossage first argument with capital letter.
3015 (aarch64_override_options): Don't start error and sorry first argument
3016 with capital letter.
3018 2017-12-11 Andi Kleen <ak@linux.intel.com>
3020 PR gcov-profile/83355
3021 * auto-profile.c (string_table::get_index_by_decl): Don't
3022 recurse when abstract origin points to itself.
3024 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3026 PR tree-optimization/83320
3027 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3028 (prune_datarefs_not_in_loop): Ditto.
3030 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3032 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3034 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3036 PR tree-optimization/83337
3037 * gimple-loop-interchange.cc (compute_access_stride): Handle
3038 bitfield DRs properly.
3040 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3042 PR tree-optimization/83338
3043 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3044 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3045 vector of unsigned integers to vector of signed integers.
3047 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3049 PR rtl-optimization/83317
3050 * lra-constraints.c (process_address_1): Add insn code check.
3052 2017-12-08 Michael Matz <matz@suse.de>
3054 Fix PR tree-optimization/83323
3055 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3056 head-controlled loops and loop BBs.
3057 * common.opt (funroll-and-jam): Remove, instead ...
3058 (floop-unroll-and-jam): ... reuse this option.
3059 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3060 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3061 (-floop-unroll-and-jam): ... this option.
3063 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3065 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3066 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3067 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3069 2017-12-08 Richard Biener <rguenther@suse.de>
3072 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3073 handle accesses outside of zero-sized vars.
3075 2017-12-08 Martin Jambor <mjambor@suse.cz>
3077 PR tree-optimization/83141
3078 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3079 test for MEM_REFs implicitely changing types with padding. Remove
3081 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3083 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3085 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3087 * config/avr/avr.c (avr_attribute_table): Likewise.
3088 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3089 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3090 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3091 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3092 of struct attribute_spec.
3093 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3095 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3098 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3100 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3103 * profile-count.c (profile_count::from_gcov_type): Move from
3104 profile-count.h; handle overflow.
3105 * profile-count.h (profile_count::from_gcov_type): Move offline.
3107 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3109 PR rtl-optimization/83304
3110 * combine.c (move_deaths): If we do not know where a register died,
3113 2017-12-08 Richard Biener <rguenther@suse.de>
3115 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3116 Provide -fopt-info-loop feedback when we interchange in a nest.
3118 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3120 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3121 for armv6 ARM CPU IDs.
3123 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3125 * common/config/arm/arm-common.c: Include <algorithm>.
3126 (INCLUDE_VECTOR): Define.
3127 (compare_opt_names): New function.
3128 (arm_rewrite_selected_arch): Only strip out extensions that can be
3129 expressed through -mfpu. Sort the remaining extensions
3132 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3134 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3135 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3136 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3137 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3139 2017-06-08 Tristan Gingold <gindold@adacore.com>
3142 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3143 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3145 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3147 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3148 of struct attribute_spec.
3150 2017-12-08 Julia Koval <julia.koval@intel.com>
3152 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3153 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3154 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3155 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3156 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3157 _mm_maskz_dpwssds_epi32): Ditto.
3159 2017-12-08 Richard Biener <rguenther@suse.de>
3161 PR tree-optimization/81303
3162 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3163 conditions try to create a comparison vector type matching
3164 the data vector type.
3165 (vectorizable_condition): Adjust.
3166 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3167 Leave invariant conditions alone in case we can vectorize those.
3169 2017-12-08 Julia Koval <julia.koval@intel.com>
3171 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3172 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3173 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3174 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3175 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3177 2017-12-08 Julia Koval <julia.koval@intel.com>
3179 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3180 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3181 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3182 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3183 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3184 _mm_maskz_dpbusds_epi32): New intrinsics.
3186 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3188 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3190 (ld<bh>io_signed): Likewise.
3191 (st<bhw>io): Likewise.
3192 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3193 SMALL_INT12 constant integer operand.
3195 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3197 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3199 (nios2_large_constant_allowed): ...this. Adjust uses.
3200 (nios2_plus_symbolic_constant_p): Rename to...
3201 (nios2_plus_large_constant_p): ...this. Adjust uses.
3202 (nios2_legitimate_address_p): Correct CONST_INT handling.
3203 (nios2_symbolic_memory_operand_p): Rename to...
3204 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3205 (nios2_large_constant_p): Check for large constant integers too.
3206 (nios2_split_large_constant): Handle constant integers.
3207 (nios2_split_symbolic_memory_operand): Rename to...
3208 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3209 (nios2_legitimize_constant_address): Handle constant integers.
3210 (r0rel_constant_p): Handle small constant integers.
3211 (nios2_print_operand_address): Handle r0-relative integer addresses.
3212 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3213 * config/nios2/nios2.md: Adjust for renamed functions.
3215 2017-12-07 Andrew Waterman <andrew@sifive.com>
3217 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3218 (TARGET_HAVE_SRODATA_SECTION): New define.
3219 (riscv_select_section): New function.
3221 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3222 Alexander Monakov <amonakov@ispras.ru>
3223 Jakub Jelinek <jakub@redhat.com>
3226 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3228 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3230 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3231 a bool success value. Don't call output_operand_lossage here.
3232 (aarch64_print_ldpstp_address): Return a bool success value.
3233 (aarch64_print_operand_address): Call output_addr_const if
3234 aarch64_print_address_internal fails.
3235 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3236 'y'; call output_operand_lossage instead. Call output_operand_lossage
3237 if aarch64_print_ldpstp_address fails.
3239 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3241 * tree-vector-builder.h
3242 (tree_vector_builder::binary_encoded_nelts): Declare.
3243 * tree-vector-builder.c
3244 (tree_vector_builder::binary_encoded_nelts): New function.
3245 * fold-const.c (negate_expr_p): Likewise.
3246 (operand_equal_p, fold_checksum_tree): Likewise.
3247 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3248 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3249 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3250 (uniform_vector_p): Likewise.
3251 * varasm.c (const_hash_1, compare_constant): Likewise.
3252 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3253 (valid_lattice_transition): Operate directly on the VECTOR_CST
3255 * ipa-icf.c: Include tree-vector-builder.h.
3256 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3257 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3259 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3261 * tree.c (build_vector): Delete.
3262 * tree.h (build_vector): Make static and move into the self-testing
3265 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3267 * vector-builder.h (vector_builder::derived): New const overload.
3268 (vector_builder::elt): New function.
3269 * tree-vector-builder.h (tree_vector_builder::type): New function.
3270 (tree_vector_builder::apply_step): Declare.
3271 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3273 * gimple-fold.h (tree_vector_builder): Declare.
3274 (gimple_build_vector): Take a tree_vector_builder instead of a
3275 type and vector of elements.
3276 * gimple-fold.c (gimple_build_vector): Likewise.
3277 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3279 (get_initial_defs_for_reduction): Likewise.
3280 (vectorizable_induction): Likewise.
3282 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3284 * tree-vector-builder.h
3285 (tree_vector_builder::new_binary_operation): Declare.
3286 * tree-vector-builder.c
3287 (tree_vector_builder::new_binary_operation): New function.
3288 * fold-const.c (fold_relational_const): Use it.
3289 (const_binop): Likewise. Check that both input vectors have
3290 the same number of elements, thus excluding things like WIDEN_SUM.
3291 Check whether it is possible to operate directly on the encodings
3294 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3296 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3297 new_unary_operation, operating only on the encoded elements.
3298 (const_unop): Likewise.
3299 (exact_inverse): Likewise.
3300 (distributes_over_addition_p): New function.
3301 (const_binop): Use tree_vector_builder and new_unary_operation
3302 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3303 on the encoded elements unless the encoding is strided and the
3304 operation does not distribute over addition.
3305 (fold_convert_const): Use tree_vector_builder and
3306 new_unary_operation. Operate only on the encoded elements
3307 for truncating integer conversions, or for non-stepped encodings.
3309 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3311 * config/sparc/sparc.c: Include tree-vector-builder.h.
3312 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3313 * expmed.c: Include tree-vector-builder.h.
3314 (make_tree): Use tree_vector_builder instead of build_vector.
3315 * fold-const.c: Include tree-vector-builder.h.
3316 (const_binop): Use tree_vector_builder instead of build_vector.
3317 (const_unop): Likewise.
3318 (native_interpret_vector): Likewise.
3319 (fold_vec_perm): Likewise.
3320 (fold_ternary_loc): Likewise.
3321 * gimple-fold.c: Include tree-vector-builder.h.
3322 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3324 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3325 (simplify_vector_constructor): Use tree_vector_builder instead
3327 * tree-vect-generic.c: Include tree-vector-builder.h.
3328 (add_rshift): Use tree_vector_builder instead of build_vector.
3329 (expand_vector_divmod): Likewise.
3330 (optimize_vector_constructor): Likewise.
3331 * tree-vect-loop.c: Include tree-vector-builder.h.
3332 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3333 of build_vector. Explicitly use a stepped encoding for
3335 * tree-vect-slp.c: Include tree-vector-builder.h.
3336 (vect_get_constant_vectors): Use tree_vector_builder instead
3338 (vect_transform_slp_perm_load): Likewise.
3339 (vect_schedule_slp_instance): Likewise.
3340 * tree-vect-stmts.c: Include tree-vector-builder.h.
3341 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3342 (vect_gen_perm_mask_any): Likewise.
3343 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3344 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3346 (build_vector_from_val): Likewise. Explicitly use a duplicate
3349 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3351 * doc/generic.texi (VECTOR_CST): Describe new representation of
3353 * vector-builder.h: New file.
3354 * tree-vector-builder.h: Likewise.
3355 * tree-vector-builder.c: Likewise.
3356 * Makefile.in (OBJS): Add tree-vector-builder.o.
3357 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3358 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3359 (tree_vector): Change the number of elements to
3360 vector_cst_encoded_nelts.
3361 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3362 (VECTOR_CST_ELTS): Delete.
3363 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3364 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3365 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3366 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3367 (VECTOR_CST_ENCODED_ELT): Likewise.
3368 (vector_cst_encoded_nelts): New function.
3369 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3370 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3371 (vector_cst_int_elt, vector_cst_elt): Declare.
3372 * tree.c: Include tree-vector-builder.h.
3373 (tree_code_size): Abort if passed VECTOR_CST.
3374 (tree_size): Update for new VECTOR_CST layout.
3375 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3376 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3377 (build_vector): Use tree_vector_builder.
3378 (vector_cst_int_elt, vector_cst_elt): New functions.
3379 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3380 encoded elements and then create the vector in the canonical form.
3381 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3382 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3383 (tree_c_tests): Call test_vector_cst_patterns.
3384 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3386 (hash_tree): Likewise.
3387 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3388 (streamer_write_tree_header): Likewise.
3389 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3390 (streamer_alloc_tree): Likewise. Update call to make_vector.
3391 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3393 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3395 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3396 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3397 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3399 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3400 Richard Biener <rguenther@suse.de>
3402 PR tree-optimization/81303
3403 * Makefile.in (gimple-loop-interchange.o): New object file.
3404 * common.opt (floop-interchange): Reuse the option from graphite.
3405 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3406 -floop-interchange and mention it for -O3.
3407 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3408 * gimple-loop-interchange.cc: New file.
3409 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3410 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3411 * passes.def (pass_linterchange): New pass.
3412 * timevar.def (TV_LINTERCHANGE): New time var.
3413 * tree-pass.h (make_pass_linterchange): New declaration.
3414 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3415 interchange. Record IV before/after increment in new parameters.
3416 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3417 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3419 (check_reduction_path): ...New function here.
3420 * tree-vectorizer.h (check_reduction_path): New declaration.
3422 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3425 PR rtl-optimization/80818
3426 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3427 always early clobbered.
3428 * lra-lives.c (process_bb_lives): Check input hard regs for early
3429 clobbered non-operand hard reg.
3431 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3434 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3435 types_compatible_p, just that TYPE_MODE is the same.
3437 2017-12-07 Martin Sebor <msebor@redhat.com>
3440 * attribs.c (empty_attribute_table): Initialize new member of
3441 struct attribute_spec.
3442 (decl_attributes): Add argument. Handle mutually exclusive
3443 combinations of attributes.
3444 (selftests::test_attribute_exclusions): New function.
3445 (selftests::attribute_c_tests): Ditto.
3446 * attribs.h (decl_attributes): Add default argument.
3447 * selftest.h (attribute_c_tests): Declare.
3448 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3449 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3451 * doc/extend.texi (Common Function Attributes): Update const and pure.
3452 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3453 of struct attribute_spec.
3454 * config/arc/arc.c (arc_attribute_table): Same.
3455 * config/arm/arm.c (arm_attribute_table): Same.
3456 * config/avr/avr.c ( avr_attribute_table): Same.
3457 * config/bfin/bfin.c (bfin_attribute_table): Same.
3458 * config/cr16/cr16.c (cr16_attribute_table): Same.
3459 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3460 * config/h8300/h8300.c (h8300_attribute_table): Same.
3461 * config/i386/i386.c (ix86_attribute_table): Same.
3462 * config/ia64/ia64.c (ia64_attribute_table): Same.
3463 * config/m32c/m32c.c (m32c_attribute_table): Same.
3464 * config/m32r/m32r.c (m32r_attribute_table): Same.
3465 * config/m68k/m68k.c (m68k_attribute_table): Same.
3466 * config/mcore/mcore.c (mcore_attribute_table): Same.
3467 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3468 * config/mips/mips.c (mips_attribute_table): Same.
3469 * config/msp430/msp430.c (msp430_attribute_table): Same.
3470 * config/nds32/nds32.c (nds32_attribute_table): Same.
3471 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3472 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3473 * config/rl78/rl78.c (rl78__attribute_table): Same.
3474 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3475 * onfig/rx/rx.c (rx_attribute_table): Same.
3476 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3477 * config/sh/sh.c (sh_attribute_table): Same.
3478 * config/sparc/sparc.c (sparc_attribute_table): Same.
3479 * config/spu/spu.c (spu_attribute_table): Same.
3480 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3481 * config/v850/v850.c (v850_attribute_table): Same.
3482 * config/visium/visium.c (visium_attribute_table): Same.
3484 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3487 * config/arm/arm.c (INCLUDE_STRING): Define.
3488 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3489 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3492 2017-12-07 Michael Matz <matz@suse.de>
3494 Add unroll and jam pass
3496 * gimple-loop-jam.c: New file.
3497 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3498 * common.opt (funroll-and-jam): New option.
3499 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3500 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3501 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3502 * passes.def: Add pass_loop_jam.
3503 * timevar.def (TV_LOOP_JAM): Add.
3504 * tree-pass.h (make_pass_loop_jam): Declare.
3505 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3506 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3507 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3508 to flow_loop_tree_node_add.
3509 (duplicate_subloops, copy_loops_to): Append to sibling list.
3510 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3511 * doc/invoke.texi (-funroll-and-jam): Document new option.
3512 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3514 2017-12-07 Richard Biener <rguenther@suse.de>
3516 PR tree-optimization/83296
3517 PR tree-optimization/67769
3518 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3519 flow sensitive info in an unrelated BB.
3520 (value_replacement): Use reset_flow_sensitive_info.
3521 (minmax_replacement): Reset flow sensitive info on the def
3522 we move. Do not reset flow sensitive info in the whole BB
3523 we move the stmt to.
3524 (abs_replacement): Likewise.
3526 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3529 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3530 rs6000_cpu to the given -mcpu=, or to the default processor.
3532 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3534 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3535 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3536 instead of rs6000_cpu_attr.
3537 (rs6000_adjust_cost): Ditto.
3538 (is_microcoded_insn): Ditto.
3539 (rs6000_adjust_priority): Ditto.
3540 (rs6000_issue_rate): Ditto.
3541 (rs6000_use_sched_lookahead): Ditto.
3542 (rs6000_use_sched_lookahead_guard): Ditto.
3543 (rs6000_sched_reorder): Ditto.
3544 (force_new_group): Ditto.
3545 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3546 (group_ending_nop): Ditto.
3548 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3550 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3551 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3552 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3553 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3555 (rs6000_reassociation_width): Ditto.
3556 (rs6000_emit_epilogue): Ditto.
3557 (rs6000_adjust_cost): Ditto.
3558 (is_microcoded_insn): Ditto.
3559 (is_cracked_insn): Ditto.
3560 (rs6000_adjust_priority): Ditto.
3561 (rs6000_sched_reorder): Ditto.
3562 (rs6000_sched_reorder2): Ditto.
3563 (insn_must_be_first_in_group): Ditto.
3564 (insn_must_be_last_in_group): Ditto.
3565 (rs6000_register_move_cost): Ditto.
3566 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3569 2017-12-07 Julia Koval <julia.koval@intel.com>
3571 * config.gcc: Add vaesintrin.h.
3572 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3574 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3575 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3577 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3578 * config/i386/immintrin.h: Include vaesintrin.h.
3579 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3580 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3581 _mm_aesdec_epi128): New intrinsics.
3583 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3585 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3586 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3587 c-family/c-spellcheck.cc.
3588 (best_macro_match::best_macro_match): Likewise.
3590 (struct edit_distance_traits<cpp_hashnode *>): Move to
3591 c-family/c-spellcheck.h.
3592 (class best_macro_match): Likewise.
3594 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3596 PR tree-optimization/83293
3597 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3598 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3599 might insert into empty bb.
3602 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3603 simplify for plus with :c added, and pointer_plus without that.
3604 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3605 with undefined overflow and the conversion is not widening,
3606 perform negation in utype and only convert to type afterwards.
3607 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3608 simplify for plus with :c added, and pointer_plus without that.
3609 If type is integral with undefined overflow and the conversion is
3610 not widening, perform minus in utype and only convert to type
3611 afterwards. Move the last pointer_diff_expr simplify into the
3614 2017-12-06 Martin Sebor <msebor@redhat.com>
3616 PR tree-optimization/82646
3617 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3618 strncpy, not maxlen.
3620 2017-12-06 Martin Sebor <msebor@redhat.com>
3622 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3625 PR tree-optimization/83075
3626 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3627 strncat/strncpy don't change length of source string.
3629 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3632 2017-11-29 Martin Aberg <maberg@gaisler.com>
3634 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3635 to prevent b2bst errata sequence.
3636 (sqrtdf2_fix): Likewise.
3638 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3640 PR tree-optimization/81945
3641 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3642 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3643 to dest_cfun has orig_loop_num set, either remap it to the new
3644 loop number if the loop got moved too, or clear it.
3646 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3648 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3650 (thunderx2t99_nothing): New insn reservation.
3651 (thunderx2t99_mrs): New insn reservation.
3652 (thunderx2t99_multiple): New insn reservation.
3653 (thunderx2t99_alu_basi): Add bfx to reservation.
3654 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3656 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3659 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3661 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3663 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3664 tree-ssa-pre.c::remove_dead_inserted_code.
3665 * tree-ssa-dce.h: New file.
3666 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3667 (remove_dead_inserted_code): Move and rename to function
3668 tree-ssa-dce.c::simple_dce_from_worklist.
3669 (pass_pre::execute): Update use.
3671 2017-12-05 Richard Biener <rguenther@suse.de>
3673 PR tree-optimization/83277
3674 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3675 to code-gen liveout vars.
3677 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3679 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3680 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3681 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3684 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3686 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3688 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3689 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3692 2017-12-05 Richard Biener <rguenther@suse.de>
3694 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3695 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3696 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3697 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3699 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3701 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3703 2017-12-05 Martin Liska <mliska@suse.cz>
3704 Jakub Jelinek <jakub@redhat.com>
3706 * doc/invoke.texi: Document the options.
3707 * flag-types.h (enum sanitize_code): Add
3708 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3709 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3710 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3711 * opts.c: Define new sanitizer options.
3712 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3713 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3715 2017-12-05 Julia Koval <julia.koval@intel.com>
3717 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3718 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3719 (ix86_handle_option): Handle -mavx512vnni.
3720 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3721 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3722 * config/i386/i386-c (__AVX512VNNI__): New.
3723 * config/i386/i386.c (ix86_target_string): Handle new option.
3724 (ix86_valid_target_attribute_inner_p): Handle new option.
3725 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3726 * config/i386/i386.opt (mavx512vnni): New option.
3728 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3731 * athlon.md: Disable for generic.
3732 * haswell.md: Enable for generic.
3733 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3734 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3736 (ix86_adjust_cost): Move generic to haswell path.
3738 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3740 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3741 instead of MEM_P in a couple more places. Fix formatting issues.
3743 2017-12-04 Jim Wilson <jimw@sifive.com>
3745 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3746 instead of GP_REG_LAST-1.
3747 (riscv_adjust_libcall_cfi_prologue): Likewise.
3748 (riscv_adjust_libcall_cri_epilogue): Likewise.
3749 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3752 2017-12-04 Luis Machado <luis.machado@linaro.org>
3754 * ipa-pure-const.c (check_decl): Add missing newline.
3755 (state_from_flags): Likewise.
3757 2017-12-04 Jeff Law <law@redhat.com>
3759 PR tree-optimizatin/78496
3760 * gimple-ssa-evrp-analyze.h
3761 (evrp_range_analyzer::get_vr_values): Simplify.
3762 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3763 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3764 and gimple-ssa-evrp-analyze.h.
3765 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3766 (simplify_stmt_for_jump_threading): Copy a blob of code from
3767 tree-vrp.c to use ranges to simplify statements.
3768 (dom_opt_dom_walker::before_dom_children): Call
3769 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3770 (dom_opt_dom_walker::after_dom_children): Similarly for
3771 evrp_range_analyzer::leave.
3772 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3775 * gimple-ssa-evrp-analyze.c
3776 (evrp_range_analyzer::extract_range_from_stmt): Always use
3777 vr_values::update_value_range so preexisting range info is
3778 medged with new range info, even if the new range is VR_VARYING.
3780 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3782 * combine.c: Adjust comment.
3783 (use_crosses_set_p): Delete.
3784 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3785 (try_combine): Ditto.
3787 2017-12-04 Richard Biener <rguenther@suse.de>
3789 PR tree-optimization/83255
3790 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3791 Re-add zero-iteration check.
3793 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3795 PR rtl-optimization/83245
3796 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3797 hard registers as earlyclobber, also if not in an asm.
3799 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3803 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3806 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3807 rs6000_cpu based on cpu_index, not tune_index.
3809 2017-12-04 Richard Biener <rguenther@suse.de>
3811 PR tree-optimization/83238
3812 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3813 code match comment, rejecting invalid SESE regions.
3815 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3817 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3818 require base operand is a REG_POINTER prior to reload on targets
3819 with non-equivalent space registers.
3821 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3823 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3824 (update_specialized_profile): Fix updating of counts.
3825 (perhaps_add_new_callers): Likewise.
3827 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3830 * x86-tune.def: Remove obsolette FIXMEs.
3831 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3832 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3833 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3835 (X86_TUNE_PAD_RETURNS): Disable for generic.
3837 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3839 PR tree-optimization/83170
3840 PR tree-optimization/83241
3841 * gimple-ssa-store-merging.c
3842 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3843 gimple_vuse (ins_stmt) in case it has changed.
3844 (imm_store_chain_info::output_merged_store): Likewise.
3846 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3850 * tree-cfg.c (pass_warn_function_return::execute): Handle
3851 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3852 with BUILTINS_LOCATION.
3856 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3857 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3858 instead of DECL_MODE.
3860 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3861 last argument to rtx pointer.
3862 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3863 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3864 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3865 depending on the chosen ISAs.
3866 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3867 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3868 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3870 * config/i386/sse.md (mov<mode>_internal): Likewise.
3871 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3873 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3875 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3877 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3879 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3880 parameters from prototype.
3881 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3882 parameters. Don't print a comment.
3883 (emit_fusion_gpr_load): Adjust.
3884 (emit_fusion_load_store): Adjust.
3885 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3886 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3887 comment on the second line.
3889 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3892 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3893 rs6000_cpu based on cpu_index, not tune_index.
3895 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3897 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3898 which_alternative instead of which_alternative + 1.
3899 (output_asm_insn): Print an extra tab if the template is short.
3901 2017-12-01 Jim Wilson <jimw@sifive.com>
3903 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3905 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3906 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3908 * doc/tm.texi: Regenerate.
3910 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3913 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3914 whether we can allocate pseudos before trying to fix an address.
3915 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3916 memory address is indexed or indirect.
3917 (floatuns_<mode>si2_hw2): Likewise.
3919 2017-12-01 Jason Merrill <jason@redhat.com>
3921 * Makefile.in (TAGS): Add c-family/*.cc.
3923 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3925 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3926 (call_value_insn): Likewise.
3927 (sibcall_insn): Likewise.
3928 (sibcall_value_insn): Likewise.
3929 (movsi_aarch64): Likewise.
3930 (movdi_aarch64): Likewise.
3931 (add_losym_): Likewise.
3932 (ldr_got_small_): Likewise.
3933 (ldr_got_small_sidi): Likewise.
3934 (ldr_got_small_28k_): Likewise.
3935 (ldr_got_small_28k_sidi): Likewise.
3936 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3937 Move output_addr_const to symbolic case. Add error check.
3939 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3941 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3942 * config/s390/s390.md: Disable MVC merging peephole if it would
3943 disable operand forwarding.
3944 (new peephole2): Split MVCs if it would turn them into up to 2
3947 2017-12-01 Richard Biener <rguenther@suse.de>
3949 PR tree-optimization/83232
3950 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3951 detection of same access. Instead of breaking the group here
3952 do not consider the duplicate. Add comment explaining real fix.
3954 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3956 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3958 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3960 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3962 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3964 * function.h (struct function): Remove cilk_frame_decl,
3965 is_cilk_function and calls_cilk_spawn fields.
3966 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3968 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3969 cilk_elemental field.
3970 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3971 * target.def: Adjust comment.
3972 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3973 Don't test cilk_elemental.
3975 PR tree-optimization/83233
3976 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3977 bswap_stat name for the struct.
3980 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3981 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3982 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3983 SWITCH_BREAK_LABEL_P set on the label.
3984 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3985 added for default case if it was missing and not all cases covered.
3986 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3987 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3989 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3990 to avoid -Wimplicit-fallthrough warning.
3991 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3992 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3994 PR tree-optimization/83221
3995 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3997 (init_reassoc): Formatting fix.
4000 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4001 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4005 * tree-cfg.c: Include asan.h.
4006 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4008 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4010 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4012 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4014 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4015 vpcmpeqd instruction.
4017 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4019 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4020 registers type for 128bit mode.
4022 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4024 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4026 * gengtype-state.c (read_state_pair): Likewise.
4027 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4028 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4029 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4030 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4031 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4032 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4033 * builtins.c (fold_builtin_3): Likewise.
4034 * graphite-scop-detection.c
4035 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4036 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4039 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4042 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4043 parameter. Return true if ignore_decl_rtl_set_p is true and other
4044 conditions are satisfied.
4045 * asan.h (asan_protect_global): Add new parameter.
4046 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4047 to asan_protect_global calls.
4049 2017-11-30 Jim Wilson <jimw@sifive.com>
4051 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4052 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4053 -mstrict-align, add info on default value. Delete redundant lines for
4054 -mabi. Add missing -mexplicit-relocs docs.
4056 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4058 * config/arc/arc.md (trap): New pattern.
4060 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4062 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4063 instruction to end into a delay slot.
4064 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4065 can be placed into a delay slot against reg_note.
4067 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4069 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4070 labels number of usages.
4072 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4074 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4076 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4078 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4081 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4082 multiplication by power of 2 constant into two shifts + comparison.
4084 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4087 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4089 2017-11-30 Richard Biener <rguenther@suse.de>
4091 PR tree-optimization/83202
4092 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4093 allow_peel argument and guard peeling.
4094 (canonicalize_loop_induction_variables): Likewise.
4095 (canonicalize_induction_variables): Pass false.
4096 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4097 peeling from cunrolli.
4099 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4101 * combine.c (try_combine): Print a message to dump file whenever
4102 I0, I1, or I2 cannot be combined into I3.
4104 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4106 PR rtl-optimization/83156
4107 PR rtl-optimization/82621
4108 * combine.c (try_combine): Don't split an I2 if one of the dests is
4109 set again before I3. Allow unused dests.
4111 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4113 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4115 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4117 PR rtl-optimization/80818
4118 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4119 recursively. Use insn code for clobber.
4120 (lra_set_insn_recog_data): Pass the new arg to
4121 collect_non_operand_hard_regs.
4122 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4124 (lra_update_insn_regno_info): Pass insn to
4125 add_regs_to_insn_regno_info.
4127 2017-11-29 Jim Wilson <jimw@sifive.com>
4128 Andrew Waterman <andrew@sifive.com>
4130 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4131 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4132 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4133 use SINGLE_SHIFT_COST.
4134 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4135 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4137 2017-11-29 Julia Koval <julia.koval@intel.com>
4139 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4140 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4141 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4142 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4143 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4144 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4145 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4146 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4147 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4148 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4149 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4150 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4151 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4152 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4153 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4154 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4155 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4156 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4157 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4158 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4159 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4160 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4161 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4162 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4163 __builtin_ia32_vpshldv_v2di_mask,
4164 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4165 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4166 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4168 2017-11-29 Julia Koval <julia.koval@intel.com>
4170 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4171 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4172 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4173 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4174 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4175 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4176 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4177 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4178 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4179 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4180 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4181 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4182 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4183 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4184 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4185 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4186 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4187 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4188 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4189 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4190 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4192 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4194 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4195 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4198 2017-11-29 Martin Aberg <maberg@gaisler.com>
4200 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4201 to prevent b2bst errata sequence.
4202 (sqrtdf2_fix): Likewise.
4204 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4206 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4207 (div_sqrt_insn_p): New function.
4208 (sparc_do_work_around_errata): Insert NOP instructions to
4209 prevent sequences that could trigger the TN-0013 errata for
4210 certain LEON3 processors.
4211 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4212 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4213 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4214 (in_branch_delay): Prevent div and sqrt in delay slot if
4216 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4218 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4220 * config/sparc/sparc.c (atomic_insn_p): New function.
4221 (sparc_do_work_around_errata): Insert NOP instructions to
4222 prevent sequences that could trigger the TN-0010 errata for
4224 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4225 instruction referable in atomic_insns_p.
4227 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4229 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4230 (atomic_compare_and_swap_leon3_1): Likewise.
4233 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4235 * config/sparc/sparc.c (fpop_insn_p): New function.
4236 (sparc_do_work_around_errata): Insert NOP instructions to
4237 prevent sequences that could trigger the TN-0012 errata for
4239 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4240 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4241 (in_branch_annul_delay): Prevent floating-point instructions
4242 in delay slot of annulled integer branch.
4244 2017-11-29 Richard Biener <rguenther@suse.de>
4246 PR tree-optimization/83202
4247 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4249 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4250 nodes vectorized to the same stmts multiple times.
4251 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4252 (vect_analyze_slp_instance): Adjust.
4253 (scalar_stmts_to_slp_tree_map_t): New typedef.
4254 (vect_schedule_slp_instance): Add a map recording the SLP node
4255 representing the vectorized stmts for a set of scalar stmts.
4256 Avoid code-generating redundancies.
4257 (vect_schedule_slp): Allocate map and pass it down.
4259 2017-11-29 Nathan Sidwell <nathan@acm.org>
4262 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4265 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4267 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4268 cases for printing LDP/STP memory addresses.
4269 (aarch64_print_address_internal): Renamed from
4270 aarch64_print_operand_address, added parameter, add Pmode check.
4271 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4272 (aarch64_print_operand_address): Indirect to
4273 aarch64_print_address_internal.
4274 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4275 'y' operand output specifier.
4277 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4280 * tree.c (build_simple_mem_ref_loc): Handle
4281 get_addr_base_and_unit_offset returning a MEM_REF.
4284 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4287 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4290 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4292 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4294 (layout::move_to_column): Likewise.
4296 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4299 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4300 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4302 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4303 Martin Jambor <mjambor@suse.cz>
4306 * tree.h (expr_type_first_operand_type_p): Declare
4307 * tree.c (expr_type_first_operand_type_p): New function.
4308 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4309 (ipa_value_from_jfunc): Adjust declaration.
4310 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4311 Use it as result type for arithmetics, unless it is NULL in which case
4312 be more conservative.
4313 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4314 ipa_get_jf_pass_through_result.
4315 (propagate_vals_across_pass_through): Likewise.
4316 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4317 is to propagate_vals_across_pass_through.
4318 (propagate_constants_across_call): Pass PARM_TYPE to
4319 propagate_scalar_across_jump_function.
4320 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4321 ipa_value_from_jfunc.
4322 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4323 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4324 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4325 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4326 target_type, pass it to ipa_value_from_jfunc.
4327 (update_indirect_edges_after_inlining): Pass parameter type to
4328 try_make_edge_direct_simple_call.
4330 2017-11-28 Jeff Law <law@redhat.com>
4332 * gimple-ssa-evrp-analyze.c
4333 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4334 refine ranges if scev_initialized_p returns true.
4335 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4337 2017-11-28 Julia Koval <julia.koval@intel.com>
4339 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4340 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4341 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4342 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4343 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4344 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4345 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4346 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4347 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4348 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4349 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4350 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4351 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4352 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4353 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4354 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4355 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4356 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4357 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4358 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4359 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4360 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4362 2017-11-28 Julia Koval <julia.koval@intel.com>
4364 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4365 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4366 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4367 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4368 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4369 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4370 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4371 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4372 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4373 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4374 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4375 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4376 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4377 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4378 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4379 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4380 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4381 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4382 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4383 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4384 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4385 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4386 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4387 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4388 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4389 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4390 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4391 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4392 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4393 __builtin_ia32_vpshld_v2di_mask): New builtins.
4394 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4395 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4397 2017-11-28 Richard Biener <rguenther@suse.de>
4399 PR tree-optimization/80776
4400 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4402 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4404 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4405 If the incoming edge is an effective fallthru because the other
4406 edge only reaches a __builtin_unreachable () then record ranges
4407 derived from the controlling condition in SSA info.
4408 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4409 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4411 2017-11-28 Olivier Hainque <hainque@adacore.com>
4413 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4414 on mingw build hosts.
4416 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4419 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4421 2017-11-28 Richard Biener <rguenther@suse.de>
4424 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4425 copies generated from memcpy use a character array as reference
4428 2017-11-28 Julia Koval <julia.koval@intel.com>
4429 Sebastian Peryt <sebastian.peryt@intel.com>
4431 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4432 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4433 cilk-common.o, cilk.h, cilk-common.c): Remove.
4435 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4436 * builtins.c (is_builtin_name): Remove cilkplus condition.
4437 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4438 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4439 cilk-builtins.def, cilkplus.def): Remove.
4440 * cif-code.def (CILK_SPAWN): Remove.
4441 * cilk-builtins.def: Delete.
4442 * cilk-common.c: Ditto.
4444 * cilkplus.def: Ditto.
4445 * config/darwin.h (fcilkplus): Delete.
4446 * cppbuiltin.c: Ditto.
4447 * doc/extend.texi: Remove cilkplus doc.
4448 * doc/generic.texi: Ditto.
4449 * doc/invoke.texi: Ditto.
4450 * doc/passes.texi: Ditto.
4451 * gcc.c (fcilkplus): Remove.
4452 * gengtype.c (cilk.h): Remove.
4453 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4455 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4457 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4459 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4460 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4461 cilkplus conditions.
4462 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4463 inline_read_section): Ditto.
4464 * ipa-inline-analysis.c (cilk.h): Remove.
4465 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4466 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4467 append_linker_options): Remove condition for fcilkplus.
4468 * lto/lto-lang.c (cilk.h): Remove.
4469 (lto_init): Remove condition for fcilkplus.
4470 * omp-expand.c (expand_cilk_for_call): Delete.
4471 (expand_omp_taskreg, expand_omp_for_static_chunk,
4472 expand_omp_for): Remove cilkplus
4474 (expand_cilk_for): Delete.
4475 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4476 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4477 execute_lower_omp, diagnose_sb_0): Ditto.
4478 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4479 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4480 * tree-nested.c: Ditto.
4481 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4482 (dump_generic_node): Ditto.
4483 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4484 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4486 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4488 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4490 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4491 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4492 for vector mode and !TARGET_SIMD.
4494 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4496 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4498 * tree.h (SWITCH_LABELS): Remove.
4499 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4500 assert SWITCH_BODY is non-NULL.
4501 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4503 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4505 PR tree-optimization/80788
4506 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4507 has TREE_OVERFLOW set, call drop_tree_overflow.
4509 2017-11-28 Richard Biener <rguenther@suse.de>
4511 PR tree-optimization/83158
4512 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4514 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4517 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4518 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4520 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4522 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4523 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4525 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4528 * varasm.c (compare_constant): Take the mode of the constants into
4529 account when comparing floating point constants.
4531 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4533 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4534 from explicit instantiation of debug_helper.
4535 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4537 2017-11-27 Richard Biener <rguenther@suse.de>
4539 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4542 2017-11-27 Richard Biener <rguenther@suse.de>
4544 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4545 (type_hash_canon): Also clear int_cst_hash_table entry for
4547 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4549 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4551 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4552 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4553 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4555 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4557 * hash-map.h (gt_cleare_cache): Avoid UB.
4559 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4561 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4562 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4563 (decide_unroll_runtime_iterations): Likewise.
4564 (decide_unroll_stupid): Likewise.
4566 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4569 * config/i386/i386.md: Add a loop with incssp.
4571 2017-11-27 Martin Jambor <mjambor@suse.cz>
4573 PR tree-optimization/81248
4574 * tree-sra.c (splice_param_accesses): Remove size check.
4575 (decide_one_param_reduction): Fix size check.
4576 * gimple-pretty-print.c (dump_profile): Silence warning.
4577 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4579 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4582 * dbxout.c (lastlineno): New variable.
4583 (dbx_debug_hooks): Use dbxout_switch_text_section as
4584 switch_text_section debug hook.
4585 (dbxout_function_end): Switch to current_function_section
4586 rather than function_section. If crtl->has_bb_partition,
4587 output just one N_FUN, depending on in_cold_section_p.
4588 (dbxout_source_line): Remember last lineno in lastlineno.
4589 (dbxout_switch_text_section): New function.
4590 (dbxout_function_decl): Adjust dbxout_block caller.
4591 (dbx_block_with_cold_children): New function.
4592 (dbxout_block): Return true if any LBRAC/RBRAC have been
4593 emitted. Use dbx_block_with_cold_children at depth == 0
4594 in second partition. Add PARENT_BLOCKNUM argument, pass
4595 it optionally adjusted to children. Output LBRAC/RBRAC
4596 around recursive call only if the block is in the current
4597 partition, if not and anything was output, emit empty
4599 * final.c (final_scan_insn): Compute cold_function_name
4600 before calling switch_text_section debug hook. Call
4601 that hook even if dwarf2out_do_frame if not emitting
4605 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4606 TREE_READONLY decls.
4608 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4610 PR rtl-optimization/82488
4611 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4613 2017-11-26 Julia Koval <julia.koval@intel.com>
4615 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4617 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4620 2017-11-26 Julia Koval <julia.koval@intel.com>
4622 * config/i386/driver-i386.c (host_detect_local_cpu):
4623 Detect skylake-avx512.
4625 2017-11-26 Julia Koval <julia.koval@intel.com>
4627 * config.gcc: Add -march=cannonlake.
4628 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4629 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4630 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4631 (PTA_CANNONLAKE): New.
4632 (processor_target_table): Add cannonlake.
4633 (ix86_option_override_internal): Ditto.
4634 (fold_builtin_cpu): Ditto.
4635 (get_builtin_code_for_version): Handle cannonlake.
4636 (M_INTEL_COREI7_CANNONLAKE): New.
4637 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4638 * doc/invoke.texi: Add -march=cannonlake.
4640 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4642 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4643 (try_init_one_plugin): Alternative implementation for MinGW.
4644 * Makefile.in (plugin_implib): New.
4645 (gengtype-lex.c): Fix broken AIX workaround.
4646 * configure: Regenerate.
4647 * doc/plugins.texi: Document support for MinGW.
4649 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4651 PR rtl-optimization/81553
4652 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4653 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4654 is a shift where C1 has different mode than the whole shift, use C1's
4655 mode for MULT rather than the shift's mode.
4658 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4659 builtins not enabled in the currently selected ISA.
4661 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4663 PR tree-optimization/71026
4664 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4665 (insert_reciprocals): Change to insert reciprocals before a division
4666 by a square and to insert the square of a reciprocal.
4667 (execute_cse_reciprocals_1): Change to consider division by a square.
4668 (register_division_in): Add importance parameter.
4670 2017-11-24 Richard Biener <rguenther@suse.de>
4672 PR tree-optimization/82402
4673 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4674 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4676 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4678 * match.pd (0-ptr): New transformation.
4680 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4683 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4684 when skipping checking.
4686 2017-11-24 Richard Biener <rguenther@suse.de>
4688 PR tree-optimization/83128
4689 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4690 (vn_reference_lookup_3): Likewise.
4692 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4695 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4696 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4699 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4700 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4701 check for the latter with gimple_call_builtin_p. Do not handle
4702 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4704 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4706 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4709 2017-11-23 Julia Koval <julia.koval@intel.com>
4711 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4712 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4713 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4714 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4715 _mm512_maskz_expandloadu_epi16): New intrinsics.
4716 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4717 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4718 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4719 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4720 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4721 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4722 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4723 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4724 _mm256_maskz_expandloadu_epi8): New intrinsics.
4725 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4726 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4727 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4728 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4729 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4730 * config/i386/sse.md (VI248_VLBW): New iterator.
4731 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4733 2017-11-23 Julia Koval <julia.koval@intel.com>
4735 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4736 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4737 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4738 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4739 _mm512_mask_compressstoreu_epi16): New.
4740 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4741 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4742 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4743 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4744 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4745 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4746 _mm256_mask_compressstoreu_epi8): New.
4747 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4748 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4749 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4750 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4751 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4752 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4753 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4754 __builtin_ia32_compresshi128_mask,
4755 __builtin_ia32_compressstoreuqi512_mask,
4756 __builtin_ia32_compressstoreuhi512_mask,
4757 __builtin_ia32_compressstoreuqi256_mask,
4758 __builtin_ia32_compressstoreuqi128_mask,
4759 __builtin_ia32_compressstoreuhi256_mask,
4760 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4761 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4763 (ix86_expand_special_args_builtin): Handle new types.
4764 (s4fma_expand): Handle new builtin array.
4765 * config/i386/immintrin.h: Include new headers.
4766 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4767 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4769 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4771 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4772 fixes. Declare temp and g variables at the top in order to avoid
4773 {} in most of the cases.
4775 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4777 * match.pd (ptr-0): New transformation.
4779 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4781 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4782 (struct addr_mode_cost_table): New.
4783 (struct tune_params): Add field addr_mode_costs.
4784 * config/arm/arm.c (generic_addr_mode_costs): New.
4785 (arm_slowmul_tune): Initialise addr_mode_costs field.
4786 (arm_fastmul_tune): Likewise.
4787 (arm_strongarm_tune): Likewise.
4788 (arm_xscale_tune): Likewise.
4789 (arm_9e_tune): Likewise.
4790 (arm_marvell_pj4_tune): Likewise.
4791 (arm_v6t2_tune): Likewise.
4792 (arm_cortex_tune): Likewise.
4793 (arm_cortex_a8_tune): Likewise.
4794 (arm_cortex_a7_tune): Likewise.
4795 (arm_cortex_a15_tune): Likewise.
4796 (arm_cortex_a35_tune): Likewise.
4797 (arm_cortex_a53_tune): Likewise.
4798 (arm_cortex_a57_tune): Likewise.
4799 (arm_exynosm1_tune): Likewise.
4800 (arm_xgene1_tune): Likewise.
4801 (arm_cortex_a5_tune): Likewise.
4802 (arm_cortex_a9_tune): Likewise.
4803 (arm_cortex_a12_tune): Likewise.
4804 (arm_cortex_a73_tune): Likewise.
4805 (arm_v7m_tune): Likewise.
4806 (arm_cortex_m7_tune): Likewise.
4807 (arm_v6m_tune): Likewise.
4808 (arm_fa726te_tune): Likewise.
4809 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4811 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4813 * config/arm/arm.c (arm_mem_costs): New function.
4814 (arm_rtx_costs_internal): Use arm_mem_costs.
4816 2017-11-23 Mark Wielaard <mark@klomp.org>
4818 * dwarf2out.c (init_sections_and_labels): Use generation to create
4819 unique ranges_section_label and ranges_base_label. Return generation.
4820 (output_rnglists): Add generation argument. Use generation to create
4821 unique ranges labels.
4822 (dwarf2out_finish): Get generation from init_sections_and_labels
4823 and pass generation to output_rnglists.
4825 2017-11-23 Mike Stump <mikestump@comcast.net>
4826 Eric Botcazou <ebotcazou@adacore.com>
4828 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4829 * cfgloop.h (struct loop): Add unroll field.
4830 * function.h (struct function): Add has_unroll bitfield.
4831 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4832 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4833 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4834 (pass_rtl_unroll_loops::gate): Likewise.
4835 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4836 for which loop->unroll==1.
4837 (decide_unroll_constant_iterations): Use note for consistency and
4838 take loop->unroll into account. Return early if loop->unroll is set.
4839 Fix thinko in existing test.
4840 (decide_unroll_runtime_iterations): Use note for consistency and
4841 take loop->unroll into account.
4842 (decide_unroll_stupid): Likewise.
4843 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4844 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4845 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4847 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4848 (print_loop): Print loop->unroll if set.
4849 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4850 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4851 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4853 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4854 loop->unroll is set and smaller than the trip count. Otherwise bypass
4855 entirely the heuristics if loop->unroll is set. Remove dead note.
4856 Fix off-by-one bug in other note.
4857 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4858 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4860 (tree_unroll_loops_completely): Make static.
4861 (pass_complete_unroll::execute): Use correct type for variable.
4862 (pass_complete_unrolli::execute): Fix formatting.
4863 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4865 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4867 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4868 enable when TARGET_PREFER_AVX128 is set.
4870 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4872 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4874 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4875 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4877 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4879 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4880 frequencies. Fix estimation of aggregate parameters.
4882 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4884 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4887 2017-11-23 Tom de Vries <tom@codesourcery.com>
4889 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4891 2017-11-23 Tom de Vries <tom@codesourcery.com>
4893 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4896 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4899 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4900 sibcall_value_pcrel_fdpic): Use local variable instead of
4902 (calli_tbr_rel): Add missing operand 2.
4903 (call_valuei_tbr_rel): Add missing operand 3.
4905 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4908 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4909 bitpos/bitsize covering the whole destination, use store_expr only if
4910 the complex mode is the same. Otherwise, use expand_normal and if
4911 it returns CONCAT, subreg each part separately instead of trying to
4912 subreg the whole result.
4914 2017-11-23 Richard Biener <rguenther@suse.de>
4916 PR tree-optimization/23094
4917 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4918 come here from walking over backedges in the first iteration.
4919 (vn_reference_lookup_3): Skip clobbers that store the same value.
4921 2017-11-23 Richard Biener <rguenther@suse.de>
4923 PR tree-optimization/81403
4924 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4925 a block we need a leader relative to.
4926 (phi_translate_1): For nary processing require a leader from
4927 get_representative_for given we run expression simplification
4928 using match-and-simplify. Remove previous fix.
4930 2017-11-22 Jeff Law <law@redhat.com>
4932 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4933 Use new method allocate_value_range rather than accessing the
4934 vrp_value_range_pool data member directly.
4935 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4936 to use extract_range_from_stmt method to avoid need for
4937 extract_range_from_assignment method.
4938 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4939 method rather than setting values_propgated data member directly.
4940 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4941 and values propagated data members and extract_range_from_assignment
4942 method. Reorder private data members to conform to standards.
4943 Add new methods set_lattice_propagation_complete and
4944 allocate_value_range.
4946 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4948 PR rtl-optimization/83030
4949 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4950 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4951 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4952 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4953 CROSSING_JUMP_P flag is already set before setting it.
4954 * cfgrtl.c (fixup_partition_crossing): Likewise.
4955 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4958 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4960 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4961 Handle the case where both arguments are using gen_const_vec_series.
4963 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4966 * pretty-print.c (pp_format): Move quoting implementation to
4967 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4968 to pass address of "quote" local.
4969 (pp_begin_quote): New function.
4970 (pp_end_quote): New function.
4971 * pretty-print.h (printer_fn): Convert penultimate param from bool
4973 (pp_begin_quote): New decl.
4974 (pp_end_quote): New decl.
4975 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4976 param from bool to bool *.
4977 * tree-diagnostic.h (default_tree_printer): Likewise.
4979 2017-11-22 Jeff Law <law@redhat.com>
4981 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4982 Initialize vr_values.
4983 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4984 vr_values via vr_values class instance rather than delegators.
4985 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4986 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4987 (evrp_range_analyzer::push_value_range): Likewise.
4988 (evrp_range_analyzer::pop_value_range): Likewise.
4989 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4990 most delegators. Those remaining are exposed as public interfaces.
4991 Make vr_values a pointer and private.
4992 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4994 (evrp_range_analyzer::get_vr_value): New method.
4995 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4996 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4997 (class evrp_dom_walker): Attach evrp_folder class, initialize
4998 it in the ctor. Remove temporary delegators.
4999 (evrp_dom_walker::before_dom_children): Call methods in attached
5000 evrp_range_analyzer class via class instance pointer. Use
5001 free value_range_constant_singleton to remove need for
5002 op_with_constant_singleton_value delegator method. Do not
5003 create a vrp_prop class instance for every call! Narrow
5004 scope of a couple variables.
5005 (evrp_dom_walker::cleanup): Call methods in attached
5006 evrp_range_analyzer class via class instance pointer.
5007 * vr-values.h (class vr_values): Privatize many methods and
5010 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5012 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5013 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5014 * expr.c (expand_expr_real_2): Likewise.
5015 * fold-const.c (const_unop): Likewise.
5016 * optabs-tree.c (optab_for_tree_code): Likewise.
5017 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5018 * tree-inline.c (estimate_operator_cost): Likewise.
5019 * tree-pretty-print.c (dump_generic_node): Likewise.
5020 (op_code_prio): Likewise.
5021 (op_symbol_code): Likewise.
5022 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5023 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5024 * internal-fn.c (direct_internal_fn_optab): New function.
5025 (direct_internal_fn_array, direct_internal_fn_supported_p
5026 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5027 * fold-const-call.c (fold_const_reduction): New function.
5028 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5030 * tree-vect-loop.c: Include internal-fn.h.
5031 (reduction_code_for_scalar_code): Rename to...
5032 (reduction_fn_for_scalar_code): ...this and return an internal
5034 (vect_model_reduction_cost): Take an internal_fn rather than
5036 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5038 (vectorizable_reduction): Use internal functions rather than tree
5039 codes for the reduction operation. Update calls to the functions
5041 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5042 Use calls to internal functions rather than REDUC tree codes.
5043 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5045 2017-11-22 Olivier Hainque <hainque@adacore.com>
5047 * config/vxworks.c (vxworks_override_options): Pick default
5048 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5049 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5050 DWARF_GNAT_ENCODINGS_DEFAULT.
5051 * config/vxworksae.h: Likewise.
5053 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5055 PR tree-optimization/83104
5056 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5059 2017-11-22 Marek Polacek <polacek@redhat.com>
5060 H.J. Lu <hongjiu.lu@intel.com>
5061 Jason Merrill <jason@redhat.com>
5066 * calls.c (initialize_argument_information): Call
5067 warn_parameter_passing_abi target hook.
5068 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5069 argument onto stack.
5070 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5071 * common.opt: Update -fabi-version description.
5072 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5073 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5075 (ix86_is_empty_record): New function.
5076 (ix86_warn_parameter_passing_abi): New function.
5077 (TARGET_EMPTY_RECORD_P): Redefine.
5078 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5079 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5080 * doc/tm.texi: Regenerated.
5081 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5082 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5083 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5084 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5085 instead of int_size_in_bytes.
5086 * expr.c (copy_blkmode_to_reg): Likewise.
5087 * function.c (aggregate_value_p): Return 0 for empty types.
5088 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5090 (locate_and_pad_parm): Call arg size_in_bytes instead
5092 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5093 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5094 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5096 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5097 (std_gimplify_va_arg_expr): Skip empty records. Call
5098 arg_size_in_bytes instead size_in_bytes.
5099 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5100 * tree-core.h (tree_type_common): Add empty_flag.
5101 (tree_decl_common): Update comments.
5102 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5104 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5105 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5107 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5108 * tree.c (default_is_empty_type): New function.
5109 (default_is_empty_record): New function.
5110 (arg_int_size_in_bytes): New function.
5111 (arg_size_in_bytes): New function.
5112 (get_ultimate_context): New function.
5113 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5114 TRANSLATION_UNIT_WARN_EMPTY_P.
5115 (default_is_empty_record, arg_int_size_in_bytes,
5116 arg_size_in_bytes, get_ultimate_context): Declare.
5118 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5120 * config/arm/arm.c (cmse_clear_registers): New function.
5121 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5122 code by call to cmse_clear_registers.
5123 (cmse_nonsecure_entry_clear_before_return): Likewise.
5125 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5127 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5128 (vdot_s32, vdotq_s32): New.
5129 (vdot_lane_u32, vdotq_lane_u32): New.
5130 (vdot_lane_s32, vdotq_lane_s32): New.
5133 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5136 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5137 for unsigned values with fewer HWIs than the precision.
5138 (test_overflow): New function.
5139 (wide_int_cc_tests): Call it.
5141 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5143 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5144 has the same mode class as Pmode.
5146 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5148 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5149 padding_bits_to_clear_ptr.
5150 (cmse_nonsecure_entry_clear_before_return): Likewise.
5152 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5154 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5155 auto_sbitap instead of integer bitfield to control register needing
5158 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5160 PR tree-optimization/83044
5161 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5162 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5163 that need the upper bound. Subtract offset from
5164 get_addr_base_and_unit_offset only if positive and subtract it
5165 before division by eltsize rather than after it.
5168 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5169 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5170 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5174 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5175 expand_binop with *mul_widen_optab, make sure at least one of the
5176 operands doesn't have VOIDmode.
5179 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5181 PR rtl-optimization/82044
5182 PR tree-optimization/82042
5183 * dse.c (record_store): Check for overflow.
5184 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5185 add_wild_read instead of clear_rhs_from_active_local_stores on
5186 overflow and log it into dump_file.
5188 2017-11-22 Richard Biener <rguenther@suse.de>
5190 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5191 fake edges to exit when looking for a place to insert.
5192 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5194 (insert_into_preds_of_block): Commit edge insertion immediately,
5195 assert that doesn't require new BBs.
5196 (fini_pre): Release expressions.
5197 (pass_pre::execute): Shuffle things around a bit, if the fn
5198 is too large do not compute AVAIL either as this is really the
5201 2017-11-22 Richard Biener <rguenther@suse.de>
5203 PR tree-optimization/83089
5204 * tree-if-conv.c (pass_if_conversion::execute): If anything
5205 changed reset SCEV and free the number of iteration estimates.
5207 2017-11-21 Martin Sebor <msebor@redhat.com>
5209 PR tree-optimization/82945
5210 * calls.h (warn_nonstring_bound): Remove unused function.
5212 2017-11-21 Martin Sebor <msebor@redhat.com>
5214 PR tree-optimization/82945
5215 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5216 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5217 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5219 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5220 * calls.h (get_attr_nonstring_decl): Declare new function.
5221 * doc/extend.texi (attribute nonstring): Update.
5222 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5223 get_attr_nonstring_decl and handle it.
5224 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5225 detection of nul-termination.
5226 (strlen_to_stridx): Change to a pointer.
5227 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5228 (pass_strlen::execute): Same.
5230 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5232 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5233 for the new option -mprefer-vector-width=[none|128|256|512].
5234 * config/i386/i386.c (ix86_target_string): remove old style options
5235 -mprefer-avx256 and make -mprefer-avx128 as alias.
5236 (ix86_option_override_internal): Apply defaults for the
5237 -mprefer-vector-width=[128|256] option.
5238 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5239 Implement macros to work with -mprefer-vector-width=.
5240 * config/i386/i386.opt: Implemented option
5241 -mprefer-vector-width=[none|128|256|512].
5242 * doc/invoke.texi: Documentation for
5243 -mprefer-vector-width=[none|128|256|512].
5245 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5247 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5249 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5251 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5253 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5255 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5257 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5258 * expr.c (expand_expr_real_2): Likewise.
5259 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5260 fold_binary_loc): Likewise.
5261 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5262 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5263 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5264 MINUS_EXPR transformations.
5265 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5266 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5267 * tree-inline.c (estimate_operator_cost): Likewise.
5268 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5269 op_symbol_code): Likewise.
5270 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5271 * vr-values.c (extract_range_from_binary_expr): Likewise.
5272 * varasm.c (initializer_constant_valid_p_1): Likewise.
5273 * tree.def: New tree code POINTER_DIFF_EXPR.
5275 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5277 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5278 integer suffix to movbe mnemonic.
5279 (*bswaphi2_movbe): Ditto.
5280 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5282 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5285 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5286 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5287 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5289 2017-11-21 Martin Liska <mliska@suse.cz>
5291 * tree-inline.c (expand_call_inline): Remove not needed
5294 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5295 Jakub Jelinek <jakub@redhat.com>
5298 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5299 Remove static keyword from f variable.
5301 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5303 PR tree-optimization/83086
5304 * gimple-ssa-store-merging.c
5305 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5306 rather than n.base_addr.
5308 2017-11-21 Martin Liska <mliska@suse.cz>
5310 PR rtl-optimization/82044
5311 PR tree-optimization/82042
5312 * dse.c (check_mem_read_rtx): Check for overflow.
5314 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5316 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5319 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5321 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5322 and its corresponding call arg location note.
5324 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5325 Andrew Burgess <andrew.burgess@embecosm.com>
5327 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5329 (arc_return_slot_offset): Likewise.
5330 (arc_eh_return_address_location): New declaration.
5331 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5332 (MUST_SAVE_REGISTER): Add exception handler case.
5333 (MUST_SAVE_RETURN_ADDR): Likewise.
5334 (arc_frame_pointer_required): Likewise.
5335 (arc_frame_pointer_needed): New function.
5336 (arc_compute_frame_size): Changed.
5337 (arc_expand_prologue): Likewise.
5338 (arc_expand_epilogue): Likewise.
5339 (arc_initial_elimination_offset): Likewise.
5340 (arc_return_slot_offset): Delete.
5341 (arc_eh_return_address_location): New function.
5342 (arc_builtin_setjmp_frame_value): Likewise.
5343 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5344 (EH_RETURN_STACKADJ_RTX): Define.
5345 (EH_RETURN_HANDLER_RTX): Likewise.
5346 * config/arc/arc.md (eh_return): Delete.
5348 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5350 * print-tree.h (debug_vec_tree): Remove prototype.
5351 * gdbinit.in (pvt): Remove macro.
5353 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5355 PR tree-optimization/83047
5356 * gimple-ssa-store-merging.c
5357 (imm_store_chain_info::output_merged_store): If the loads with the
5358 same vuse are in different basic blocks, for load_gsi pick a load
5359 location that is dominated by the other loads.
5362 * config/i386/i386.c (ix86_memmodel_check): Start
5363 -Winvalid-memory-model diagnostics with lowercase letter.
5366 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5367 set in_cold_section_p to the partition containing loc_list->first.
5368 When seeing loc_list->last_before_switch node, update secname and
5369 perform range_across_switch second partition handling only after that.
5372 * run-rtl-passes.c: Include debug.h.
5373 (run_rtl_passes): Call debug_hooks->assembly_start.
5374 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5378 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5379 OPTAB_DIRECT in calls to expand_simple_binop.
5381 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5384 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5386 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5389 * spellcheck.h (best_match::blithely_get_best_candidate): New
5392 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5394 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5395 with lower case letter.
5397 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5399 * config/i386/i386.md (bswaphi2): New expander.
5400 (*bswaphi2_movbe): New insn pattern.
5401 (bswaphi -> rorhi pepehole2): New peephole pattern.
5403 2017-11-20 Jeff Law <law@redhat.com>
5405 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5406 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5407 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5408 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5409 gimple-ssa-evrp-analyze.h.
5411 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5412 set BB_VISITED here.
5413 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5415 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5417 * tree-predcom.c: Add general comment on Store-Store chains.
5418 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5420 (get_chain_last_ref_at): Rename into...
5421 (get_chain_last_write_at): ...this.
5422 (get_chain_last_write_before_load): New function.
5423 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5424 CT_STORE_STORE when write reference is added.
5425 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5426 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5427 (initialize_root_vars_store_elim_1): Ditto.
5428 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5429 definition is created.
5430 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5431 chain by replacing it with dominant stored value.
5433 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5435 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5437 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5439 * vr-values.c (extract_range_from_binary_expr): Use a full range
5442 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5444 * config/arm/arm.md (R4_REGNUM): Define constant.
5445 (nonsecure_call_internal): Remove r4 clobber.
5446 (nonsecure_call_value_internal): Likewise.
5447 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5448 clobber and resequence match_operands.
5449 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5450 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5451 (nonsecure_call_value_reg_thumb2): Likewise.
5453 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5455 PR tree-optimization/78821
5456 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5457 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5458 to the address of the base rather than the base itself.
5459 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5460 (find_bswap_or_nop_finalize): New function.
5461 (find_bswap_or_nop): Use it.
5462 (bswap_replace): Return a tree rather than bool, change first
5463 argument from gimple * to gimple_stmt_iterator, allow inserting
5464 into an empty sequence, allow ins_stmt to be NULL - then emit
5465 all stmts into gsi. Fix up MEM_REF address gimplification.
5466 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5467 (struct store_immediate_info): Add N and INS_STMT non-static
5469 (store_immediate_info::store_immediate_info): Initialize them
5470 from newly added ctor args.
5471 (merged_store_group::apply_stores): Formatting fixes. Sort by
5473 (stmts_may_clobber_ref_p): For stores call also
5474 refs_anti_dependent_p.
5475 (gather_bswap_load_refs): New function.
5476 (imm_store_chain_info::try_coalesce_bswap): New method.
5477 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5478 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5479 (imm_store_chain_info::output_merged_store): Fail if number of
5480 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5481 and NOP_EXPR rhs_code.
5482 (pass_store_merging::process_store): Compute n and ins_stmt, if
5483 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5484 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5486 (pass_store_merging::execute): Calculate dominators.
5488 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5489 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5490 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5491 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5492 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5493 class pass_optimize_bswap, bswap_replace,
5494 pass_optimize_bswap::execute): Moved to ...
5495 * gimple-ssa-store-merging.c: ... this file.
5496 Include optabs-tree.h.
5497 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5498 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5499 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5500 anonymous namespace, remove static keywords.
5501 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5502 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5504 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5507 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5509 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5511 * vec.h (debug_helper): New function.
5512 (DEFINE_DEBUG_VEC): New macro.
5513 * hash-set.h (debug_helper): New function.
5514 (DEFINE_DEBUG_HASH_SET): New macro.
5515 * cfg.c (debug_slim (edge)): New function.
5516 Call DEFINE_DEBUG_VEC for edges.
5517 Call DEFINE_DEBUG_HASH_SET for edges.
5518 * cfghooks.c (debug_slim (basic_block)): New function.
5519 Call DEFINE_DEBUG_VEC for basic blocks.
5520 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5521 * print-tree.c (debug_slim): New function to handle trees.
5522 Call DEFINE_DEBUG_VEC for trees.
5523 Call DEFINE_DEBUG_HASH_SET for trees.
5524 (debug (vec<tree, va_gc>) &): Remove.
5525 (debug (<vec<tree, va_gc>) *): Remove.
5526 * print-rtl.c (debug_slim): New function to handle const_rtx.
5527 Call DEFINE_DEBUG_VEC for rtx_def.
5528 Call DEFINE_DEBUG_VEC for rtx_insn.
5529 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5530 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5531 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5532 (debug (vec<rtx_insn *> *ptr): Remove.
5533 (debug_insn_vector): Remove.
5534 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5536 2017-11-20 Tom de Vries <tom@codesourcery.com>
5538 PR rtl-optimization/82020
5539 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5540 IF_THEN_ELSE condition.
5542 2017-11-19 Jeff Law <law@redhat.com>
5544 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5545 of degenerates resulting from ignoring an edge.
5547 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5550 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5552 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5555 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5556 for uninitialied counts.
5558 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5561 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5563 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5566 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5568 2017-11-19 Tom de Vries <tom@codesourcery.com>
5570 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5573 2017-11-19 Tom de Vries <tom@codesourcery.com>
5575 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5578 2017-11-19 Tom de Vries <tom@codesourcery.com>
5580 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5583 2017-11-19 Tom de Vries <tom@codesourcery.com>
5585 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5587 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5589 2017-11-19 Tom de Vries <tom@codesourcery.com>
5591 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5593 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5594 semicolon after MCORE_EXPORT_NAME call.
5596 2017-11-19 Tom de Vries <tom@codesourcery.com>
5599 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5600 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5602 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5604 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5605 (cgraph_edge::clone): Cleanup updating of profile.
5606 * ipa-cp.c (update_profiling_info): Likewise.
5607 * ipa-inline-transform.c (inline_transform): Likewise.
5608 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5609 * ipa-split.c (execute_split_functions): Do not split when function
5611 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5612 * profile-count.c (profile_count::dump): Add global0.
5613 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5615 (profile_count::to_sreal_scale): Likewise.
5616 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5617 (profile_count::combine_with_ipa_count): New function.
5618 * profile-count.h (profile_guessed_global0adjusted): New.
5619 (profile_count::adjusted_zero): New.
5620 (profile_count::global0adjusted): New.
5621 (profile_count::combine_with_ipa_count): New.
5622 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5623 correct profile of return block of split functions.
5624 (copy_cfg_body): Remove unused profile_count.
5625 (copy_body): Likewise.
5626 (expand_call_inline): Update.
5627 (tree_function_versioning): Update.
5629 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5631 * hash-set.h (hash_set::empty): New.
5632 * tree-ssa-threadbackward.h: Delete.
5633 * tree-ssa-threadbackward.c (class thread_jumps): New.
5634 Move max_threaded_paths into class.
5635 (fsm_find_thread_path): Remove arguments that are now in class.
5636 (profitable_jump_thread_path): Rename to...
5637 (thread_jumps::profitable_jump_thread_path): ...this.
5638 (convert_and_register_jump_thread_path): Rename to...
5639 (thread_jumps::convert_and_register_current_path): ...this.
5640 (check_subpath_and_update_thread_path): Rename to...
5641 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5642 (register_jump_thread_path_if_profitable): Rename to...
5643 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5644 (handle_phi): Rename to...
5645 (thread_jumps::handle_phi): ...this.
5646 (handle_assignment): Rename to...
5647 (thread_jumps::handle_assignment): ...this.
5648 (fsm_find_control_statement_thread_paths): Rename to...
5649 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5650 (find_jump_threads_backwards): Rename to...
5651 (thread_jumps::find_jump_threads_backwards): ...this.
5652 Initialize path local data.
5653 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5654 from within thread_jumps class.
5655 (pass_early_thread_jumps::execute): Same.
5657 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5659 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5661 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5663 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5664 to canonical location.
5666 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5669 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5671 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5673 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5675 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5676 rcrt1.o%s/grcrt1.o%s for -static-pie.
5678 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5680 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5681 ix86_shift_rotate_cost): Break out from ...
5682 (ix86_rtx_costs): ... here.
5683 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5686 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5688 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5689 when entry block was promoted unlikely.
5690 (estimate_bb_frequencies): Increase frequency scale.
5691 * profile-count.h (profile_count): Export precision info.
5693 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5695 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5696 disturbin profile of entry block.
5698 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5700 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5703 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5705 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5706 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5708 2017-11-17 Jeff Law <law@redhat.com>
5710 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5711 from evrp_dom_walker class. Various methods moved into new class.
5712 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5713 (evrp_range_analyzer::enter): New method.
5714 (evrp_range_analyzer::leave): New method.
5715 (evrp_dom_walker): Remove delegators no longer needed by this class.
5716 Replace vr_values data member with evrp_range_analyzer
5718 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5719 method extracted from evrp_dom_walker::before_dom_children.
5720 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5721 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5723 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5724 Add private copy constructor and move assignment operators.
5725 Privatize methods and class data where trivially possible.
5726 (evrp_dom_walker::cleanup): New function, extracted from
5727 execute_early_vrp. Simplify access to class data.
5729 * vr-values.h (get_output_for_vrp): Prototype.
5730 * vr-values.c (get_output_for_vrp): New function extracted from
5731 vrp_visit_assignment_or_call and extract_range_from_stmt.
5732 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5734 2017-11-17 Luis Machado <luis.machado@linaro.org>
5736 * config/aarch64/aarch64.c
5737 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5738 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5739 tune_params::AUTOPREFETCHER_WEAK.
5741 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5744 * config/arm/arm.c (arm_valid_target_attribute_rec):
5745 Parse "arch=" and "+<ext>".
5746 (arm_valid_target_attribute_tree): Re-init global options.
5747 (arm_option_override): Make non-static.
5748 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5749 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5750 (__ARM_FEATURE_CRC32): Support undef.
5751 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5752 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5754 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5756 * gdbinit.in (break-on-diagnostic): New command.
5758 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5760 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5761 return to indirect jump for EH return if control-flow
5762 protection is enabled. Change explicit 'false' argument in
5763 pro_epilogue_adjust_stack with a value of
5765 * config/i386/i386.md (simple_return_indirect_internal):
5766 Remove SImode restriction to support 64-bit.
5768 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5770 * combine.c (added_notes_insn): New.
5771 (try_combine): Handle added_notes_insn like added_links_insn.
5772 Rewrite return value code.
5773 (distribute_notes): Set added_notes_insn to the earliest insn we added
5776 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5778 PR rtl-optimization/82621
5779 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5780 dest of one of those SETs is unused.
5782 2017-11-17 Richard Biener <rguenther@suse.de>
5785 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5786 * tree-pretty-print.c (dump_generic_node): Handle
5787 annot_expr_parallel_kind.
5788 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5789 * gimplify.c (gimple_boolify): Likewise.
5791 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5793 * config.gcc (extra_headers): Add cet.h for x86 targets.
5794 * config/i386/cet.h: New file.
5795 * doc/install.texi: Add --enable-cet/--disable-cet.
5797 2017-11-17 Richard Biener <rguenther@suse.de>
5799 PR tree-optimization/83017
5800 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5801 (gen_parallel_loop): Properly count iterations.
5802 (parallelize_loops): Handle loop->can_be_parallel independent
5803 of flag_loop_parallelize_all. Make static profitability test match
5805 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5806 * invoke.texi (parloops-min-per-thread): Document.
5808 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5810 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5811 upstreaming review comments.
5813 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5815 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5816 with fast unaligned access.
5817 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5819 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5821 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5822 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5824 2017-11-17 Richard Biener <rguenther@suse.de>
5826 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5827 folding of references.
5829 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5832 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5833 of replacing call to strncmp with corresponding call to strcmp when
5836 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5838 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5839 option prefer-avx256 for skylake-avx512 configuration.
5840 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5841 (get_builtin_code_for_version): Ditto.
5843 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5844 Monk Chiang <sh.chiang04@gmail.com>
5846 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5847 (FIXED_REGISTERS): Reserve more register numbers.
5848 (CALL_USED_REGISTERS): Likewise.
5849 (REG_ALLOC_ORDER): Likewise.
5850 (REG_CLASS_CONTENTS): Likewise.
5851 (REGISTER_NAMES): Likewise.
5853 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5854 Kito Cheng <kito.cheng@gmail.com>
5856 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5858 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5860 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5862 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5863 parameter list for vec_splats.
5865 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5867 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5869 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5871 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5874 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5875 with the required version of automake 1.11.6.
5877 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5879 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5881 (DU_C2_3_power9): Correct reservation combinations.
5882 (FP_DIV_power9, VEC_DIV_power9): New.
5883 (power9-alu): Split out rotate/shift...
5884 (power9-rot): ...to here, correct dispatch resource.
5885 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5887 (power9-fp): Correct latency.
5888 (power9-sdiv): Add div/sqrt resource.
5889 (power9-ddiv): Correct latency, add div/sqrt resource.
5890 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5891 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5893 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5895 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5897 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5898 switch statement mapping KF built-ins to TF built-ins if we don't
5899 have the proper ISA 3.0 assembler support.
5901 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5903 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5904 (lower_emutls_function_body): Do not compute it.
5906 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5908 * ipa-split.c (split_bb_info): Turn time to sreal.
5909 (split_point): Likewise.
5910 (dump_split_point): Likewise.
5911 (fine_split_points): Likewise.
5912 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5914 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5916 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5919 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5921 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5923 (expensive_function_p): Remove useless assert.
5924 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5926 2017-11-16 Martin Sebor <msebor@redhat.com>
5928 PR tree-optimization/82588
5929 PR tree-optimization/82583
5930 * tree-vrp.c (check_array_ref): Handle flexible array members,
5931 string literals, and inner indices.
5932 (search_for_addr_array): Add detail to diagnostics.
5934 2017-11-16 Nathan Sidwell <nathan@acm.org>
5938 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5939 (SET_DECL_ASSEMBLER_NAME): Forward to
5940 overwrite_decl_assembler_name.
5941 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5942 (overwrite_decl_assembler_name): Declare.
5943 * tree.c (overwrite_decl_assembler_name): New.
5944 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5945 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5946 (LANG_HOOKS_INITIALIZER): Add it.
5947 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5948 * langhooks.c (lhd_set_decl_assembler_name): Use
5949 SET_DECL_ASSEMBLER_NAME.
5950 (lhd_overwrite_decl_assembler_name): Default implementation.
5952 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5953 Jackson Woodruff <jackson.woodruff@arm.com>
5955 PR tree-optimization/71026
5956 * match.pd: Canonicalize constant multiplies in division.
5958 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5960 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5961 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5962 Remove OPT_fomit_frame_pointer entry.
5963 * common/config/arc/arc-common.c: Likewise.
5964 * common/config/arm/arm-common.c: Likewise.
5965 * common/config/avr/avr-common.c: Likewise.
5966 * common/config/c6x/c6x-common.c: Likewise.
5967 * common/config/cr16/cr16-common.c: Likewise.
5968 * common/config/cris/cris-common.c: Likewise.
5969 * common/config/epiphany/epiphany-common.c: Likewise.
5970 * common/config/fr30/fr30-common.c: Likewise.
5971 * common/config/frv/frv-common.c: Likewise.
5972 * common/config/ia64/ia64-common.c: Likewise.
5973 * common/config/iq2000/iq2000-common.c: Likewise.
5974 * common/config/lm32/lm32-common.c: Likewise.
5975 * common/config/m32r/m32r-common.c: Likewise.
5976 * common/config/mcore/mcore-common.c: Likewise.
5977 * common/config/microblaze/microblaze-common.c: Likewise.
5978 * common/config/mips/mips-common.c: Likewise.
5979 * common/config/mmix/mmix-common.c: Likewise.
5980 * common/config/mn10300/mn10300-common.c: Likewise.
5981 * common/config/nios2/nios2-common.c: Likewise.
5982 * common/config/pa/pa-common.c: Likewise.
5983 * common/config/pdp11/pdp11-common.c: Likewise.
5984 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5985 * common/config/riscv/riscv-common.c: Likewise.
5986 * common/config/rs6000/rs6000-common.c: Likewise.
5987 * common/config/rx/rx-common.c: Likewise.
5988 * common/config/s390/s390-common.c: Likewise.
5989 * common/config/sh/sh-common.c: Likewise.
5990 * common/config/sparc/sparc-common.c: Likewise.
5991 * common/config/tilegx/tilegx-common.c: Likewise.
5992 * common/config/tilepro/tilepro-common.c: Likewise.
5993 * common/config/v850/v850-common.c: Likewise.
5994 * common/config/visium/visium-common.c: Likewise.
5995 * common/config/xstormy16/xstormy16-common.c: Likewise.
5996 * common/config/xtensa/xtensa-common.c: Likewise.
5997 * invoke.texi (-fomit-frame-pointer): Update documentation.
5999 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6001 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6003 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6005 * cfg.c (scale_bbs_frequencies_int,
6006 cale_bbs_frequencies_gcov_type): Remove.
6007 * cfg.h (scale_bbs_frequencies_int,
6008 cale_bbs_frequencies_gcov_type): Remove.
6010 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6012 * tree-ssa-loop-manip.c
6013 (scale_dominated_blocks_in_loop): Update to profile counts.
6014 (tree_transform_and_unroll_loop): Likewise.
6016 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6018 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6019 scale_bbs_frequencies_int.
6021 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6023 * final.c (compute_alignments): Use counts rather than frequencies.
6025 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6027 * cfgloopanal.c: Include sreal.h
6028 (average_num_loop_insns): Use counts and sreal for accounting.
6030 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6032 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6035 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6036 Kito Cheng <kito.cheng@gmail.com>
6038 * config/nds32/constraints.md: Provide more constraints.
6039 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6040 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6041 support constraints usage.
6043 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6045 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6047 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6048 Kito Cheng <kito.cheng@gmail.com>
6050 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6051 * config/nds32/nds32.opt: Refine the layout.
6052 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6053 TARGET_EXT_STRING): Support new options.
6054 * config/nds32/nds32.h: Likewise.
6055 * config/nds32/nds32.md: Likewise.
6056 * config/nds32/nds32-predicates.c: Likewise.
6057 * config/nds32/constraints.md: Likewise.
6058 * common/config/nds32/nds32-common.c: Likewise.
6060 2017-11-16 Julia Koval <julia.koval@intel.com>
6063 * config/i386/gfniintrin.h: Add sse check.
6064 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6066 2017-11-16 Julia Koval <julia.koval@intel.com>
6068 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6069 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6070 (ix86_handle_option): Handle -mavx512vbmi2.
6071 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6072 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6073 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6074 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6075 (ix86_valid_target_attribute_inner_p): Ditto.
6076 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6077 * config/i386/i386.opt (mavx512vbmi2): New option.
6078 * doc/invoke.texi: Add new option.
6080 2017-11-16 Julia Koval <julia.koval@intel.com>
6082 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6083 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6084 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6085 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6086 _mm512_gf2p8mul_epi8): New intrinsics.
6087 * config/i386/i386-builtin-types.def
6088 (V64QI_FTYPE_V64QI_V64QI): New type.
6089 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6090 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6091 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6092 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6093 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6094 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6096 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6098 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6099 explicitly as a stream of bytes.
6101 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6103 * config/rs6000/altivec.h (vec_xst_be): New #define.
6104 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6105 and externalize from *altivec_vperm_<mode>_internal.
6106 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6108 (XL_BE_V8HI): Likewise.
6109 (XL_BE_V4SI): Likewise.
6110 (XL_BE_V4SI): Likewise.
6111 (XL_BE_V2DI): Likewise.
6112 (XL_BE_V4SF): Likewise.
6113 (XL_BE_V2DF): Likewise.
6114 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6115 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6116 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6117 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6118 VSX_BUILTIN_VEC_XST_BE.
6119 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6120 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6122 (altivec_init_builtins): Replace conditional calls to def_builtin
6123 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6124 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6125 with unconditional calls. Remove calls to def_builtin for
6126 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6127 __builtin_vec_xst_be.
6128 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6129 to define_expand, and add alternate RTL generation for P8.
6130 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6131 vsx_ld_elemrev_v8hi.
6132 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6133 add alternate RTL generation for P8.
6134 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6135 vsx_ld_elemrev_v16qi.
6136 (vsx_st_elemrev_v8hi): Convert define_insn
6137 to define_expand, and add alternate RTL generation for P8.
6138 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6139 vsx_st_elemrev_v8hi.
6140 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6141 add alternate RTL generation for P8.
6142 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6143 vsx_st_elemrev_v16qi.
6145 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6148 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6149 TARGET_AVX512ER check.
6150 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6151 neither -mzeroupper nor -mno-zeroupper is used and
6152 TARGET_EMIT_VZEROUPPER is set.
6153 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6154 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6156 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6158 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6159 folding of vector compares.
6160 (fold_build_vec_cmp): New helper function.
6161 (fold_compare_helper): New helper function.
6162 (builtin_function_type): Add compare builtins to the list of functions
6163 having unsigned arguments. Cosmetic updates to comment indentation.
6164 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6165 the not+eq combination.
6167 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6169 PR tree-optimization/82726
6170 PR tree-optimization/70754
6171 * tree-predcom.c (order_drefs_by_pos): New function.
6172 (combine_chains): Move code setting has_max_use_after to...
6173 (try_combine_chains): ...here. New parameter. Sort combined chains
6174 according to position information.
6175 (tree_predictive_commoning_loop): Update call to above function.
6176 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6178 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6180 PR tree-optimization/82726
6182 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6184 PR tree-optimization/70754
6185 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6186 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6187 combined stmt before it if not NULL.
6188 (combine_chains): Process refs reversely and compute dominance point
6192 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6194 PR tree-optimization/79663
6195 * tree-predcom.c (combine_chains): Process refs in reverse order
6196 only for ZERO length chains, and add explaining comment.
6198 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6200 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6201 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6202 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6204 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6206 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6208 2017-11-15 Martin Liska <mliska@suse.cz>
6210 * tree-cfg.c (pass_warn_function_return::execute):
6211 Compare warn_return_type for greater than zero.
6213 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6217 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6218 to return true on Xeon and not on Xeon Phi.
6219 (ix86_check_avx256_register): Changed to ...
6220 (ix86_check_avx_upper_register): ... this. Add extra check for
6221 VALID_AVX512F_REG_OR_XI_MODE.
6222 (ix86_avx_u128_mode_needed): Changed
6223 ix86_check_avx256_register to ix86_check_avx_upper_register.
6224 (ix86_check_avx256_stores): Changed to ...
6225 (ix86_check_avx_upper_stores): ... this. Changed
6226 ix86_check_avx256_register to ix86_check_avx_upper_register.
6227 (ix86_avx_u128_mode_after): Changed
6228 avx_reg256_found to avx_upper_reg_found. Changed
6229 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6230 (ix86_avx_u128_mode_entry): Changed
6231 ix86_check_avx256_register to ix86_check_avx_upper_register.
6232 (ix86_avx_u128_mode_exit): Ditto.
6233 * config/i386/i386.h: (host_detect_local_cpu): New define.
6235 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6237 * config/arm/xgene1.md (xgene1): Split into automatons
6238 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6239 (xgene1_f_load): Adjust reservations and/or types.
6240 (xgene1_f_store): Likewise.
6241 (xgene1_load_pair): Likewise.
6242 (xgene1_store_pair): Likewise.
6243 (xgene1_fp_load1): Likewise.
6244 (xgene1_load1): Likewise.
6245 (xgene1_store1): Likewise.
6246 (xgene1_move): Likewise.
6247 (xgene1_alu): Likewise.
6248 (xgene1_simd): Likewise.
6249 (xgene1_bfm): Likewise.
6250 (xgene1_neon_load1): Likewise.
6251 (xgene1_neon_store1): Likewise.
6252 (xgene1_neon_logic): Likewise.
6253 (xgene1_neon_st1): Likewise.
6254 (xgene1_neon_ld1r): Likewise.
6255 (xgene1_alu_cond): Added.
6256 (xgene1_shift_reg): Likwise.
6257 (xgene1_bfx): Likewise.
6258 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6260 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6263 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6265 (can_widen_mult_without_libcall): New function.
6266 (expand_mul_overflow): If only checking unsigned mul overflow,
6267 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6268 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6269 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6270 (expand_DIVMOD): Formatting fix.
6271 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6272 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6273 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6275 PR tree-optimization/82977
6276 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6277 constructed temporary to strlen_to_stridx.put.
6279 2017-11-15 Martin Liska <mliska@suse.cz>
6281 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6282 * configure: Regenerate.
6284 2017-11-15 Martin Liska <mliska@suse.cz>
6287 * config/sh/sh-mem.cc: Use proper probability for
6290 2017-11-14 Jeff Law <law@redhat.com>
6292 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6293 the red zone for stack_clash_protection_final_dynamic_probe targets
6294 when the total dynamic stack size is zero bytes.
6296 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6297 blocks is post order.
6299 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6301 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6302 * final.c (rest_of_clean_state): Set it for the
6303 -fcompare-debug dump.
6304 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6305 class when TDF_COMPARE_DEBUG is set.
6307 * dwarf2out.c (gen_producer_string): Discard
6310 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6313 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6314 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6315 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6316 (__TGMATH_REAL_2_3): Remove macros.
6317 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6318 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6320 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6322 (remquo): Define using __TGMATH_REAL_3.
6324 2017-11-14 Jeff Law <law@redhat.com>
6326 * vr-values.c: New file with contents extracted from tree-vrp.c.
6327 * Makefile.in (OBJS): Add vr-values.o
6328 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6329 (set_value_range, set_and_canonicalize_value_range): Likewise.
6330 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6331 (value_range_constant_singleton, symbolic_range_p): Likewise.
6332 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6333 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6334 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6335 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6336 (find_case_label_range, find_case_label_index): Likewise.
6337 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6338 (range_int_cst_singleton_p, value_inside_range): Likewise.
6339 (get_single_symbol): Likewise.
6340 (switch_update): Move structure definition here.
6341 (to_remove_edges, to_update_switch_stmts): Provide externs.
6342 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6343 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6344 (vrp_val_is_min, set_value_range): Likewise.
6345 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6346 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6347 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6348 (range_is_nonnull, range_int_cst_p): Likewwise.
6349 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6350 (get_single_symbol, operand_less_p): Likewise
6351 (compare_values_warnv, compare_values): Likewise.
6352 (value_inside_range, value_range_constant_singleton): Likewise.
6353 (zero_nonzero_bitgs_from_vr): Likewise.
6354 (extract_range_from_binary_expr_1): Likewise.
6355 (overflow_comparison_p): Likewise.
6356 (to_remove_edges, to_update_switch_stmts): Likewise.
6357 (find_case_label-index, find_case_label_range): Likewise.
6358 (switch_update, set_value_range_to_nonnegative): Remove.
6359 (set_value_range_to_truthvalue): Likewise.
6360 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6361 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6362 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6363 (find_case_label_ranges, test_for_singularity): Likewise.
6364 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6365 (x_vr_values): Move to its remaining use site.
6367 2017-11-10 Jeff Law <law@redhat.com>
6369 * vr-values.h (VR_INITIALIZER): Move #define here.
6370 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6371 * Makefile.in (OBJS): Add tree-evrp.o
6372 * tree-vrp.h (assert_info): Move structure definition here.
6373 (set_value_range_to_varying): Prototype.
6374 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6375 (infer_value_range, register_edge_assert_for): Likewise.
6376 (stmt_interesting_for_vrp): Likewise.
6377 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6378 (set_value_range_to_varying): No longer static.
6379 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6380 (infer_value_range, register_edge_assert_for): Likewise.
6382 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6384 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6385 generate the XXBRD instruction.
6387 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6388 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6389 as being compatible if -mabi=ieeelongdouble.
6390 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6391 to setup float128 built-ins with hardware support.
6392 (BU_FLOAT128_HW_2): Likewise.
6393 (BU_FLOAT128_HW_3): Likewise.
6394 (BU_FLOAT128_HW_VSX_1): Likewise.
6395 (BU_FLOAT128_HW_VSX_2): Likewise.
6396 (scalar_extract_expq): Change float128 built-in functions to
6397 accommodate having both KFmode and TFmode functions. Use the
6398 KFmode variant as the default.
6399 (scalar_extract_sigq): Likewise.
6400 (scalar_test_neg_qp): Likewise.
6401 (scalar_insert_exp_q): Likewise.
6402 (scalar_insert_exp_qp): Likewise.
6403 (scalar_test_data_class_qp): Likewise.
6404 (sqrtf128_round_to_odd): Delete processing the round to odd
6405 built-in functions as special built-in functions, and define them
6406 as float128 built-ins. Use the KFmode variant as the default.
6407 (truncf128_round_to_odd): Likewise.
6408 (addf128_round_to_odd): Likewise.
6409 (subf128_round_to_odd): Likewise.
6410 (mulf128_round_to_odd): Likewise.
6411 (divf128_round_to_odd): Likewise.
6412 (fmaf128_round_to_odd): Likewise.
6413 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6414 support for KFmode and TFmode xststdcqp calls.
6415 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6416 point, switch the built-in handlers for the get/set float128
6417 exponent, get float128 mantissa, float128 test built-ins, and the
6418 float128 round to odd built-in functions. Eliminate creating the
6419 float128 round to odd built-in functions as special built-ins.
6420 (rs6000_init_builtins): Eliminate special creation of the float128
6421 round to odd built-in functions.
6422 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6423 function insns to support both TFmode and KFmode variants.
6424 (xsxsigqp_<mode>): Likewise.
6425 (xsiexpqpf_<mode>): Likewise.
6426 (xsiexpqp_<mode>): Likewise.
6427 (xststdcqp_<mode>): Likewise.
6428 (xststdcnegqp_<mode>): Likewise.
6429 (xststdcqp_<mode>): Likewise.
6431 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6433 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6434 unused path_in_freq_ptr parameter.
6435 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6437 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6439 * ipa-inline.c (edge_badness): Dump sreal frequency.
6440 (compute_inlined_call_time): Match natural implementaiton ...
6441 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6442 forgotten division by CGRAPH_FREQ_BASE.
6444 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6446 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6447 Solaris 11. Update comment.
6448 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6450 * config/sol2.h (STARTFILE_SPEC): Likewise.
6451 * configure: Regenerate.
6453 2017-11-14 Carl Love <cel@us.ibm.com>
6455 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6456 le_ and be_ prefixes to swap* variables. Remove
6457 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6460 2017-11-14 Jason Merrill <jason@redhat.com>
6462 Support GTY((cache)) on hash_map.
6463 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6464 (ggc_cache_remove): Override it instead of ggc_mx.
6465 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6466 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6467 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6468 (simple_cache_map_traits): Override maybe_mx.
6469 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6470 (hash_map): Friend gt_cleare_cache.
6471 (gt_cleare_cache): New.
6472 * tree.h (tree_cache_traits): New hash_map traits class.
6473 (tree_cache_map): New typedef.
6475 2017-11-14 Richard Biener <rguenther@suse.de>
6477 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6478 paramter and handling.
6479 (cleanup_control_flow_bb): Likewise.
6480 (cleanup_control_flow_pre): New helper performing a DFS walk
6481 to call cleanup_control_flow_bb in PRE order.
6482 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6483 via cleanup_control_flow_pre.
6485 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6487 * config/aarch64/aarch64-simd.md
6488 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6489 (*aarch64_simd_bsl<mode>_alt): Likewise.
6490 (aarch64_simd_bsldi_internal): New.
6491 (aarch64_simd_bsldi_alt): Likewise.
6493 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6495 * tracer.c (better_p): Do not compare frequencies.
6496 * reg-stack.c (better_edge): Likewise.
6497 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6500 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6502 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6503 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6504 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6505 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6506 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6507 * predict.c (maybe_hot_frequency_p): Inline to ...
6508 (maybe_hot_count_p): ... here; rewrite to counts.
6509 (counts_to_freqs): Rename to ...
6510 (update_max_bb_count): ... this one.
6511 (expensive_function_p): Use counts.
6512 (estimate_bb_frequencies): Update.
6513 (rebuild_frequencies): Update.
6514 * predict.h (counts_to_freqs): Rename to ...
6515 (update_max_bb_count): ... this one.
6516 * profile.c (compute_branch_probabilities): Add debug info
6517 * tree-inline.c (expand_call_inline): Update debug info.
6518 (optimize_inline_calls): Use update_max_bb_count..
6519 (tree_function_versioning): Use update_max_bb_count..
6520 * value-prof.c (gimple_value_profile_transformations):
6521 Do not use update_max_bb_count.
6523 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6525 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6526 always use frequencies.
6528 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6530 * bb-reorder.c: Remove frequencies from comments.
6531 (better_edge_p): Use profile counts.
6532 (find_traces): Dump profile counts.
6533 (rotate_loop): Use profile counts.
6534 (find_traces_1_round): Likewise.
6535 (connect_better_edge_p): Use counts instead of probabilities for
6537 (copy_bb_p): Drop early check for non-0 frequency.
6538 (sanitize_hot_paths): Update comments.
6540 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6542 * ipa-split.c (struct split_point): Add count.
6543 (consider_split): Do not compute incoming frequency; compute incoming
6544 count and store it to split_point.
6545 (split_function): Set count of the call to split part correctly.
6547 2017-11-13 Carl Love <cel@us.ibm.com>
6549 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6551 2017-11-13 Tom Tromey <tom@tromey.com>
6553 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6555 2017-11-13 Carl Love <cel@us.ibm.com>
6557 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6558 Add support for builtins:
6559 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6560 vector {un,}signed {char,int,short},
6561 vector {un,}signed {char,int,short}) arguments.
6562 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6563 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6564 Add BU_P9V_AV_2 expansions for the builtins.
6565 * config/rs6000/altivec.h (vec_first_match_index,
6566 vec_first_mismatch_index, vec_first_match_or_eos_index,
6567 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6568 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6569 new extern declaration.
6570 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6571 * config/rs6000/vsx.md (first_match_index_<mode>,
6572 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6573 first_mismatch_or_eos_index_<mode>): Add define expand.
6574 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6575 * doc/extend.texi: Update the built-in documenation file for the new
6578 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6580 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6581 into the min/max operations for _Float<N> and _Float<N>X types.
6583 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6586 * dwarf2out.c (do_eh_frame): New static variable.
6587 (dwarf2out_begin_prologue): Set it.
6588 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6590 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6592 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6594 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6596 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6599 * debug.h (dwarf2out_do_eh_frame): Declare.
6600 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6601 (dwarf2out_do_frame): Use it.
6602 (dwarf2out_do_cfi_asm): Likewise.
6603 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6604 (dwarf2out_assembly_start): Likewise.
6605 (dwarf2out_begin_prologue): Fix comment.
6606 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6607 if the target needs either debug or unwind DWARF2 info.
6608 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6609 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6610 -ftrapping-math, -ftrapv and -fwrapv.
6612 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6614 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6615 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6616 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6617 (estimate_edge_size_and_time): Likewise.
6618 (ipa_merge_fn_summary_after_inlining): Likewise.
6619 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6620 (compute_uninlined_call_time): Use sreal_frequency.
6621 (compute_inlined_call_time): Likewise.
6622 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6623 * profile-count.c: Include sreal.h.
6624 (profile_count::to_sreal_scale): New.
6625 * profile-count.h: Forward declare sreal.
6626 (profile_count::to_sreal_scale): Declare.
6628 2017-11-13 Nathan Sidwell <nathan@acm.org>
6630 * diagnostic.c (maybe_line_and_column): New.
6631 (diagnostic_get_location_text): Use it.
6632 (diagnostic_report_current_module): Likewise.
6633 (test_diagnostic_get_location_text): Add tests.
6635 2017-11-13 Luis Machado <luis.machado@linaro.org>
6637 * doc/md.texi (Specifying processor pipeline description): Fix
6638 incorrect latency for the div instruction example.
6640 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6642 PR tree-optimization/78821
6643 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6644 that bit_not_p is the same.
6645 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6646 (split_group): Count precisely bit_not_p bits in each statement.
6647 (invert_op): New function.
6648 (imm_store_chain_info::output_merged_store): Use invert_op to
6649 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6650 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6652 2017-11-13 Martin Liska <mliska@suse.cz>
6654 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6655 (struct source_info): Likewise.
6656 (add_branch_counts): Likewise.
6657 (add_line_counts): Likewise.
6658 (function_summary): Likewise.
6659 (output_intermediate_line): Likewise.
6660 (generate_results): Likewise.
6662 2017-11-13 Martin Liska <mliska@suse.cz>
6664 * gcov.c (struct block_info): Remove typedef for block_t.
6665 (struct line_info): Likewise.
6666 (line_info::has_block): Likewise.
6667 (EXIT_BLOCK): Likewise.
6668 (unblock): Likewise.
6669 (circuit): Likewise.
6670 (get_cycles_count): Likewise.
6671 (process_file): Likewise.
6672 (read_graph_file): Likewise.
6673 (solve_flow_graph): Likewise.
6674 (find_exception_blocks): Likewise.
6675 (add_line_counts): Likewise.
6676 (accumulate_line_info): Likewise.
6677 (output_line_details): Likewise.
6679 2017-11-13 Martin Liska <mliska@suse.cz>
6681 * gcov.c (struct arc_info): Remove typedef for arc_t.
6682 (struct line_info): Likewise.
6683 (add_branch_counts): Likewise.
6684 (output_branch_count): Likewise.
6685 (function_info::~function_info): Likewise.
6686 (circuit): Likewise.
6687 (output_intermediate_line): Likewise.
6688 (read_graph_file): Likewise.
6689 (solve_flow_graph): Likewise.
6690 (find_exception_blocks): Likewise.
6691 (add_line_counts): Likewise.
6692 (accumulate_line_info): Likewise.
6693 (output_line_details): Likewise.
6694 (output_function_details): Likewise.
6696 2017-11-13 Martin Liska <mliska@suse.cz>
6698 * gcov.c (struct function_info): Remove typedef for function_t.
6699 (struct source_info): Likewise.
6700 (source_info::get_functions_at_location): Likewise.
6701 (solve_flow_graph): Likewise.
6702 (find_exception_blocks): Likewise.
6703 (add_line_counts): Likewise.
6704 (output_intermediate_file): Likewise.
6705 (process_file): Likewise.
6706 (generate_results): Likewise.
6707 (release_structures): Likewise.
6708 (read_graph_file): Likewise.
6709 (read_count_file): Likewise.
6710 (accumulate_line_counts): Likewise.
6711 (output_lines): Likewise.
6713 2017-11-13 Martin Liska <mliska@suse.cz>
6715 * gcov.c (function_info::function_info): Remove num_counts
6716 and add vector<gcov_type>.
6717 (function_info::~function_info): Use the vector.
6718 (process_file): Likewise.
6719 (read_graph_file): Likewise.
6720 (read_count_file): Likewise.
6721 (solve_flow_graph): Likewise.
6723 2017-11-13 Martin Liska <mliska@suse.cz>
6725 * gcov.c (function_info::is_artificial): New function.
6726 (process_file): Erase all artificial early.
6727 (generate_results): Skip as all artificial are already
6730 2017-11-13 Martin Liska <mliska@suse.cz>
6732 * gcov.c (read_graph_file): Store to global vector of functions.
6733 (read_count_file): Iterate the vector.
6734 (process_file): Likewise.
6735 (generate_results): Likewise.
6736 (release_structures): Likewise.
6738 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6740 PR tree-optimization/82954
6741 * gimple-ssa-store-merging.c
6742 (imm_store_chain_info::coalesce_immediate_stores): If
6743 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6745 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6747 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6748 Upddate call to ENDIAN_LANE_N.
6749 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6750 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6751 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6752 and use aarch64_endian_lane_rtx.
6753 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6755 2017-11-12 Tom de Vries <tom@codesourcery.com>
6757 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6759 2017-11-12 Tom de Vries <tom@codesourcery.com>
6761 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6763 2017-11-12 Tom de Vries <tom@codesourcery.com>
6765 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6767 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6768 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6769 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6771 2017-11-11 Martin Sebor <msebor@redhat.com>
6774 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6777 * prefic.c (translate_name): Replace strncpy with memcpy to
6778 avoid -Wstringop-truncation.
6780 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6782 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6784 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6786 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6788 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6790 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6793 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6795 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6796 enable generating XXBRH if the value is in a vector register.
6797 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6798 value is in a vector register.
6799 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6800 register to register bswap64's instead of doing the GPR sequence
6801 used on previous machines.
6802 (bswapdi2_xxbrd): New insn.
6803 (bswapdi2_reg): Disallow on ISA 3.0.
6804 (register to register bswap64 splitter): Do not split the insn on
6805 ISA 3.0 systems that use XXBRD.
6807 2017-11-10 Martin Sebor <msebor@redhat.com>
6810 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6812 (find_subframework_file): Same.
6814 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6816 * auto-profile.c (afdo_indirect_call): Drop frequency.
6817 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6818 (cgraph_node::create_edge): Drop frequency argument.
6819 (cgraph_node::create_indirect_edge): Drop frequency argument.
6820 (cgraph_edge::make_speculative): Drop frequency arguments.
6821 (cgraph_edge::resolve_speculation): Do not update frequencies
6822 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6823 (cgraph_node::dump): Check consistency in IPA mode.
6824 (cgraph_edge::maybe_hot_p): Use IPA counter.
6825 (cgraph_edge::verify_count_and_frequency): Rename to ...
6826 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6827 (cgraph_node::verify_node): Update.
6828 * cgraph.h (struct cgraph_edge): Drop frequency.
6829 (cgraph_edge::frequency): New function.
6830 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6832 (cgraph_edge::rebuild_edges): Likewise.
6833 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6834 (duplicate_thunk_for_node): Do not pass frequency.
6835 (cgraph_node::create_clone): Scale only counts.
6836 (cgraph_node::create_virtual_clone): Do not pass frequency.
6837 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6838 (cgraph_node::create_version_clone): Do not pass frequency.
6839 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6840 (cgraph_node::expand_thunk): Do not pass frequency.
6841 (cgraph_node::create_wrapper): Do not pass frequency.
6842 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6844 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6845 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6846 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6847 (gather_caller_stats): Use frequency function.
6848 (ipcp_cloning_candidate_p): Use frequency function.
6849 (ipcp_propagate_stage): Use frequency function.
6850 (get_info_about_necessary_edges): Use frequency function.
6851 (update_profiling_info): Update only IPA profile.
6852 (update_specialized_profile): Use frequency functoin.
6853 (perhaps_add_new_callers): Update only IPA profile.
6854 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6855 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6856 (dump_ipa_call_summary): Use frequency function.
6857 (estimate_edge_size_and_time): Use frequency function.
6858 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6859 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6860 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6861 (update_noncloned_counts): ... ths one; scale counts only.
6862 (clone_inlined_nodes): Do not scale frequency.
6863 (inline_call): Do not pass frequency.
6864 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6865 (compute_inlined_call_time): Use IPA profile.
6866 (want_inline_small_function_p): Use IPA profile.
6867 (want_inline_self_recursive_call_p): Use IPA profile.
6868 (edge_badness): Use IPA profile.
6869 (lookup_recursive_calls): Use IPA profile.
6870 (recursive_inlining): Do not pass frequency.
6871 (resolve_noninline_speculation): Do not update frequency.
6872 (inline_small_functions): Collect max of IPA profile.
6873 (dump_overall_stats): Dump IPA porfile.
6874 (dump_inline_stats): Dump IPA porfile.
6875 (ipa_inline): Collect IPA stats.
6876 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6877 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6878 (ipa_propagate_frequency): Use frequency function.
6879 (ipa_profile): Cleanup.
6880 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6881 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6882 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6883 (input_node): Do not stream frequency.
6884 (input_edge): Do not stream frequency.
6885 (merge_profile_summaries): Scale only IPA profiles.
6886 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6887 * predict.c (drop_profile): Do not recompute frequency.
6888 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6889 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6890 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6891 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6892 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6893 * tree-inline.c (copy_bb): Do not scale frequency.
6894 (expand_call_inline): Do not scale frequency.
6895 (tree_function_versioning): Do not scale frequency.
6896 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6898 2017-11-10 Julia Koval <julia.koval@intel.com>
6900 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6901 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6902 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6903 (_mm256_maskz_gf2p8affine_epi64_epi8)
6904 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6905 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6906 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6907 (__builtin_ia32_vgf2p8affineqb_v32qi)
6908 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6909 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6911 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6915 (arm_option_override): Refactor.
6916 (arm_option_reconfigure_globals): New.
6917 (arm_options_perform_arch_sanity_checks): New.
6918 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6920 (arm_options_perform_arch_sanity_checks): Likewise
6922 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6924 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6925 (power9-qpmul): New.
6926 * rs6000/rs6000.md ("type" attr): Add qmul.
6927 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6928 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6929 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6931 2017-11-10 Martin Sebor <msebor@redhat.com>
6934 * builtins.c (compute_objsize): Handle arrays that
6935 compute_builtin_object_size likes to fail for. Make extern.
6936 * builtins.h (compute_objsize): Declare.
6937 (check_strncpy_sizes): New function.
6938 (expand_builtin_strncpy): Call check_strncpy_sizes.
6939 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6940 -Wstringop-truncation.
6941 (gimple_fold_builtin_strncat): Same.
6942 * gimple.c (gimple_build_call_from_tree): Set call location.
6943 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6944 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6945 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6946 (handle_builtin_strlen): Use strlen_to_stridx.
6947 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6949 Use strlen_to_stridx.
6950 (pass_strlen::execute): Release strlen_to_stridx.
6951 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6952 (-Wstringop-truncation): Document new option.
6954 2017-11-10 Martin Liska <mliska@suse.cz>
6956 PR gcov-profile/82702
6957 * gcov.c (main): Handle intermediate files in a different
6959 (get_gcov_intermediate_filename): New function.
6960 (output_gcov_file): Remove support of intermediate files.
6961 (generate_results): Allocate intermediate file.
6962 (release_structures): Clean-up properly fn_end.
6963 (output_intermediate_file): Start iterating with line 1.
6965 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6967 PR tree-optimization/82929
6968 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6969 ops_swapped_p non-static data member.
6970 (store_immediate_info::store_immediate_info): Clear it.
6971 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6972 ops set ops_swapped_p.
6973 (count_multiple_uses): Handle ops_swapped_p.
6975 2017-11-10 Martin Liska <mliska@suse.cz>
6977 * coverage.c (coverage_init): Stream information about
6978 support of has_unexecuted_blocks.
6979 * doc/gcov.texi: Document that.
6980 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6981 * gcov.c (read_graph_file): Likewise.
6982 (output_line_beginning): Fix a small issue with
6985 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6987 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6988 reference of trivial component.
6990 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6993 * gimple-ssa-store-merging.c
6994 (pass_store_merging::terminate_all_aliasing_chains): For
6995 gimple_store_p stmts also call refs_output_dependent_p.
6997 PR rtl-optimization/82913
6998 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7001 2017-11-09 Jeff Law <law@redhat.com>
7003 * vr-values.h: New file with vr_values class.
7004 * tree-vrp.c: Include vr-values.h
7005 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7006 data objects into the vr_values class.
7007 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7008 (get_value_range): Make it a member function within vr_values class.
7009 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7010 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7011 (op_with_constant_singleton_value_range): Likewise.
7012 (extract_range_for_var_from_comparison_expr): Likewise.
7013 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7014 (extract_range_from_binary_expr): Likewise.
7015 (extract_range_from_unary_expr): Likewise.
7016 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7017 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7018 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7019 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7020 (compare_name_with_value, compare_names): Likewise.
7021 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7022 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7023 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7024 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7025 (extract_range_from_phi_node): Likewise.
7026 (simplify_truth_ops_using_ranges): Likewise.
7027 (simplify_div_or_mod_using_ranges): Likewise.
7028 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7029 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7030 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7031 (simplify_float_conversion_using_ranges): Likewise.
7032 (simplify_internal_call_using_ranges): Likewise.
7033 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7034 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7035 poitner via x_vr_values for calls into gimple folder.
7036 (vrp_initialize_lattice): Make this the vr_values ctor.
7037 (vrp_free_lattice): Make this the vr_values dtor.
7038 (set_vr_value): New function.
7039 (class vrp_prop): Add vr_values data member. Add various member
7040 functions as well as member functions that delegate to vr_values.
7041 (check_array_ref): Make a member function within vrp_prop class.
7042 (search_for_addr_array, vrp_initialize): Likewise.
7043 (vrp_finalize): Likewise. Revamp to avoid direct access to
7044 vr_value, values_propagated, etc.
7045 (check_array_bounds): Extract vrp_prop class instance pointer from
7046 walk info structure. Use it to call member functions.
7047 (check_all_array_refs): Make a member function within vrp_prop class.
7048 Smuggle class instance pointer via walk info structure.
7049 (x_vr_values): New local static.
7050 (vrp_valueize): Use x_vr_values to get class instance.
7051 (vr_valueize_1): Likewise.
7052 (class vrp_folder): Add vr_values data member. Add various member
7053 functions as well as member functions that delegate to vr_values.
7054 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7055 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7056 class instance from vr_values. Use it to call member functions.
7057 (vrp_dom_walker): Add vr_values data member.
7058 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7059 instance via x_vr_values.
7060 (identify_jump_threads): Accept vr_values as argument. Store
7061 it into the walker structure.
7062 (evrp_dom_walker): Add vr_values class data member. Add various
7064 (evrp_dom_walker::try_find_new_range): Use vr_values data
7065 member to access the memory allocator.
7066 (evrp_dom_walker::before_dom_children): Store vr_values class
7067 instance into the vrp_folder class.
7068 (evrp_dom_walker::push_value_range): Rework to avoid direct
7069 access to num_vr_values and vr_value.
7070 (evrp_dom_walker::pop_value_range): Likewise.
7071 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7072 Use vr_values to get to dump_all_value_ranges member function.
7073 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7074 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7075 Pass vr_values class instance down to identify_jump_threads.
7076 Remove call to vrp_free_lattice.
7077 (debug_all_value_ranges): Remove.
7079 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7080 (vrp_folder): Likewise.
7082 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7083 Get it from the existing bitmap instead.
7084 (vrp_intersect_ranges_1): Likewise.
7086 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7088 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7090 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7091 set bit_not_p to it.
7092 (imm_store_chain_info::coalesce_immediate_stores): Break group
7093 if bit_not_p is different.
7094 (count_multiple_uses, split_group,
7095 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7096 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7097 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7098 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7099 to store_immediate_info ctor.
7101 2017-11-09 Jim Wilson <jimw@sifive.com>
7103 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7104 (scan_prog_file): Likewise.
7106 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7108 * bb-reorder.c (max_entry_frequency): Remove.
7109 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7110 connect_traces, push_to_next_round_p): Remove prototypes.
7111 (find_traces_1_round): Use counts only.
7112 (push_to_next_round_p): Likewise.
7113 (find_traces): Likewise.
7114 (rotate_loop): Likewise.
7115 (find_traces_1_round): Likewise.
7116 (connect_traces): Likewise.
7117 (edge_order): Likewise.
7119 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7121 * config/arm/arm.c (output_return_instruction): Add comments to
7122 indicate requirement for cmse_nonsecure_entry return to account
7123 for the size of clearing instruction output here.
7124 (thumb_exit): Likewise.
7125 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7126 return in hardfloat mode.
7128 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7130 * config/rs6000/rs6000.c (machine_function): Add a bool,
7131 "toc_is_wrapped_separately".
7132 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7133 if it wasn't explicitly set or unset, we are optimizing for speed, and
7134 doing separate shrink-wrapping.
7135 (rs6000_get_separate_components): Enable the TOC component if
7136 saving the TOC register in the prologue.
7137 (rs6000_components_for_bb): Handle the TOC component.
7138 (rs6000_emit_prologue_components): Store the TOC register where needed.
7139 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7140 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7142 2017-11-09 Martin Jambor <mjambor@suse.cz>
7144 * ipa-param-manipulation.c: New file.
7145 * ipa-param-manipulation.h: Likewise.
7146 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7147 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7148 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7149 (ipa_parm_adjustment): Likewise.
7150 (ipa_parm_adjustment_vec): Likewise.
7151 (ipa_get_vector_of_formal_parms): Moved declaration to
7152 ipa-param-manipulation.h.
7153 (ipa_get_vector_of_formal_parm_types): Likewise.
7154 (ipa_modify_formal_parameters): Likewise.
7155 (ipa_modify_call_arguments): Likewise.
7156 (ipa_combine_adjustments): Likewise.
7157 (ipa_dump_param_adjustments): Likewise.
7158 (ipa_modify_expr): Likewise.
7159 (ipa_get_adjustment_candidate): Likewise.
7160 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7161 ipa-param-manipulation.c.
7162 (ipa_get_vector_of_formal_parm_types): Likewise.
7163 (ipa_modify_formal_parameters): Likewise.
7164 (ipa_modify_call_arguments): Likewise.
7165 (ipa_modify_expr): Likewise.
7166 (get_ssa_base_param): Likewise.
7167 (ipa_get_adjustment_candidate): Likewise.
7168 (index_in_adjustments_multiple_times_p): Likewise.
7169 (ipa_combine_adjustments): Likewise.
7170 (ipa_dump_param_adjustments): Likewise.
7171 * tree-sra.c: Also include ipa-param-manipulation.h
7172 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7175 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7176 Alan Hayward <alan.hayward@arm.com>
7177 David Sherwood <david.sherwood@arm.com>
7179 * doc/sourcebuild.texi (vect_masked_store): Document.
7181 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7182 Alan Hayward <alan.hayward@arm.com>
7183 David Sherwood <david.sherwood@arm.com>
7185 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7187 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7188 Alan Hayward <alan.hayward@arm.com>
7189 David Sherwood <david.sherwood@arm.com>
7191 * doc/sourcebuild.texi (vect_variable_length): Document.
7193 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7194 Alan Hayward <alan.hayward@arm.com>
7195 David Sherwood <david.sherwood@arm.com>
7197 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7199 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7200 Alan Hayward <alan.hayward@arm.com>
7201 David Sherwood <david.sherwood@arm.com>
7203 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7205 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7206 Alan Hayward <alan.hayward@arm.com>
7207 David Sherwood <david.sherwood@arm.com>
7209 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7210 previously undocumented selectors.
7211 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7213 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7215 * doc/rtl.texi (const_vector): Say that elements can be
7216 const_wide_ints too.
7217 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7218 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7219 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7220 * optabs.c (expand_vector_broadcast): Likewise.
7222 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7223 Alan Hayward <alan.hayward@arm.com>
7224 David Sherwood <david.sherwood@arm.com>
7226 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7227 scaled index even if the unscaled address was invalid.
7228 Don't increase the complexity of using a scale in that case.
7230 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7231 Alan Hayward <alan.hayward@arm.com>
7232 David Sherwood <david.sherwood@arm.com>
7234 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7235 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7236 UNITS_PER_WORD bytes.
7237 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7238 based on REGMODE_NATURAL_SIZE of the inner mode.
7239 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7240 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7241 * expmed.c (lowpart_bit_field_p): Divide the value up into
7242 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7243 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7244 whether something is likely to occupy more than one register.
7246 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7249 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7250 profile_count::adjust_for_ipa_scaling.
7251 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7252 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7254 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7256 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7258 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7259 (split_group): Add total_orig and total_new arguments, estimate the
7260 number of statements related to the store group without store merging
7261 and with store merging.
7262 (imm_store_chain_info::output_merged_store): Adjust split_group
7263 callers, punt if estimated number of statements with store merging
7264 is not smaller than estimated number of statements without it.
7266 (handled_load): Remove has_single_use checks.
7267 (pass_store_merging::process_store): Likewise.
7269 2017-11-09 Richard Biener <rguenther@suse.de>
7271 PR tree-optimization/82902
7272 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7274 2017-11-09 Martin Liska <mliska@suse.cz>
7277 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7280 2017-11-09 Martin Liska <mliska@suse.cz>
7282 PR tree-optimization/82669
7283 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7285 2017-11-09 Martin Liska <mliska@suse.cz>
7287 PR gcov-profile/48463
7288 * coverage.c (coverage_begin_function): Output also end locus
7289 of a function and information whether the function is
7291 * gcov-dump.c (tag_function): Parse and print the information.
7292 * gcov.c (INCLUDE_MAP): Add include.
7293 (INCLUDE_SET): Likewise.
7294 (struct line_info): Move earlier in the source file because
7295 of vector<line_info> in function_info structure.
7296 (line_info::line_info): Likewise.
7297 (line_info::has_block): Likewise.
7298 (struct source_info): Add new member index.
7299 (source_info::get_functions_at_location): New function.
7300 (function_info::group_line_p): New function.
7301 (output_intermediate_line): New function.
7302 (output_intermediate_file): Use the mentioned function.
7303 (struct function_start): New.
7304 (struct function_start_pair_hash): Likewise.
7305 (process_file): Add code that identifies group functions.
7306 Assign lines either to global or function scope.
7307 (generate_results): Skip artificial functions.
7308 (find_source): Assign index for each source file.
7309 (read_graph_file): Read new flag artificial and end_line.
7310 (add_line_counts): Assign it either to global of function scope.
7311 (accumulate_line_counts): Isolate core of the function to
7312 accumulate_line_info and call it for both function and global
7314 (accumulate_line_info): New function.
7315 (output_line_beginning): Fix GNU coding style.
7316 (print_source_line): New function.
7317 (output_line_details): Likewise.
7318 (output_function_details): Likewise.
7319 (output_lines): Iterate both source (global) scope and function
7321 (struct function_line_start_cmp): New class.
7322 * doc/gcov.texi: Reflect changes in documentation.
7324 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7327 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7328 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7329 and similarly for not instead of neg.
7331 2017-11-08 Andi Kleen <ak@linux.intel.com>
7333 * config/i386/i386.opt: Add -mforce-indirect-call.
7334 * config/i386/predicates.md: Check for flag_force_indirect_call.
7335 * doc/invoke.texi: Document -mforce-indirect-call
7337 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7339 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7341 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7342 config/riscv/riscv.c (predict.h): New include.
7343 (riscv_slow_unaligned_access_p): No longer static.
7344 (riscv_block_move_straight): Add require.
7345 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7347 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7350 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7351 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7352 nonimmediate_operand predicate for operand 1 instead of
7355 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7357 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7359 * config/aarch64/constraints.md (Uml): New constraint.
7360 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7363 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7365 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7366 of two vec_duplicates into a vec_concat.
7368 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7370 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7371 Simplify vec_merge of vec_duplicate and vec_concat.
7372 * config/aarch64/constraints.md (Utq): New constraint.
7373 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7376 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7378 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7379 Simplify vec_merge of vec_duplicate and const_vector.
7380 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7382 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7383 mode iterator. Update predicate on operand 1 to
7384 handle non-const_vec constants. Delete constraints.
7385 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7387 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7389 PR tree-optimization/78821
7390 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7392 (store_operand_info::store_operand_info): Initialize it to false.
7393 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7394 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7395 store in the group, and if chain_info is non-NULL, to ignore altogether
7397 (compatible_load_p): Fail if bit_not_p does not match.
7398 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7399 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7400 (pass_store_merging::process_store): Adjust
7401 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7402 call terminate_all_aliasing_chains newly when adding a store into
7403 a chain with non-NULL chain_info.
7405 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7407 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7409 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7411 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7413 (aarch64_layout_frame): Initialise emit_frame_chain.
7414 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7415 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7417 2017-11-08 Martin Liska <mliska@suse.cz>
7419 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7420 of gimple_call_internal_p.
7422 2017-11-07 Tom de Vries <tom@codesourcery.com>
7424 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7426 2017-11-07 Tom de Vries <tom@codesourcery.com>
7428 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7431 2017-11-08 Martin Liska <mliska@suse.cz>
7434 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7436 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7438 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7440 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7443 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7444 (*cmp<mode>_ccz_1): New insn with $k alternative.
7447 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7448 RTX_COMM_COMPARE as commutative as well.
7449 (ix86_binary_operator_ok): Formatting fix.
7450 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7451 *<code><mode>3<mask_name><round_saeonly_name>,
7452 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7453 *<s>mul<mode>3_highpart<mask_name>,
7454 *vec_widen_umult_even_v16si<mask_name>,
7455 *vec_widen_umult_even_v8si<mask_name>,
7456 *vec_widen_umult_even_v4si<mask_name>,
7457 *vec_widen_smult_even_v16si<mask_name>,
7458 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7459 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7460 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7461 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7462 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7463 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7464 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7465 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7466 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7467 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7468 ix86_binary_operator_ok. Formatting fixes.
7469 (*<plusminus_insn><mode>3<mask_name><round_name>,
7470 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7473 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7475 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7476 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7477 eq and ne if TARGET_ISEL.
7478 (cmp): New code_iterator.
7479 (UNS, UNSU_, UNSIK): New code_attrs.
7480 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7481 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7483 ("eq<mode>3"): ... this.
7484 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7486 ("ne<mode>3"): ... this.
7488 2017-11-07 Julia Koval <julia.koval@intel.com>
7491 * common/config/i386/i386-common.c
7492 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7493 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7494 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7495 * config/i386/i386.opt: Ditto.
7496 * config/i386/i386.c (ix86_target_string): Ditto.
7497 (ix86_option_override_internal): Ditto.
7498 (ix86_init_mpx_builtins): Move MPX to args2.
7499 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7500 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7501 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7502 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7503 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7504 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7505 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7507 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7510 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7511 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7512 (zero-extendsidi peephole2): Remove peephole.
7514 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7517 * stor-layout.c: Include attribs.h.
7518 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7519 explicit lookup of "aligned" attribute.
7521 2017-11-07 Andrew Waterman <andrew@sifive.com>
7523 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7524 (riscv_expand_block_move): Likewise.
7525 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7527 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7528 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7529 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7530 (riscv_adjust_block_mem): Likewise.
7531 (riscv_block_move_loop): Likewise.
7532 (riscv_expand_block_move): Likewise.
7533 * config/riscv/riscv.md (movmemsi): New pattern.
7535 2017-11-07 Michael Clark <michaeljclark@mac.com>
7537 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7538 (MUSL_DYNAMIC_LINKER): Likewise.
7540 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7542 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7543 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7546 2017-11-07 Richard Biener <rguenther@suse.de>
7548 * match.pd: Fix build.
7550 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7551 Jackson Woodruff <jackson.woodruff@arm.com>
7553 PR tree-optimization/71026
7554 * match.pd: Canonicalize negate in division.
7556 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7559 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7561 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7563 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7564 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7566 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7568 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7569 non-scalar integral types.
7570 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7571 (-(A-B), -(~A)): New transformations.
7573 2017-11-07 Tom de Vries <tom@codesourcery.com>
7575 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7576 semicolon after "do {} while (0)".
7577 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7578 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7579 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7580 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7581 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7583 2017-11-07 Tom de Vries <tom@codesourcery.com>
7585 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7586 after "do {} while (0)".
7587 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7588 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7589 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7590 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7591 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7593 2017-11-07 Tom de Vries <tom@codesourcery.com>
7596 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7598 (arm_rtx_costs_internal): Add missing semicolon after
7599 HANDLE_NARROW_SHIFT_ARITH call.
7601 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7603 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7604 disable isel if it was not set explicitly.
7606 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7608 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7609 (add_type_attribute) likewise.
7611 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7613 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7614 of struct ix86_frame.
7615 (ix86_initial_elimination_offset): Likewise.
7616 (ix86_expand_split_stack_prologue): Likewise.
7618 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7620 * tree-vrp.h (enum value_range_type): Update stale comment.
7622 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7623 Alan Hayward <alan.hayward@arm.com>
7624 David Sherwood <david.sherwood@arm.com>
7626 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7627 (aarch64_expand_vec_perm_const): Take the number of units too.
7628 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7629 (aarch64_expand_vec_perm_const): Likewise.
7630 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7631 (vec_perm<mode>): Update accordingly.
7633 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7634 Alan Hayward <alan.hayward@arm.com>
7635 David Sherwood <david.sherwood@arm.com>
7637 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7638 Take the number of units too.
7639 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7640 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7641 but check for a vector mode before rather than after the call.
7642 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7643 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7644 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7645 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7646 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7647 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7648 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7649 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7650 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7651 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7652 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7653 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7654 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7655 (aarch64_sqdmull2_n<mode>): Update accordingly.
7657 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7658 Alan Hayward <alan.hayward@arm.com>
7659 David Sherwood <david.sherwood@arm.com>
7661 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7662 the number of units too.
7663 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7664 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7665 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7666 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7667 (vec_store_lanesxi<mode>): Update accordingly.
7669 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7670 Alan Hayward <alan.hayward@arm.com>
7671 David Sherwood <david.sherwood@arm.com>
7673 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7674 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7675 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7676 of units rather than the mode.
7677 * config/aarch64/iterators.md (nunits): New mode attribute.
7678 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7679 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7680 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7681 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7682 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7683 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7684 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7685 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7686 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7687 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7688 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7689 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7690 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7691 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7692 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7693 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7694 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7695 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7696 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7697 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7698 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7699 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7700 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7701 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7702 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7703 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7704 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7705 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7706 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7707 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7708 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7709 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7710 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7711 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7712 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7713 (aarch64_simd_vec_setv2di): Likewise.
7715 2017-11-06 Carl Love <cel@us.ibm.com>
7717 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7719 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7720 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7722 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7724 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7725 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7726 Add power 8 macro expansions.
7727 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7728 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7729 power 8 instructions. (VSX_XXBR): Add iterator.
7731 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7733 * config/arm/arm.md (predicable_short_it): Change default to "no",
7734 improve documentation, remove uses that are identical to the default.
7735 (enabled_for_depr_it): Rename to enabled_for_short_it.
7736 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7737 Remove default uses.
7738 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7739 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7740 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7741 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7743 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7746 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7747 float128 helper macros, which are no longer used after deleting
7748 the old 'q' built-in functions, and moving the round to odd
7749 built-in functions to being special built-in functions.
7750 (BU_FLOAT128_2): Likewise.
7751 (BU_FLOAT128_1_HW): Likewise.
7752 (BU_FLOAT128_2_HW): Likewise.
7753 (BU_FLOAT128_3_HW): Likewise.
7754 (FABSQ): Delete old 'q' built-in functions.
7755 (COPYSIGNQ): Likewise.
7756 (SQRTF128_ODD): Move round to odd built-in functions to be
7757 special built-in functions, so that we can handle
7758 -mabi=ieeelongdouble.
7759 (TRUNCF128_ODD): Likewise.
7760 (ADDF128_ODD): Likewise.
7761 (SUBF128_ODD): Likewise.
7762 (MULF128_ODD): Likewise.
7763 (DIVF128_ODD): Likewise.
7764 (FMAF128_ODD): Likewise.
7765 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7766 built-in names to 'f128'.
7767 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7768 old 'q' built-in functions, as the machine independent code for
7769 'f128' built-in functions handles this.
7770 (rs6000_expand_builtin): Add expansion for float128 round to odd
7771 functions, keying off on -mabi=ieeelongdouble of whether to use
7772 the KFmode or TFmode variant.
7773 (rs6000_init_builtins): Initialize the _Float128 round to odd
7775 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7776 _Float128 'q' built-in functions are now mapped into the new
7777 'f128' built-in functions.
7779 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7781 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7782 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7784 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7787 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7788 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7789 * toplev.c: Include "ipa-fnsummary.h".
7790 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7792 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7794 PR tree-optimization/82838
7795 * gimple-ssa-store-merging.c
7796 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7797 on a separate gimple_seq which is then appended to seq.
7799 2017-11-06 Jeff Law <law@redhat.com>
7802 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7803 (get_probe_interval): New functions.
7804 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7805 (ix86_adjust_stack_and_probe): Likewise.
7806 (output_adjust_stack_and_probe): Likewise.
7807 (ix86_emit_probe_stack_range): Likewise.
7808 (ix86_expand_prologue): Likewise.
7810 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7812 PR tree-optimization/82816
7813 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7814 if the modes of the two types are the same.
7815 (convert_plusminus_to_widen): Likewise.
7817 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7819 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7821 (usadv16qi): New define_expand.
7822 (usadv8hi): New define_expand.
7824 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7827 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7829 (inline_transform): Likewise.
7830 * predict.c (counts_to_freqs): Remove useless conditional.
7831 * profile-count.h (profile_count::apply_scale): Move sanity check.
7832 * tree-inline.c (copy_bb): Always scale.
7833 (copy_cfg_body): Likewise.
7835 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7838 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7840 (*cmp_ite0): Add enabled_for_depr_it attribute.
7841 (*cmp_ite1): Likewise.
7843 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7845 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7848 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7850 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7851 return the value by pointer.
7852 (extract_range_from_multiplicative_op_1): Update accordingly.
7853 Return as soon as an operation fails.
7855 2017-11-05 Tom de Vries <tom@codesourcery.com>
7858 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7859 (DEF_SANITIZER_BUILTIN): ... here.
7860 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7861 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7863 2017-11-05 Tom de Vries <tom@codesourcery.com>
7866 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7868 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7869 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7870 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7872 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7873 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7874 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7876 2017-11-05 Tom de Vries <tom@codesourcery.com>
7879 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7882 2017-11-04 Michael Clark <michaeljclark@mac.com>
7884 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7885 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7887 (*addsi3_extended): Likewise.
7888 (*addsi3_extended2): Likewise.
7889 (<optab>si3): Likewise.
7890 (<optab>di3): Likewise.
7891 (<optab><mode>3): Likewise.
7892 (<*optabe>si3_internal): Likewise.
7893 (zero_extendqi<SUPERQI:mode>2): Likewise.
7894 (*add<mode>hi3): Likewise.
7895 (*xor<mode>hi3): Likewise.
7896 (<optab>di3): Likewise.
7897 (*<optab>si3_extend): Likewise.
7898 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7899 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7900 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7902 2017-11-04 Andrew Waterman <andrew@sifive.com>
7904 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7905 TARGET_STRICT_ALIGN based upon -mtune argument.
7907 2017-11-04 Andrew Waterman <andrew@sifive.com>
7909 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7911 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7913 * config/i386/i386.c (choose_basereg): Use optional scratch
7914 register and add assertion.
7915 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7916 needed, and don't allocate stack.
7917 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7918 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7919 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7921 2017-11-03 Jeff Law <law@redhat.com>
7923 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7924 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7925 to probe at the start of a noreturn function.
7927 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7929 PR tree-optimization/78821
7930 * gimple-ssa-store-merging.c: Update the file comment.
7931 (MAX_STORE_ALIAS_CHECKS): Define.
7932 (struct store_operand_info): New type.
7933 (store_operand_info::store_operand_info): New constructor.
7934 (struct store_immediate_info): Add rhs_code and ops data members.
7935 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7936 and op1r arguments to the ctor, initialize corresponding data members.
7937 (struct merged_store_group): Add load_align_base and load_align
7939 (merged_store_group::merged_store_group): Initialize them.
7940 (merged_store_group::do_merge): Update them.
7941 (merged_store_group::apply_stores): Pick the constant for
7942 encode_tree_to_bitpos from one of the two operands, or skip
7943 encode_tree_to_bitpos if neither operand is a constant.
7944 (class pass_store_merging): Add process_store method decl. Remove
7945 bool argument from terminate_all_aliasing_chains method decl.
7946 (pass_store_merging::terminate_all_aliasing_chains): Remove
7947 var_offset_p argument and corresponding handling.
7948 (stmts_may_clobber_ref_p): New function.
7949 (compatible_load_p): New function.
7950 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7951 if there is overlap and rhs_code is not INTEGER_CST. For
7952 non-overlapping stores terminate group if rhs is not mergeable.
7953 (get_alias_type_for_stmts): Change first argument from
7954 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7955 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7956 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7958 (get_location_for_stmts): Change first argument from
7959 auto_vec<gimple *> & to vec<gimple *> &.
7960 (struct split_store): Remove orig_stmts data member, add orig_stores.
7961 (split_store::split_store): Create orig_stores rather than orig_stmts.
7962 (find_constituent_stmts): Renamed to ...
7963 (find_constituent_stores): ... this. Change second argument from
7964 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7965 to info structures rather than the statements.
7966 (split_group): Rename ALLOW_UNALIGNED argument to
7967 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7968 it. Adjust find_constituent_stores caller.
7969 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7970 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7971 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7972 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7973 (mem_valid_for_store_merging): New function.
7974 (handled_load): New function.
7975 (pass_store_merging::process_store): New method.
7976 (pass_store_merging::execute): Use process_store method. Adjust
7977 terminate_all_aliasing_chains caller.
7979 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7981 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7982 Return true for more constants, symbols and label references.
7983 (aarch64_valid_floating_const): Remove unused function.
7985 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7988 * config/aarch64/aarch64.c (aarch64_layout_frame):
7989 Undo forcing of LR at bottom of frame.
7991 2017-11-03 Jeff Law <law@redhat.com>
7994 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7995 for int_registers_saved.
7997 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7998 extracted from tree-ssa-dom.c.
7999 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8000 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8001 (record_equivalences_from_incoming_edge): Add additional argument
8002 to single_pred_edge_ignoring_loop_edges call.
8003 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8004 (uncprop_dom_walker::before_dom_children): Add additional argument
8005 to single_pred_edge_ignoring_loop_edges call.
8006 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8007 single_pred_edge_ignoring_loop_edges rather than open coding.
8008 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8010 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8012 * match.pd (-(-A)): Rewrite.
8014 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8016 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8017 (rs6000_emit_int_cmove): New declaration.
8018 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8019 (rs6000_emit_sISEL): Delete.
8020 (rs6000_emit_int_cmove): Make non-static.
8021 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8022 instead of rs6000_emit_sISEL.
8024 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8026 * asan.c (create_cond_insert_point): Maintain profile.
8027 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8029 * basic-block.h (struct basic_block_def): Remove frequency.
8030 (EDGE_FREQUENCY): Use to_frequency
8031 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8033 (find_traces): Update to use to_frequency.
8034 (find_traces_1_round): Likewise; use only IPA counts.
8035 (bb_to_key): Likewise.
8036 (connect_traces): Use IPA counts only.
8037 (copy_bb_p): Update to use to_frequency.
8038 (fix_up_crossing_landing_pad): Likewise.
8039 (sanitize_hot_paths): Likewise.
8040 * bt-load.c (basic_block_freq): Likewise.
8041 * cfg.c (init_flow): Set count_max to uninitialized.
8042 (check_bb_profile): Remove frequencies; check counts.
8043 (dump_bb_info): Do not dump frequencies.
8044 (update_bb_profile_for_threading): Update counts only.
8045 (scale_bbs_frequencies_int): Likewise.
8046 (MAX_SAFE_MULTIPLIER): Remove.
8047 (scale_bbs_frequencies_gcov_type): Update counts only.
8048 (scale_bbs_frequencies_profile_count): Update counts only.
8049 (scale_bbs_frequencies): Update counts only.
8050 * cfg.h (struct control_flow_graph): Add count-max.
8051 (update_bb_profile_for_threading): Update prototype.
8052 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8053 (find_many_sub_basic_blocks): Likewise.
8054 * cfgcleanup.c (try_forward_edges): Likewise.
8055 (try_crossjump_to_edge): Likewise.
8056 * cfgexpand.c (expand_gimple_cond): Likewise.
8057 (expand_gimple_tailcall): Likewise.
8058 (construct_init_block): Likewise.
8059 (construct_exit_block): Likewise.
8060 * cfghooks.c (verify_flow_info): Check consistency of counts.
8061 (dump_bb_for_graph): Do not dump frequencies.
8062 (split_block_1): Do not update frequencies.
8063 (split_edge): Do not update frequencies.
8064 (make_forwarder_block): Do not update frequencies.
8065 (duplicate_block): Do not update frequencies.
8066 (account_profile_record): Do not update frequencies.
8067 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8068 for global heuristics.
8069 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8070 (expected_loop_iterations_unbounded): Use counts only.
8071 * cfgloopmanip.c (scale_loop_profile): Simplify.
8072 (create_empty_loop_on_edge): Simplify
8074 (duplicate_loop_to_header_edge): Simplify
8075 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8076 (update_br_prob_note): Take care of removing note when profile
8078 (relink_block_chain): Do not dump frequency.
8079 (rtl_account_profile_record): Use to_frequency.
8080 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8081 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8082 (cgraph_update_edges_for_call_stmt_node): Likewise.
8083 (cgraph_edge::verify_count_and_frequency): Update.
8084 (cgraph_node::verify_node): Temporarily disable frequency verification.
8085 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8086 to_cgraph_frequency.
8087 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8088 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8090 (cgraph_node::expand_thunk): Update profile.
8091 * except.c (dw2_build_landing_pads): Do not update frequency.
8092 * final.c (compute_alignments): Use to_frequency.
8093 (dump_basic_block_info): Do not dump frequency.
8094 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8095 (dump_gimple_bb_header): Do not dump frequency.
8096 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8098 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8099 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8100 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8101 (init_before_recovery): Do not update frequency.
8102 (sched_create_recovery_edges): Do not update frequency.
8103 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8104 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8105 (ipa_cp_c_finalize): Set max_count to uninitialized.
8106 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8107 (param_change_prob): Use counts.
8108 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8110 * ipa-split.c (consider_split): Use to_frequency.
8111 (split_function): Use to_frequency.
8112 * ira-build.c (loop_compare_func): Likewise.
8113 (mark_loops_for_removal): Likewise.
8114 (mark_all_loops_for_removal): Likewise.
8115 * loop-doloop.c (doloop_modify): Do not update frequency.
8116 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8118 * lto-streamer-in.c (input_function): Update count_max.
8119 * omp-expand.c (expand_omp_taskreg): Update count_max.
8120 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8121 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8122 (maybe_hot_count_p): Use ipa counts only.
8123 (maybe_hot_bb_p): Simplify.
8124 (maybe_hot_edge_p): Simplify.
8125 (probably_never_executed): Do not take frequency argument.
8126 (probably_never_executed_bb_p): Do not pass frequency.
8127 (probably_never_executed_edge_p): Likewise.
8128 (combine_predictions_for_bb): Check that profile is nonzero.
8129 (propagate_freq): Do not set frequency.
8130 (drop_profile): Simplify.
8131 (counts_to_freqs): Simplify.
8132 (expensive_function_p): Use to_frequency.
8133 (propagate_unlikely_bbs_forward): Simplify.
8134 (determine_unlikely_bbs): Simplify.
8135 (estimate_bb_frequencies): Add hack to silence graphite issues.
8136 (compute_function_frequency): Use ipa counts.
8137 (pass_profile::execute): Update.
8138 (rebuild_frequencies): Use counts only.
8139 (force_edge_cold): Use counts only.
8140 * profile-count.c (profile_count::dump): Dump new count types.
8141 (profile_count::differs_from_p): Check compatiblity.
8142 (profile_count::to_frequency): New function.
8143 (profile_count::to_cgraph_frequency): New function.
8144 * profile-count.h (struct function): Declare.
8145 (enum profile_quality): Add profile_guessed_local and
8146 profile_guessed_global0.
8147 (class profile_proability): Decrease number of bits to 29;
8148 update from_reg_br_prob_note and to_reg_br_prob_note.
8149 (class profile_count: Update comment; decrease number of bits
8150 to 61. Check compatibility.
8151 (profile_count::compatible_p): New private member function.
8152 (profile_count::ipa_p): New member function.
8153 (profile_count::operator<): Handle global zero correctly.
8154 (profile_count::operator>): Handle global zero correctly.
8155 (profile_count::operator<=): Handle global zero correctly.
8156 (profile_count::operator>=): Handle global zero correctly.
8157 (profile_count::nonzero_p): New member function.
8158 (profile_count::force_nonzero): New member function.
8159 (profile_count::max): New member function.
8160 (profile_count::apply_scale): Handle IPA scalling.
8161 (profile_count::guessed_local): New member function.
8162 (profile_count::global0): New member function.
8163 (profile_count::ipa): New member function.
8164 (profile_count::to_frequency): Declare.
8165 (profile_count::to_cgraph_frequency): Declare.
8166 * profile.c (OVERLAP_BASE): Delete.
8167 (compute_frequency_overlap): Delete.
8168 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8169 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8170 * sched-ebb.c (rank): Use counts only.
8171 * shrink-wrap.c (handle_simple_exit): Use counts only.
8172 (try_shrink_wrapping): Use counts only.
8173 (place_prologue_for_one_component): Use counts only.
8174 * tracer.c (find_best_predecessor): Use to_frequency.
8175 (find_trace): Use to_frequency.
8176 (tail_duplicate): Use to_frequency.
8177 * trans-mem.c (expand_transaction): Do not update frequency.
8178 * tree-call-cdce.c: Do not update frequency.
8179 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8180 (gimple_merge_blocks): Likewise.
8181 (gimple_split_edge): Likewise.
8182 (gimple_duplicate_sese_region): Likewise.
8183 (gimple_duplicate_sese_tail): Likewise.
8184 (move_sese_region_to_fn): Likewise.
8185 (gimple_account_profile_record): Likewise.
8186 (insert_cond_bb): Likewise.
8187 * tree-complex.c (expand_complex_div_wide): Likewise.
8188 * tree-eh.c (lower_resx): Update profile.
8189 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8191 (initialize_cfun): Do not initialize frequencies
8192 (freqs_to_counts): Delete.
8193 (copy_cfg_body): Ignore count parameter.
8194 (copy_body): Update.
8195 (expand_call_inline): Update count_max.
8196 (optimize_inline_calls): Update count_max.
8197 (tree_function_versioning): Update count_max.
8198 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8199 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8201 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8202 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8203 (try_peel_loop): Likewise.
8204 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8206 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8207 (tree_transform_and_unroll_loop): Do not use frequencies
8208 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8209 Use reliable prediction only.
8210 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8211 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8212 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8213 probability scaling.
8214 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8215 not update frequency
8216 (any_remaining_duplicated_blocks): Likewise.
8217 (update_profile): Likewise.
8218 (estimated_freqs_path): Delete.
8219 (freqs_to_counts_path): Delete.
8220 (clear_counts_path): Delete.
8221 (ssa_fix_duplicate_block_edges): Likewise.
8222 (duplicate_thread_path): Likewise.
8223 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8224 * tree-tailcall.c (decrease_profile): Do not update frequency.
8225 (eliminate_tail_call): Likewise.
8226 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8227 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8228 (optimize_mask_stores): Likewise.
8229 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8230 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8231 (ubsan_expand_ptr_ifn): Update profile.
8232 * value-prof.c (gimple_ic): Simplify.
8233 * value-prof.h (gimple_ic): Update prototype.
8234 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8235 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8237 (want_inline_self_recursive_call_p): Likewise.
8238 (resolve_noninline_speculation): Only cummulate defined counts.
8239 (inline_small_functions): Use nonzero_p.
8240 (ipa_inline): Do not access freed node.
8242 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8244 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8245 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8247 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8249 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8250 non-legitimate address.
8252 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8254 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8255 (*lt0_<mode>di, *lt0_<mode>si): New.
8257 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8259 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8260 TARGET_PAIRED_FLOAT.
8262 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8263 Jim Wilson <jim.wilson@linaro.org>
8265 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8266 * config/aarch64/aarch64-tune.md: Regenerated.
8267 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8268 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8270 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8272 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8273 (arc_expand_prologue): Restore blink for millicode.
8274 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8276 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8279 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8280 gen_vec_duplicate after forcing the scalar into a register.
8282 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8284 * combine (try_combine): Print the insns input to try_combine to the
8287 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8290 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8291 Remove second argument from aarch64_process_target_attr call.
8292 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8294 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8296 (aarch64_handle_attr_arch): Remove second argument.
8297 (aarch64_handle_attr_cpu): Ditto.
8298 (aarch64_handle_attr_tune): Ditto.
8299 (aarch64_handle_attr_isa_flags): Ditto.
8300 (aarch64_process_one_target_attr): Ditto.
8301 (aarch64_process_target_attr): Ditto.
8302 (aarch64_option_valid_attribute_p): Remove second argument.
8303 on aarch64_process_target_attr call.
8305 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8307 * diagnostic.c: Include "selftest-diagnostic.h".
8308 (selftest::assert_location_text): New function.
8309 (selftest::test_diagnostic_get_location_text): New function.
8310 (selftest::diagnostic_c_tests): Call it.
8312 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8314 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8315 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8316 (class selftest::test_diagnostic_context): Move to...
8317 * selftest-diagnostic.c: New file.
8318 * selftest-diagnostic.h: New file.
8320 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8322 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8323 offset range for FT32B.
8324 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8325 * config/ft32/ft32.md: Add TARGET_NOPM.
8326 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8327 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8329 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8331 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8333 2017-11-02 Jeff Law <law@redhat.com>
8335 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8336 virtual keyword on FINAL OVERRIDE members.
8338 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8339 virtuals together. Add virtual destructor.
8340 (substitute_and_fold_engine): Similarly.
8342 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8344 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8346 2017-11-02 Richard Biener <rguenther@suse.de>
8348 PR tree-optimization/82795
8349 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8351 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8353 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8355 (gcc_GAS_CHECK_FEATURE): Remove.
8356 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8357 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8358 * configure: Regenerate.
8360 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8362 * config/arc/arc.c (hwloop_optimize): Account for empty
8365 2017-11-02 Richard Biener <rguenther@suse.de>
8368 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8369 Truncate ARRAY_REF index and element size.
8371 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8373 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8375 2017-11-01 Jeff Law <law@redhat.com>
8377 * tree-ssa-ccp.c (ccp_folder): New class derived from
8378 substitute_and_fold_engine.
8379 (ccp_folder::get_value): New member function.
8380 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8381 (ccp_fold_stmt): Remove prototype.
8382 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8383 * tree-ssa-copy.c (copy_folder): New class derived from
8384 substitute_and_fold_engine.
8385 (copy_folder::get_value): Renamed from get_value.
8386 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8387 * tree-vrp.c (vrp_folder): New class derived from
8388 substitute_and_fold_engine.
8389 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8390 (vrp_folder::get_value): New member function.
8391 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8392 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8393 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8394 provide a class interface to folder/substitute routines.
8395 (ssa_prop_fold_stmt_fn): Remove typedef.
8396 (ssa_prop_get_value_fn): Likewise.
8397 (subsitute_and_fold): Remove prototype.
8398 (replace_uses_in): Likewise.
8399 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8400 Renamed from replace_uses_in. Call the virtual member function
8401 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8402 (substitute_and_fold_dom_walker): Remove initialization of
8403 data member entries for calbacks. Add substitute_and_fold_engine
8404 member and initialize it.
8405 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8406 member functions for get_value, replace_phi_args_in c
8407 replace_uses_in, and fold_stmt calls.
8408 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8409 substitute_and_fold. Remove assert. Update ctor call.
8411 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8412 (ssa_prop_visit_phi_fn): Likewise.
8413 (class ssa_propagation_engine): New class to provide an interface
8415 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8417 (ssa_prop_visit_phi): Likewise.
8418 (ssa_propagation_engine::simulate_stmt): Moved into class.
8419 Call visit_phi/visit_stmt from the class rather than via
8420 file scoped static variables.
8421 (ssa_propagation_engine::simulate_block): Moved into class.
8422 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8423 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8424 set file scoped statics for the visit_stmt/visit_phi callbacks.
8425 * tree-complex.c (complex_propagate): New class derived from
8426 ssa_propagation_engine.
8427 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8428 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8429 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8431 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8432 ssa_propagation_engine.
8433 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8434 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8435 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8436 * tree-ssa-copy.c (copy_prop): New class derived from
8437 ssa_propagation_engine.
8438 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8439 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8440 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8441 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8442 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8443 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8444 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8446 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8448 PR rtl-optimization/82778
8449 PR rtl-optimization/82597
8450 * compare-elim.c (struct comparison): Add in_a_setter field.
8451 (find_comparison_dom_walker::before_dom_children): Remove killed
8452 bitmap and df_simulate_find_defs call, instead walk the defs.
8453 Compute last_setter and initialize in_a_setter. Merge definitions
8454 with first initialization for a few variables.
8455 (try_validate_parallel): Use insn_invalid_p instead of
8456 recog_memoized. Return insn rather than just the pattern.
8457 (try_merge_compare): Fix up comment. Don't uselessly test if
8458 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8460 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8463 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8464 Alan Hayward <alan.hayward@arm.com>
8465 David Sherwood <david.sherwood@arm.com>
8467 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8468 aarch64_hard_regno_nregs to get the number of registers
8471 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8472 Alan Hayward <alan.hayward@arm.com>
8473 David Sherwood <david.sherwood@arm.com>
8475 * config/aarch64/constraints.md (Upl): Rename to...
8477 * config/aarch64/aarch64.md
8478 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8481 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8482 Alan Hayward <alan.hayward@arm.com>
8483 David Sherwood <david.sherwood@arm.com>
8485 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8486 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8489 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8490 Alan Hayward <alan.hayward@arm.com>
8491 David Sherwood <david.sherwood@arm.com>
8493 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8494 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8495 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8497 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8498 of a single element.
8499 * config/aarch64/iterators.md: Add a comment above the permute
8500 unspecs to say that they are generated directly by
8501 aarch64_expand_vec_perm_const.
8502 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8504 2017-11-01 Nathan Sidwell <nathan@acm.org>
8506 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8508 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8510 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8511 and medany code models, and describe what they do.
8513 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8515 Revert accidental duplicate:
8517 * combine.c (can_change_dest_mode): Reject changes in
8518 REGMODE_NATURAL_SIZE.
8520 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8522 PR rtl-optimization/64682
8523 PR rtl-optimization/69567
8524 PR rtl-optimization/69737
8525 PR rtl-optimization/82683
8526 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8527 register mentioned in the note, drop the note, unless it came from I3,
8528 in which case it should go to I3 again.
8530 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8532 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8533 and return false if not.
8534 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8536 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8538 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8539 range is known to be empty.
8541 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8542 Alan Hayward <alan.hayward@arm.com>
8543 David Sherwood <david.sherwood@arm.com>
8545 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8546 and CONST_VECTOR_NUNITS instead of computing the number of units from
8547 the byte sizes of the vector and element.
8548 (simplify_binary_operation_1): Likewise.
8549 (simplify_const_binary_operation): Likewise.
8550 (simplify_ternary_operation): Likewise.
8552 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8553 Alan Hayward <alan.hayward@arm.com>
8554 David Sherwood <david.sherwood@arm.com>
8556 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8557 (int_mem_offset): ...this new function.
8558 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8559 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8560 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8563 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8564 Alan Hayward <alan.hayward@arm.com>
8565 David Sherwood <david.sherwood@arm.com>
8567 * lower-subreg.c (interesting_mode_p): New function.
8568 (compute_costs, find_decomposable_subregs, decompose_register)
8569 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8570 (resolve_clobber, dump_choices): Use it.
8572 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8573 Alan Hayward <alan.hayward@arm.com>
8574 David Sherwood <david.sherwood@arm.com>
8576 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8578 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8579 Alan Hayward <alan.hayward@arm.com>
8580 David Sherwood <david.sherwood@arm.com>
8582 * alias.c (find_base_value, find_base_term): Only process integer
8583 truncations. Check the precision rather than the size.
8585 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8586 Alan Hayward <alan.hayward@arm.com>
8587 David Sherwood <david.sherwood@arm.com>
8589 * machmode.h (is_narrower_int_mode): New function
8590 * optabs.c (expand_float, expand_fix): Use it.
8591 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8593 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8594 Alan Hayward <alan.hayward@arm.com>
8595 David Sherwood <david.sherwood@arm.com>
8597 * rtl.h (narrower_subreg_mode): New function.
8598 * ira-color.c (update_costs_from_allocno): Use it.
8600 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8601 Alan Hayward <alan.hayward@arm.com>
8602 David Sherwood <david.sherwood@arm.com>
8604 * optabs-query.h (convert_optab_p): New function, split out from...
8605 (convert_optab_handler): ...here.
8606 (widening_optab_handler): Delete.
8607 (find_widening_optab_handler): Remove permit_non_widening parameter.
8608 (find_widening_optab_handler_and_mode): Likewise. Provide an
8609 override that operates on mode class wrappers.
8610 * optabs-query.c (widening_optab_handler): Delete.
8611 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8612 parameter. Assert that the two modes are the same class and that
8613 the "from" mode is narrower than the "to" mode. Use
8614 convert_optab_handler instead of widening_optab_handler.
8615 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8616 instead of widening_optab_handler.
8617 * expr.c (expand_expr_real_2): Update calls to
8618 find_widening_optab_handler.
8619 * optabs.c (expand_widen_pattern_expr): Likewise.
8620 (expand_binop_directly): Take the insn_code as a parameter.
8621 (expand_binop): Only call find_widening_optab_handler for
8622 conversion optabs; use optab_handler otherwise. Update calls
8623 to find_widening_optab_handler and expand_binop_directly.
8624 Use convert_optab_handler instead of widening_optab_handler.
8625 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8626 find_widening_optab_handler and use scalar_mode rather than
8628 (convert_plusminus_to_widen): Likewise.
8630 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8631 Alan Hayward <alan.hayward@arm.com>
8632 David Sherwood <david.sherwood@arm.com>
8634 * machmode.h (fixed_size_mode): New class.
8635 * rtl.h (get_pool_mode): Return fixed_size_mode.
8636 * gengtype.c (main): Add fixed_size_mode.
8637 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8638 (get_raw_arg_mode): Likewise.
8639 * doc/tm.texi: Regenerate.
8640 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8641 * targhooks.c (default_get_reg_raw_mode): Likewise.
8642 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8643 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8644 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8645 (msp430_get_raw_result_mode): Likewise.
8646 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8647 * dbxout.c (dbxout_parms): Require fixed-size modes.
8648 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8649 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8650 * omp-low.c (lower_oacc_reductions): Likewise.
8651 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8652 (simplify_subreg): Update accordingly.
8653 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8654 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8655 that aren't fixed-size.
8656 (get_pool_mode): Return a fixed_size_mode.
8657 (output_constant_pool_2): Take a fixed_size_mode.
8659 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8660 Alan Hayward <alan.hayward@arm.com>
8661 David Sherwood <david.sherwood@arm.com>
8663 * doc/rtl.texi (vec_series): Document.
8664 (const): Say that the operand can be a vec_series.
8665 * rtl.def (VEC_SERIES): New rtx code.
8666 * rtl.h (const_vec_series_p_1): Declare.
8667 (const_vec_series_p): New function.
8668 * emit-rtl.h (gen_const_vec_series): Declare.
8669 (gen_vec_series): Likewise.
8670 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8671 (gen_vec_series): Likewise.
8672 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8673 * simplify-rtx.c (simplify_unary_operation): Handle negations
8675 (simplify_binary_operation_series): New function.
8676 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8677 (test_vector_ops_series): New function.
8678 (test_vector_ops): Call it.
8679 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8680 gen_const_vec_series.
8681 (altivec_lvsr): Likewise.
8682 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8684 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8685 Alan Hayward <alan.hayward@arm.com>
8686 David Sherwood <david.sherwood@arm.com>
8688 * doc/rtl.texi (const): Update description of address constants.
8689 Say that vector constants are allowed too.
8690 * common.md (E, F): Use CONSTANT_P instead of checking for
8692 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8693 checking for CONST_VECTOR.
8694 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8696 * expr.c (expand_expr_real_2): Check for vector modes instead
8697 of checking for CONST_VECTOR.
8698 * rtl.h (const_vec_p): New function.
8699 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8700 (unwrap_const_vec_duplicate): Handle them here too.
8702 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8703 David Malcolm <dmalcolm@redhat.com>
8704 Alan Hayward <alan.hayward@arm.com>
8705 David Sherwood <david.sherwood@arm.com>
8707 * rtl.h (vec_duplicate_p): New function.
8708 * selftest-rtl.c (assert_rtx_eq_at): New function.
8709 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8710 (assert_rtx_eq_at): Declare.
8711 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8712 * selftest-run-tests.c (selftest::run_tests): Call it.
8713 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8714 (simplify_unary_operation_1): Recursively handle vector duplicates.
8715 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8717 (simplify_subreg): Handle subregs of vector duplicates.
8718 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8719 (selftest::simplify_rtx_c_tests): New functions.
8721 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8722 Alan Hayward <alan.hayward@arm.com>
8723 David Sherwood <david.sherwood@arm.com>
8725 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8726 (gen_vec_duplicate): Likewise.
8727 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8729 (gen_const_vector): ...here.
8730 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8731 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8732 whose elements are all equal.
8733 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8734 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8735 (simplify_relational_operation): Likewise.
8736 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8738 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8739 (aarch64_expand_vector_init): Likewise.
8740 * config/arm/arm.c (neon_vdup_constant): Likewise.
8741 (neon_expand_vector_init): Likewise.
8742 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8743 (arm_block_set_unaligned_vect): Likewise.
8744 (arm_block_set_aligned_vect): Likewise.
8745 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8746 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8747 (expand_vec_perm_even_odd_pack): Likewise.
8748 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8749 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8750 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8751 gen_const_vec_duplicate.
8752 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8753 * config/mips/mips.c (mips_gen_const_int_vector): Use
8754 gen_const_vec_duplicate.
8755 (mips_expand_vector_init): Use CONST0_RTX.
8756 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8757 (define_split): Use gen_const_vec_duplicate.
8758 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8759 (define_split): Use gen_const_vec_duplicate.
8760 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8761 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8762 * config/spu/spu.c (spu_const): Likewise.
8764 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8765 Alan Hayward <alan.hayward@arm.com>
8766 David Sherwood <david.sherwood@arm.com>
8768 * combine.c (can_change_dest_mode): Reject changes in
8769 REGMODE_NATURAL_SIZE.
8771 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8773 * configure.ac (--enable-libssp): New.
8774 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8775 trying to determine target-specific default. Adjust indentation.
8776 * configure: Regenerated.
8777 * doc/install.texi (Configuration): Expand --disable-libssp
8780 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8782 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8785 2017-10-31 Martin Jambor <mjambor@suse.cz>
8788 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8790 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8792 * auto-profile.c (autofdo_source_profile::read): Use
8793 UNKNOWN_LOCATION rather than 0.
8794 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8795 (warning_at): ...this overload.
8796 (warning_at_rich_loc_n): Rename to...
8797 (warning_n): ...this overload.
8798 (error_at_rich_loc): Rename to...
8799 (error_at): ...this overload.
8800 (pedwarn_at_rich_loc): Rename to...
8801 (pedwarn): ...this overload.
8802 (permerror_at_rich_loc): Rename to...
8803 (permerror): ...this overload.
8804 (inform_at_rich_loc): Rename to...
8805 (inform): ...this overload.
8806 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8807 (diagnostic_n_impl_richloc): Rename to...
8808 (diagnostic_n_impl): ...this rich_location *-based decl.
8809 (inform_at_rich_loc): Rename to...
8810 (inform): ...this, and add an assertion.
8811 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8812 (warning_at_rich_loc): Rename to...
8813 (warning_at): ...this, and add an assertion.
8814 (warning_at_rich_loc_n): Rename to...
8815 (warning_n): ...this, and add an assertion.
8816 (warning_n): Update location_t-based implementation for removal of
8817 location_t-based diagnostic_n_impl.
8818 (pedwarn_at_rich_loc): Rename to...
8819 (pedwarn): ...this, and add an assertion.
8820 (permerror_at_rich_loc): Rename to...
8821 (permerror): ...this, and add an assertion.
8822 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8823 (error_at_rich_loc): Rename to...
8824 (error_at): ...this, and add an assertion.
8825 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8826 (driver::do_spec_on_infiles): Likewise.
8827 * substring-locations.c (format_warning_va): Update for renaming
8828 of inform_at_rich_loc.
8830 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8832 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8833 _Float<N>X built-in functions so that the variant without the
8834 "__builtin_" prefix is only enabled for the GNU C and Objective C
8835 languages when they are in non-strict ANSI/ISO mode.
8836 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8837 * target.def (floatn_builtin_p): Add a target hook to control
8838 whether _Float<N> and _Float<N>X built-in functions without the
8839 "__builtin_" prefix are enabled, and return true for C and
8840 Objective C in the default hook. Include langhooks.h in
8842 * targhooks.h (default_floatn_builtin_p): Likewise.
8843 * targhooks.c (default_floatn_builtin_p): Likewise.
8844 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8845 floatn_builtin_p target hook.
8846 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8848 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8849 Eric Botcazou <ebotcazou@adacore.com>
8851 PR rtl-optimization/81803
8852 * lra-constraints.c (curr_insn_transform): Also reload the whole
8853 register for a strict subreg no wider than a word if this is for
8854 a WORD_REGISTER_OPERATIONS target.
8856 2017-10-31 Jason Merrill <jason@redhat.com>
8858 * gdbinit.in: Skip over inlines from timevar.h.
8860 2017-10-31 Martin Liska <mliska@suse.cz>
8862 * doc/gcov.texi: Document new option.
8863 * gcov.c (print_usage): Likewise print it.
8864 (process_args): Support the argument.
8865 (format_count): New function.
8866 (format_gcov): Use the function.
8868 2017-10-31 Martin Liska <mliska@suse.cz>
8870 * gcov.c (struct name_map): do not use typedef.
8871 Define operator== and operator<.
8872 (name_search): Remove.
8873 (name_sort): Remove.
8874 (main): Do not allocate names.
8875 (process_file): Add vertical space.
8876 (generate_results): Use std::find.
8877 (release_structures): Do not release memory.
8878 (find_source): Use std::find.
8880 2017-10-31 Martin Liska <mliska@suse.cz>
8882 * gcov.c (struct line_info): Remove it's typedef.
8883 (line_info::line_info): Add proper ctor.
8884 (line_info::has_block): Do not use a typedef.
8885 (struct source_info): Do not use typedef.
8886 (circuit): Likewise.
8887 (get_cycles_count): Likewise.
8888 (output_intermediate_file): Iterate via vector iterator.
8889 (add_line_counts): Use std::vector methods.
8890 (accumulate_line_counts): Likewise.
8891 (output_lines): Likewise.
8893 2017-10-31 Martin Liska <mliska@suse.cz>
8895 * gcov.c (struct source_info): Remove typedef.
8896 (source_info::source_info): Add proper ctor.
8897 (accumulate_line_counts): Use struct, not it's typedef.
8898 (output_gcov_file): Likewise.
8899 (output_lines): Likewise.
8900 (main): Do not allocate an array.
8901 (output_intermediate_file): Use size of vector container.
8902 (process_file): Resize the vector.
8903 (generate_results): Do not preallocate, use newly added vector
8905 (release_structures): Do not release sources.
8906 (find_source): Use vector methods.
8907 (add_line_counts): Do not use typedef.
8909 2017-10-31 Martin Liska <mliska@suse.cz>
8911 * doc/gcov.texi: Document that.
8912 * gcov.c (add_line_counts): Mark lines with a non-executed
8914 (output_line_beginning): Handle such lines.
8915 (output_lines): Pass new argument.
8916 (output_intermediate_file): Print it in intermediate format.
8918 2017-10-31 Martin Liska <mliska@suse.cz>
8920 * color-macros.h: New file.
8921 * diagnostic-color.c: Factor out color related to macros to
8923 * doc/gcov.texi: Document -k option.
8924 * gcov.c (INCLUDE_STRING): Include string.h.
8925 (print_usage): Add -k option.
8926 (process_args): Parse it.
8927 (pad_count_string): New function.
8928 (output_line_beginning): Likewise.
8929 (DEFAULT_LINE_START): New macro.
8930 (output_lines): Support color output.
8932 2017-10-31 Martin Liska <mliska@suse.cz>
8934 PR gcov-profile/82633
8935 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8936 their interaction with GCOV infrastructure.
8937 * configure.ac: Add -fkeep-{inline,static}-functions to
8939 * configure: Regenerate.
8941 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8944 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8946 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8949 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8950 into a register if it does not fit into an immediate offset field.
8952 2017-10-31 Olivier Hainque <hainque@adacore.com>
8954 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8956 2017-10-31 Julia Koval <julia.koval@intel.com>
8958 * config.gcc: Add gfniintrin.h.
8959 * config/i386/gfniintrin.h: New.
8960 * config/i386/i386-builtin-types.def
8961 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8962 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8963 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8964 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8965 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8966 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8967 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8968 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8969 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8970 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8971 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8972 * config/i386/immintrin.h: Include gfniintrin.h.
8973 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8975 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8977 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8979 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8981 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8982 (arm_ashldi3_1bit): Remove pattern.
8983 (ashrdi3): Remove shift by 1 expansion.
8984 (arm_ashrdi3_1bit): Remove pattern.
8985 (lshrdi3): Remove shift by 1 expansion.
8986 (arm_lshrdi3_1bit): Remove pattern.
8987 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8988 cost of ashldi3 by 1.
8989 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8990 (<shift>di3_neon): Likewise.
8992 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8994 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8995 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8996 and (*aarch64_simd_mov<VQ:mode>).
8997 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8998 pattern alternative.
8999 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9000 attributes to match pattern alternative.
9002 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9004 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9005 * config/rs6000/emmintrin.h: New file.
9006 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9008 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9010 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9011 * (movdi_vfp_cortexa8): Remove pattern.
9013 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9015 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9017 (Specific, alpha*-dec-osf5.1): Remove.
9018 (Specific, mips-sgi-irix5): Remove.
9019 (Specific, mips-sgi-irix6): Remove.
9021 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9024 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9025 arguments to clear_bit_region_be.
9027 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9029 * gimplify.c: Include memmodel.h.
9031 2017-10-30 Martin Jambor <mjambor@suse.cz>
9033 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9034 condition whether loop should be executed at all.
9036 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9038 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9039 gimple folding of vec_madd() intrinsics.
9040 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9041 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9042 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9044 2017-10-30 Richard Biener <rguenther@suse.de>
9046 PR tree-optimization/82762
9048 2017-10-23 Richard Biener <rguenther@suse.de>
9050 PR tree-optimization/82129
9052 2017-08-01 Richard Biener <rguenther@suse.de>
9054 PR tree-optimization/81181
9055 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9056 (compute_antic): ... end of iteration here.
9058 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9060 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9061 -std=iso9899:2017 and -std=gnu17.
9062 * doc/standards.texi (C Language): Document C17 support.
9063 * doc/cpp.texi (Overview): Mention -std=c17.
9064 (Standard Predefined Macros): Document C11 and C17 values of
9065 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9066 * doc/extend.texi (Inline): Do not list individual options for
9067 standards newer than C99.
9068 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9070 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9073 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9075 * asan.c (asan_finish_file): Align asan globals array by shadow
9078 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9081 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9082 (struct store_immediate_info): Add bitregion_start and bitregion_end
9084 (store_immediate_info::store_immediate_info): Add brs and bre
9085 arguments and initialize bitregion_{start,end} from those.
9086 (struct merged_store_group): Add bitregion_start, bitregion_end,
9087 align_base and mask fields. Drop unnecessary struct keyword from
9088 struct store_immediate_info. Add do_merge method.
9089 (clear_bit_region_be): Use memset instead of loop storing zeros.
9090 (merged_store_group::do_merge): New method.
9091 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9092 stores as long as the surrounding bitregions have no gaps.
9093 (merged_store_group::merge_overlapping): Use do_merge.
9094 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9095 is byte aligned, rather than requiring that start and width are
9096 byte aligned. Drop unnecessary struct keyword from
9097 struct store_immediate_info. Allocate and populate also mask array.
9098 Make start of the arrays relative to bitregion_start rather than
9099 start and size them according to bitregion_{end,start} difference.
9100 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9101 struct store_immediate_info.
9102 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9103 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9104 struct keyword from struct store_immediate_info.
9105 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9106 between stores as long as the surrounding bitregions have no gaps.
9108 (struct split_store): Add orig non-static data member.
9109 (split_store::split_store): Initialize orig to false.
9110 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9111 if there is exactly a single original stmt. Change stmts argument
9112 to pointer from reference, if NULL, don't push anything to it. Add
9113 first argument, use it to optimize skipping over orig stmts that
9114 are known to be before bitpos already. Simplify.
9115 (split_group): Return unsigned int count how many stores are or
9116 would be needed rather than a bool. Add allow_unaligned argument.
9117 Change split_stores argument from reference to pointer, if NULL,
9118 only do a dry run computing how many stores would be produced.
9119 Rewritten algorithm to use both alignment and misalign if
9120 !allow_unaligned and handle bitfield stores with gaps.
9121 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9122 from bitregion_start instead of start. Compute allow_unaligned
9123 here, if true, do 2 split_group dry runs to compute which one
9124 produces fewer stores and prefer aligned if equal. Punt if
9125 new count is bigger or equal than original before emitting any
9126 statements, rather than during that. Remove no longer needed
9127 new_ssa_names tracking. Replace num_stmts with
9128 split_stores.length (). Use 32-bit stack allocated entries
9129 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9130 if possible. Handle bitfields with gaps.
9131 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9132 Compute bitregion_{start,end} for the stores and construct
9133 store_immediate_info with that. Formatting fixes.
9135 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9138 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9139 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9141 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9143 * gimplify.c: Include tm_p.h.
9145 * common.opt (gcoff): Re-add as ignored option.
9146 (gcoff1, gcoff2, gcoff3): Likewise.
9148 * Makefile.in (OBJS): Delete sdbout.o.
9149 (GTFILES): Delete $(srcdir)/sdbout.c.
9150 * debug.h: Delete sdb_debug_hooks.
9151 * final.c: Delete sdbout.h include.
9152 (final_scan_insn): Delete SDB_DEBUG check.
9153 (rest_of_clean_state): Likewise.
9154 * output.h: Delete sdb_begin_function_line.
9157 * toplev.c: Delete sdbout.h include.
9158 (process_options): Delete SDB_DEBUG check.
9159 * tree-core.h (tree_type_common): Delete pointer field of
9161 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9162 TYPE_SYMTAB_POINTER.
9163 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9164 (TYPE_SYMTAB_IS_POINTER): Delete.
9165 (TYPE_SYMTAB_IS_DIE): Renumber.
9166 * xcoffout.c: Refer to former sdbout.c file.
9167 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9169 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9170 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9171 (Debugging Options): Delete -gcoff.
9172 (-gstabs): Delete SDB reference.
9174 (-gcoff@var{level}): Delete.
9175 * doc/passes.texi (Debugging information output): Delete SDB and
9176 sdbout.c references.
9177 * doc/tm.texi: Regenerate.
9178 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9179 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9181 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9182 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9183 (SDB_DEBUGGING_INFO): Delete.
9184 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9185 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9186 * target.def (output_source_filename): Delete COFF reference.
9188 * common.opt (gcoff): Delete.
9189 (gxcoff+): Update Negative chain.
9190 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9192 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9193 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9194 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9196 (expand_function_start): Change sdb reference to past tense.
9197 (expand_function_end): Change sdb reference to past tense.
9198 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9199 * opts.c (debug_type_names): Delete coff entry.
9200 (common_handle_option): Delete OPT_gcoff case.
9201 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9203 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9204 * config/cris/cris.h: Delete SDB reference in comment.
9205 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9206 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9207 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9208 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9210 (ix86_expand_prologue): Likewise.
9211 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9212 * config/ia64/ia64.h: Likewise.
9213 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9214 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9216 * config/mmix/mmix.h: Likewise.
9217 * config/nds32/nds32.c: Likewise.
9218 * config/stormy/storym16.h: Likewise.
9219 * config/visium/visium.h: Likewise.
9220 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9222 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9224 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9225 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9226 FRAME_POINTER_REGNUM point at high end of local var area.
9228 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9230 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9231 Move comment around. Do not reset best_edge for a copiable
9232 destination if the copy would cause a partition change.
9233 (better_edge_p): Remove redundant check.
9235 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9237 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9239 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9241 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9242 for math functions that have _Float<N> and _Float<N>X variants.
9243 (mathfn_built_in_2): Add support for math functions that have
9244 _Float<N> and _Float<N>X variants.
9245 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9246 (expand_builtin_mathfn_ternary): Add support for fma with
9247 _Float<N> and _Float<N>X variants.
9248 (expand_builtin): Likewise.
9249 (fold_builtin_3): Likewise.
9250 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9251 create math function _Float<N> and _Float<N>X variants as external
9253 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9254 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9255 the __builtin_ prefix and if not strict ansi, without the prefix.
9256 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9257 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9258 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9259 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9260 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9261 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9262 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9263 function signatures for fma _Float<N> and _Float<N>X variants.
9264 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9265 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9266 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9267 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9268 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9269 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9270 * gencfn-macros.c (print_case_cfn): Add support for math functions
9271 that have _Float<N> and _Float<N>X variants.
9272 (print_define_operator_list): Likewise.
9273 (fltfn_suffixes): Likewise.
9275 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9276 for math functions that have _Float<N> and _Float<N>X variants.
9277 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9278 and _Float<N>X variants.
9279 (COPYSIGN): Likewise.
9282 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9283 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9285 (integer_valued_read_call_p): Likewise.
9286 * fold-const-call.c (fold_const_call_ss): Likewise.
9287 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9288 _Float<N> and _Float<N>X variants.
9289 (fold_const_call_ssss): Add support for fma _Float<N> and
9290 _Float<N>X variants.
9291 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9292 support for copysign and fma _Float<N> and _Float<N>X variants.
9293 (backprop::process_builtin_call_use): Likewise.
9294 * tree-call-cdce.c (can_test_argument_range); Add support for
9295 sqrt _Float<N> and _Float<N>X variants.
9296 (edom_only_function): Likewise.
9297 (get_no_error_domain): Likewise.
9298 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9299 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9300 copysign _Float<N> and _Float<N>X variants.
9301 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9302 handled by machine independent code.
9303 (FMAF128): Likewise.
9304 * doc/cpp.texi (Common Predefined Macros): Document defining
9305 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9306 fma _Float<N> and _Float<N>X variants.
9308 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9311 * config/i386/i386-modes.def (CCFPU): Remove definition.
9312 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9313 (ix86_cc_modes_compatible): Ditto.
9314 (ix86_expand_carry_flag_compare): Ditto.
9315 (ix86_expand_int_movcc): Ditto.
9316 (ix86_expand_int_addcc): Ditto.
9317 (ix86_reverse_condition): Ditto.
9318 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9319 Return true/false for unordered/ordered fp comparisons.
9320 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9321 (ix86_prepare_fp_compare_args): Update for rename.
9322 (ix86_expand_fp_compare): Update for rename. Generate unordered
9323 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9324 (ix86_expand_sse_compare_and_jump): Ditto.
9325 * config/i386/predicates.md (fcmov_comparison_operator):
9326 Remove CCFPU mode handling.
9327 (ix86_comparison_operator): Ditto.
9328 (ix86_carry_flag_operator): Ditto.
9329 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9330 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9331 (*cmpu<mode>_cc_i387): Ditto.
9332 (FPCMP): Remove mode iterator.
9333 (unord): Remove mode attribute.
9334 (unord_subst): New define_subst transformation
9335 (unord): New define_subst attribute.
9337 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9338 (*cmpi<unord>xf_i387): Ditto.
9339 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9340 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9341 using unord_subst transformation.
9342 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9343 (round_saeonly): Also handle CCFP mode.
9344 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9345 Remove UNSPEC_SAHF unspec handling.
9347 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9349 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9351 2017-10-27 Jeff Law <law@redhat.com>
9353 * gimple-ssa-sprintf.c: Include domwalk.h.
9354 (class sprintf_dom_walker): New class, derived from dom_walker.
9355 (sprintf_dom_walker::before_dom_children): New function.
9356 (struct call_info): Moved into sprintf_dom_walker class
9357 (compute_formath_length, handle_gimple_call): Likewise.
9358 (sprintf_length::execute): Call the dominator walker rather
9359 than walking the statements.
9361 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9362 gimple statement locations.
9363 (check_array_bounds): Corresponding changes. Get the statement's
9364 location directly from wi->stmt.
9366 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9369 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9371 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9373 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9374 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9376 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9379 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9380 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9381 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9382 maybe_get_pool_constant.
9383 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9386 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9388 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9389 2.26 caveat. Update gas and gld versions.
9390 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9393 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9395 * cgraph.h (set_malloc_flag): Declare.
9396 * cgraph.c (set_malloc_flag_1): New function.
9397 (set_malloc_flag): Likewise.
9398 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9399 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9401 (read_ipa_call_summary): Add support for reading is_return_callee.
9402 (write_ipa_call_summary): Stream is_return_callee.
9403 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9404 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9405 ipa-prop.h, ipa-fnsummary.h.
9406 (pure_const_names): Change to static.
9407 (malloc_state_e): Define.
9408 (malloc_state_names): Define.
9409 (funct_state_d): Add field malloc_state.
9410 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9411 (check_retval_uses): New function.
9412 (malloc_candidate_p): Likewise.
9413 (analyze_function): Add support for malloc attribute.
9414 (pure_const_write_summary): Stream malloc_state.
9415 (pure_const_read_summary): Add support for reading malloc_state.
9416 (dump_malloc_lattice): New function.
9417 (propagate_malloc): New function.
9418 (warn_function_malloc): New function.
9419 (ipa_pure_const::execute): Call propagate_malloc and
9420 ipa_free_fn_summary.
9421 (pass_local_pure_const::execute): Add support for malloc attribute.
9422 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9423 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9425 2017-10-27 Martin Liska <mliska@suse.cz>
9427 PR gcov-profile/82457
9428 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9429 for fork-like functions to be properly instrumented.
9431 2017-10-27 Richard Biener <rguenther@suse.de>
9434 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9435 info when we redirected EH.
9437 2017-10-26 Michael Collison <michael.collison@arm.com>
9439 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9441 (<optab>_trunchf<GPI:mode>2: New pattern.
9442 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9443 * config/aarch64/iterators.md (wv): New mode attribute.
9444 (vf, VF): New mode attributes.
9445 (vgp, VGP): New mode attributes.
9446 (s): Update attribute with SImode and DImode prefixes.
9448 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9450 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9451 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9452 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9453 (nios2_option_overide): Initialize it. Don't allow R0-relative
9454 addressing with PIC.
9455 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9456 (nios2_symbolic_constant_p): Likewise.
9457 (nios2_legitimate_address_p): Likewise.
9458 (nios2_r0rel_section_name_p): New.
9459 (nios2_symbol_ref_in_r0rel_data_p): New.
9460 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9461 (r0rel_constant_p): New.
9462 (nios2_print_operand_address): Handle r0rel_constant_p.
9463 (nios2_cdx_narrow_form_p): Likewise.
9464 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9465 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9466 (Nios II Options): Document -mr0rel-sec.
9468 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9470 * config/nios2/nios2.c: Include xregex.h.
9471 (nios2_gprel_sec_regex): New.
9472 (nios2_option_overide): Initialize it. Don't allow GP-relative
9473 addressing with PIC.
9474 (nios2_small_section_name_p): Check for regex match.
9475 * config/nios2/nios2.opt (mgprel-sec=): New option.
9476 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9477 (Nios II Options): Document -mgprel-sec.
9479 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9481 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9483 2017-10-26 Tom de Vries <tom@codesourcery.com>
9485 PR tree-optimization/82707
9486 * gimple.c (gimple_copy): Fix unsharing of
9487 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9489 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9491 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9492 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9493 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9494 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9495 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9496 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9497 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9498 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9499 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9500 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9501 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9502 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9503 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9504 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9505 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9506 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9507 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9509 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9511 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9513 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9514 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9515 warning to rs6000.c. Remove the Undocumented flag, since it has
9517 (-mabi=ibmlongdouble): Likewise.
9518 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9519 already set, set the default format for long double.
9520 (rs6000_debug_reg_global): Print whether long double is IBM or
9522 (rs6000_option_override_internal): Rework setting long double
9523 format. Only warn if the user is changing the long double default
9524 and they did not use -Wno-psabi.
9525 * doc/invoke.texi (PowerPC options): Update the documentation for
9526 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9528 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9529 Alan Hayward <alan.hayward@arm.com>
9530 David Sherwood <david.sherwood@arm.com>
9532 * rtl.h (wider_subreg_mode): New function.
9533 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9534 rather than an unsigned int *.
9535 * ira-color.c (regno_max_ref_width): Replace with...
9536 (regno_max_ref_mode): ...this new variable.
9537 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9538 Use wider_subreg_mode.
9539 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9540 rather than an unsigned int *.
9541 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9542 (process_alt_operands): Likewise.
9543 (invariant_p): Likewise.
9544 * lra-spills.c (assign_mem_slot): Likewise.
9545 (add_pseudo_to_slot): Likewise.
9546 * lra.c (collect_non_operand_hard_regs): Likewise.
9547 (add_regs_to_insn_regno_info): Likewise.
9548 * reload1.c (regno_max_ref_width): Replace with...
9549 (regno_max_ref_mode): ...this new variable.
9550 (reload): Update accordingly. Update call to
9551 ira_sort_regnos_for_alter_reg.
9552 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9553 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9554 (scan_paradoxical_subregs): Likewise.
9556 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9558 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9559 (aarch64_frame): Add emit_frame_chain boolean.
9560 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9561 Move eh_return case to aarch64_layout_frame.
9562 (aarch64_layout_frame): Initialize emit_frame_chain.
9563 (aarch64_expand_prologue): Use emit_frame_chain.
9565 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9567 * config/aarch64/aarch64.c (aarch64_layout_frame):
9568 Ensure LR is always stored at the bottom of the callee-saves.
9569 Remove rarely used frame layout which saves callee-saves at top of
9570 frame, so the store of LR can be used as a valid probe in all cases.
9572 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9574 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9575 Improve unaligned TImode/TFmode base/offset split.
9577 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9578 Alan Hayward <alan.hayward@arm.com>
9579 David Sherwood <david.sherwood@arm.com>
9581 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9582 * combine.c (find_single_use_1): Likewise.
9583 (expand_field_assignment): Likewise.
9584 (move_deaths): Likewise.
9585 * lra-constraints.c (simplify_operand_subreg): Likewise.
9586 (curr_insn_transform): Likewise.
9587 * lra.c (collect_non_operand_hard_regs): Likewise.
9588 (add_regs_to_insn_regno_info): Likewise.
9589 * rtlanal.c (reg_referenced_p): Likewise.
9590 (covers_regno_no_parallel_p): Likewise.
9592 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9594 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9595 Don't print any bits outside the precision of the value.
9596 * wide-int.cc (test_printing): Add some new tests.
9598 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9600 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9601 supports -xbrace_comment option.
9602 * configure: Regenerate.
9603 * config.in: Regenerate.
9604 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9605 (ASM_CPU_SPEC): Use it.
9607 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9609 * target.def (static_rtx_alignment): New hook.
9610 * targhooks.h (default_static_rtx_alignment): Declare.
9611 * targhooks.c (default_static_rtx_alignment): New function.
9612 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9613 * doc/tm.texi: Regenerate.
9614 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9615 instead of targetm.constant_alignment. Remove call to
9617 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9618 (cris_preferred_mininum_alignment): New function, split out from...
9619 (cris_constant_alignment): ...here.
9620 (cris_static_rtx_alignment): New function.
9621 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9623 (ix86_constant_alignment): ...here.
9624 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9625 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9626 (mmix_static_rtx_alignment): New function.
9627 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9628 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9630 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9633 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9634 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9636 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9638 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9639 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9640 mark as a sign-extending load.
9641 (local_pic_load_u): Define.
9643 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9646 * fold-const.c (operand_equal_for_comparison_p): Also return true
9647 if ARG0 is a simple variant of ARG1 with narrower precision.
9648 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9650 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9652 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9654 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9655 scatter_static, scatter_per_elt.
9656 * x86-tune-costs.h: Add new cost entries.
9658 2017-10-25 Richard Biener <rguenther@suse.de>
9660 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9661 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9662 class pass_fre): Move to ...
9663 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9664 class pass_fre): ... here and adjust for statistics.
9666 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9669 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9670 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9671 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9672 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9675 2017-10-25 Richard Biener <rguenther@suse.de>
9677 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9678 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9679 eliminate_push_avail, eliminate_insert): Move inside...
9680 (class eliminate_dom_walker): ... this class in preparation
9682 (fini_eliminate): Remove by merging with ...
9683 (eliminate): ... this function. Adjust for class changes.
9684 (pass_pre::execute): Remove fini_eliminate call.
9685 (pass_fre::execute): Likewise.
9687 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9690 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9691 (VPERMI2, VPERMI2I): New mode iterators.
9692 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9693 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9695 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9696 mode iterator. Remove 3 old define_insn patterns.
9697 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9698 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9699 VPERMI2 mode iterator, remove the other two expanders.
9700 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9701 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9702 instructions, remove the other two patterns.
9703 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9704 mode iterator, remove the other two patterns.
9705 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9706 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9707 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9708 and adjust argument order accordingly.
9709 (ix86_expand_vec_perm): Adjust caller.
9710 (expand_vec_perm_1): Likewise.
9711 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9712 (expand_vec_perm_vpermt2_vpshub2): ... this.
9713 (ix86_expand_vec_perm_const_1): Adjust caller.
9714 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9717 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9718 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9719 (vec_shl_<mode>): Remove unused expander.
9720 (avx512bw_<shift_insn><mode>3): New define_insn.
9721 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9722 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9724 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9727 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9730 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9732 PR rtl-optimization/82396
9733 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9734 (autopref_multipass_init): Simplify initialization.
9735 (autopref_rank_data): Simplify sort order.
9736 * gcc/sched-int.h (autopref_multipass_data_): Remove
9737 multi_mem_insn_p, min_offset and max_offset.
9739 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9742 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9743 Check special value of flag_omit_frame_pointer.
9744 (aarch64_can_eliminate): Likewise.
9745 (aarch64_override_options_after_change_1): Simplify handling of
9746 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9748 2017-10-24 Richard Biener <rguenther@suse.de>
9750 PR tree-optimization/82697
9751 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9752 zero for conditional load and unconditional store.
9754 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9756 * doc/install.texi: Document bootstrap-cet.
9758 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9761 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9762 ENDBR instruction at function entrance if function is only
9765 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9768 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9769 patterns to better describe from which operation the CF is computed.
9770 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9771 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9772 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9773 is 0, use _0 suffixed expanders instead of emitting a comparison
9776 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9778 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9779 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9781 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9784 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9785 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9786 * loop-iv.c (iv_get_reaching_def): Likewise.
9787 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9788 variable is promoted and the partition contains undefined values.
9790 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9792 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9794 (nios2_address_cost): Define.
9795 (nios2_legitimize_address): Recognize (exp + constant) directly.
9796 (TARGET_ADDRESS_COST): Define.
9798 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9800 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9801 (nios2_symbolic_memory_operand_p): Declare.
9802 (nios2_split_large_constant): Declare.
9803 (nios2_split_symbolic_memory_operand): Declare.
9804 * config/nios2/nios2.c: Adjust includes.
9805 (nios2_symbolic_constant_allowed): New.
9806 (nios2_symbolic_constant_p): New.
9807 (nios2_plus_symbolic_constant_p): New.
9808 (nios2_valid_addr_expr_p): Recognize addresses involving
9810 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9811 (nios2_symbolic_memory_operand_p): New.
9812 (nios2_large_constant_p): New.
9813 (nios2_split_large_constant): New.
9814 (nios2_split_plus_large_constant): New.
9815 (nios2_split_symbolic_memory_operand): New.
9816 (nios2_legitimize_address): Code refactoring. Handle addresses
9817 involving symbolic constants.
9818 (nios2_emit_move_sequence): Likewise.
9819 (nios2_print_operand): Improve error output.
9820 (nios2_print_operand_address): Handle LO_SUM.
9821 (nios2_cdx_narrow_form_p): Likewise.
9822 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9823 operands involving symbolic constants.
9824 (movhi_internal, movsi_internal): Likewise.
9825 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9826 (extendhisi2, extendqi<mode>2): Likewise.
9828 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9830 * tree-pass.h (PROP_rtl_split_insns): Define.
9831 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9833 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9835 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9837 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9840 * target.def (const_not_ok_for_debug_p): Default to
9841 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9842 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9843 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9844 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9845 which targetm.const_not_ok_for_debug_p returned true.
9846 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9848 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9850 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9851 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9852 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9854 (ix86_const_not_ok_for_debug_p): New function.
9855 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9856 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9858 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9861 * system.h: Conditionally include "unique-ptr.h" if
9862 INCLUDE_UNIQUE_PTR is defined.
9863 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9864 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9866 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9868 * config/rl78/rl78.md: New define_expand "subdi3".
9870 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9873 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9874 DF_REF_INSN if DF_REF_INSN_INFO is false.
9876 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9878 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9879 xmm_move instead of sse_move.
9880 (sse_store_index): New function.
9881 (ix86_register_move_cost): Be more sensible about mismatch stall;
9882 model AVX moves correctly; make difference between sse->integer and
9884 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9885 moves; make difference between SSE and AVX.
9886 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9887 and zmm_move. Increase size of sse load and store tables;
9888 add unaligned load and store tables; add ssemmx_to_integer.
9889 * x86-tune-costs.h: Update all entries according to real
9890 move latencies from Agner Fog's manual and chip documentation.
9892 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9895 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9896 * config/i386/constraints.md (Wf): New constraint.
9897 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9898 (cmp<dwi>_doubleword): Removed.
9899 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9900 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9901 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9902 expand with cmp<dwi>_doubleword. For LTU and GEU use
9903 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9905 * common.opt (gcolumn-info): Enable by default.
9906 * doc/invoke.texi (gcolumn-info): Document new default.
9908 2017-10-23 Richard Biener <rguenther@suse.de>
9910 PR tree-optimization/82672
9911 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9912 Fold the stmt if we propagated into it.
9914 2017-10-23 Richard Biener <rguenther@suse.de>
9916 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9917 (bitmap_remove_expr_from_set): ... this. All callers call this
9918 for non-constant values.
9919 (bitmap_set_subtract): Rename to...
9920 (bitmap_set_subtract_expressions): ... this. Adjust and
9922 (bitmap_set_contains_value): Remove superfluous check.
9923 (bitmap_set_replace_value): Inline into single caller ...
9924 (bitmap_value_replace_in_set): ... here and simplify.
9925 (dependent_clean): Merge into ...
9926 (clean): ... this using an overload. Adjust.
9927 (prune_clobbered_mems): Adjust.
9928 (compute_antic_aux): Likewise.
9929 (compute_partial_antic_aux): Likewise.
9931 2017-10-23 Richard Biener <rguenther@suse.de>
9933 PR tree-optimization/82129
9935 2017-08-01 Richard Biener <rguenther@suse.de>
9937 PR tree-optimization/81181
9938 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9939 (compute_antic): ... end of iteration here.
9941 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9943 * target.def (starting_frame_offset): New hook.
9944 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9945 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9946 * doc/tm.texi.in: Regenerate.
9947 * hooks.h (hook_hwi_void_0): Declare.
9948 * hooks.c (hook_hwi_void_0): New function.
9949 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9950 STARTING_FRAME_OFFSET.
9951 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9952 * reload1.c (reload): Likewise.
9953 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9954 instead of STARTING_FRAME_OFFSET.
9955 * function.c (try_fit_stack_local): Likewise.
9956 (assign_stack_local_1): Likewise
9957 (instantiate_virtual_regs): Likewise.
9958 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9959 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9960 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9961 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9962 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9963 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9964 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9965 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9966 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9967 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9968 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9969 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9970 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9971 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9972 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9973 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9974 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9975 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9976 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9977 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9978 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9979 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9980 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9981 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9982 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9983 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9984 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9985 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9986 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9987 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9988 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9989 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9990 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9991 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9992 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9993 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9994 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9995 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9996 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9997 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9998 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9999 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10000 return a HOST_WIDE_INT.
10001 (avr_builtin_setjmp_frame_value): Use it instead of
10002 STARTING_FRAME_OFFSET.
10003 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10004 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10005 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10007 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10008 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10009 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10010 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10011 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10012 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10013 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10014 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10015 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10016 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10017 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10018 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10020 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10021 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10022 * config/mips/mips.c (mips_compute_frame_info): Refer to
10023 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10024 (mips_starting_frame_offset): New function.
10025 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10026 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10027 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10028 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10029 and return a HOST_WIDE_INT.
10030 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10031 (mmix_initial_elimination_offset): Refer to
10032 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10033 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10034 * config/pa/pa.c (pa_starting_frame_offset): New function.
10035 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10036 (pa_expand_prologue): Likewise.
10037 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10038 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10039 !FRAME_GROWS_DOWNWARD handling to...
10040 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10041 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10042 !FRAME_GROWS_DOWNWARD handling to...
10043 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10044 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10045 !FRAME_GROWS_DOWNWARD handling to...
10046 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10047 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10049 (rs6000_starting_frame_offset): New function.
10050 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10051 !FRAME_GROWS_DOWNWARD handling to...
10052 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10053 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10054 !FRAME_GROWS_DOWNWARD handling to...
10055 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10056 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10057 !FRAME_GROWS_DOWNWARD handling to...
10058 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10059 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10060 (rs6000_starting_frame_offset): New function.
10061 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10062 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10063 * config/vax/vax.c (vax_starting_frame_offset): New function.
10064 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10065 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10066 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10067 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10068 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10069 * system.h (STARTING_FRAME_OFFSET): Poison.
10071 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10073 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10074 SCALAR_TYPE_MODE instead of TYPE_MODE.
10076 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10077 Alan Hayward <alan.hayward@arm.com>
10078 David Sherwood <david.sherwood@arm.com>
10080 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10082 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10083 Alan Hayward <alan.hayward@arm.com>
10084 David Sherwood <david.sherwood@arm.com>
10086 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10088 2017-10-23 Richard Biener <rguenther@suse.de>
10090 PR tree-optimization/82129
10091 * tree-ssa-pre.c (bitmap_set_and): Remove.
10092 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10093 canonicalizing expressions in the set to those with lowest
10094 ID rather than taking that from the first edge.
10096 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10098 * combine.c (rtx_equal_for_field_assignment_p): Use
10099 byte_lowpart_offset.
10101 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10102 Alan Hayward <alan.hayward@arm.com>
10103 David Sherwood <david.sherwood@arm.com>
10105 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10106 to a SUBREG_PROMOTED_VAR.
10108 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10109 Alan Hayward <alan.hayward@arm.com>
10110 David Sherwood <david.sherwood@arm.com>
10112 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10113 (expand_debug_source_expr): Likewise.
10114 * combine.c (combine_simplify_rtx): Likewise.
10115 * cse.c (fold_rtx): Likewise.
10116 * optabs.c (expand_float): Likewise.
10117 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10118 (simplify_binary_operation_1): Likewise.
10120 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10121 Alan Hayward <alan.hayward@arm.com>
10122 David Sherwood <david.sherwood@arm.com>
10124 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10125 (record_promoted_value): Likewise.
10126 * expr.c (expand_expr_real_2): Likewise.
10127 * ree.c (update_reg_equal_equiv_notes): Likewise.
10128 (combine_set_extension): Likewise.
10129 * rtlanal.c (low_bitmask_len): Likewise.
10130 * simplify-rtx.c (neg_const_int): Likewise.
10131 (simplify_binary_operation_1): Likewise.
10133 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10134 Alan Hayward <alan.hayward@arm.com>
10135 David Sherwood <david.sherwood@arm.com>
10137 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10138 * regcprop.c (maybe_mode_change): Likewise.
10139 * reload1.c (alter_reg): Likewise.
10141 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10143 * inchash.h (inchash::hash::add_wide_int): New function.
10144 * lto-streamer-out.c (hash_tree): Use it.
10146 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10148 * inchash.h (inchash::hash::add_wide_int): Rename to...
10149 (inchash::hash::add_hwi): ...this.
10150 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10151 (polymorphic_call_target_hasher::hash): Likewise.
10152 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10153 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10154 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10155 * lto-streamer-out.c (hash_tree): Likewise.
10156 * optc-save-gen.awk: Likewise.
10157 * tree.c (add_expr): Likewise.
10159 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10162 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10163 for ordered inequality comparisons even with TARGET_IEEE_FP.
10165 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10168 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10169 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10170 Expand with cmp<dwi>_doubleword.
10172 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10174 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10175 List CET intrinsics.
10176 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10177 specific to -fcf-protection option.
10179 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10181 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10182 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10183 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10184 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10185 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10186 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10187 (extra_objs): Add cet.o for Linux/x86 targets.
10188 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10189 * config/i386/cet.c: New file.
10190 * config/i386/cetintrin.h: Likewise.
10191 * config/i386/t-cet: Likewise.
10192 * config/i386/cpuid.h (bit_SHSTK): New.
10193 (bit_IBT): Likewise.
10194 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10195 pass IBT and SHSTK bits.
10196 * config/i386/i386-builtin-types.def
10197 (VOID_FTYPE_UNSIGNED_PVOID): New.
10198 (VOID_FTYPE_UINT64_PVOID): Likewise.
10199 * config/i386/i386-builtin.def: Add CET intrinsics.
10200 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10201 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10202 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10203 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10205 * config/i386/i386.c (rest_of_insert_endbranch): New.
10206 (pass_data_insert_endbranch): Likewise.
10207 (pass_insert_endbranch): Likewise.
10208 (make_pass_insert_endbranch): Likewise.
10209 (ix86_notrack_prefixed_insn_p): Likewise.
10210 (ix86_target_string): Add -mibt, -mshstk flags.
10211 (ix86_option_override_internal): Add flag_cf_protection
10213 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10214 (ix86_print_operand): Add 'notrack' prefix output.
10215 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10216 (ix86_expand_builtin): Expand CET intrinsics.
10217 (x86_output_mi_thunk): Add 'endbranch' instruction.
10218 * config/i386/i386.h (TARGET_IBT): New.
10219 (TARGET_IBT_P): Likewise.
10220 (TARGET_SHSTK): Likewise.
10221 (TARGET_SHSTK_P): Likewise.
10222 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10223 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10224 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10225 (builtin_setjmp_setup): New pattern.
10226 (builtin_longjmp): Likewise.
10227 (rdssp<mode>): Likewise.
10228 (incssp<mode>): Likewise.
10229 (saveprevssp): Likewise.
10230 (rstorssp): Likewise.
10231 (wrss<mode>): Likewise.
10232 (wruss<mode>): Likewise.
10233 (setssbsy): Likewise.
10234 (clrssbsy): Likewise.
10235 (nop_endbr): Likewise.
10236 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10238 * config/i386/immintrin.h: Include <cetintrin.h>.
10239 * config/i386/linux-common.h
10240 (file_end_indicate_exec_stack_and_cet): New prototype.
10241 (TARGET_ASM_FILE_END): New.
10243 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10245 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10246 latencies instead of having separate table; make difference between
10247 integer and float costs.
10248 * i386.h (processor_costs): Remove scalar_stmt_cost,
10249 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10250 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10252 * x86-tune-costs.h: Remove entries which has been removed in
10253 procesor_costs from all tables; make cond_taken_branch_cost
10254 and cond_not_taken_branch_cost COST_N_INSNS based.
10256 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10258 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10260 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10262 * config/i386/i386.md (isa): Remove fma_avx512f.
10263 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10264 <avx512>_fmadd_<mode>_mask3<round_name>,
10265 <avx512>_fmsub_<mode>_mask<round_name>,
10266 <avx512>_fmsub_<mode>_mask3<round_name>,
10267 <avx512>_fnmadd_<mode>_mask<round_name>,
10268 <avx512>_fnmadd_<mode>_mask3<round_name>,
10269 <avx512>_fnmsub_<mode>_mask<round_name>,
10270 <avx512>_fnmsub_<mode>_mask3<round_name>,
10271 <avx512>_fmaddsub_<mode>_mask<round_name>,
10272 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10273 <avx512>_fmsubadd_<mode>_mask<round_name>,
10274 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10275 (*vec_widen_umult_even_v16si<mask_name>,
10276 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10277 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10279 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10281 * extend.texi: Add 'nocf_check' documentation.
10282 * gimple.texi: Add second parameter to
10283 gimple_build_call_from_tree.
10284 * invoke.texi: Add -fcf-protection documentation.
10285 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10287 2017-10-20 Richard Biener <rguenther@suse.de>
10289 PR tree-optimization/82473
10290 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10291 the largest input type.
10293 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10295 * c-attribs.c (handle_nocf_check_attribute): New function.
10296 (c_common_attribute_table): Add 'nocf_check' handling.
10297 * gimple-parser.c: Add second argument NULL to
10298 gimple_build_call_from_tree.
10299 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10300 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10302 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10303 * common.opt: Add fcf-protection flag.
10304 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10305 * flag-types.h: Add enum cf_protection_level.
10306 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10307 Add 'nocf_check' attribute propagation to gimple call.
10308 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10309 (gimple_build_call_from_tree): Update prototype.
10310 (gimple_call_nocf_check_p): New function.
10311 (gimple_call_set_nocf_check): Likewise.
10312 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10313 * ipa-icf.c: Add nocf_check attribute in statement hash.
10314 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10315 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10316 * toplev.c (process_options): Add flag_cf_protection handling.
10318 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10320 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10322 2017-10-20 Richard Biener <rguenther@suse.de>
10324 PR tree-optimization/82603
10325 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10326 remove false predicated stores.
10328 2017-10-20 Richard Biener <rguenther@suse.de>
10330 * graphite-isl-ast-to-gimple.c
10331 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10332 Remove return value and simplify, dump copied stmt after lhs
10334 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10335 Reduce dump verbosity.
10336 (gsi_insert_earliest): Likewise.
10337 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10338 * graphite.c (print_global_statistics): Adjust dumping.
10339 (print_graphite_scop_statistics): Likewise.
10340 (print_graphite_statistics): Do not dump loops here.
10341 (graphite_transform_loops): But here.
10343 2017-10-20 Nicolas Roche <roche@adacore.com>
10345 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10346 * configure: Regenerate.
10348 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10351 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10352 functions when optimizing replace GIMPLE_RETURN stmts with
10353 calls to __builtin_unreachable ().
10356 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10357 for -fsanitize=thread link of executables.
10358 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10359 link of executables.
10362 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10363 New mode iterators.
10364 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10365 define_insns for logical vector shifts to use VI248_AVX512BW
10366 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10367 condition, useless isa and prefix attributes. Change the first
10368 2 of these define_insns to ...
10369 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10370 define_insn for avx512vl.
10371 (<shift_insn><mode>3): ... and this, new define_insn without
10372 masking for non-avx512vl.
10375 * config/i386/sse.md (*andnot<mode>3,
10376 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10377 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10378 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10379 not applied use empty suffix even for TARGET_AVX512VL.
10380 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10381 is applied, supply evex,evex or evex,evex,evex instead of just
10384 2017-10-20 Julia Koval <julia.koval@intel.com>
10386 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10387 (OPTION_MASK_ISA_GFNI_UNSET): New.
10388 (ix86_handle_option): Handle OPT_mgfni.
10389 * config/i386/cpuid.h (bit_GFNI): New.
10390 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10391 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10392 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10393 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10394 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10395 * config/i386/i386.opt: Add mgfni.
10397 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10399 * config/msp430/msp430.c (msp430_option_override): Disable
10400 -fdelete-null-pointer-checks.
10401 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10403 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10405 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10406 of x87 and SSE instructions.
10408 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10410 * asan.c (create_cond_insert_point): Do not update edge count.
10411 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10412 (afdo_propagate_circuit): Likewise.
10413 (afdo_calculate_branch_prob): Likewise.
10414 (afdo_annotate_cfg): Likewise.
10415 * basic-block.h (struct edge_def): Remove count.
10416 (edge_def::count): New accessor.
10417 * bb-reorder.c (rotate_loop): Update.
10418 (find_traces_1_round): Update.
10419 (connect_traces): Update.
10420 (sanitize_hot_paths): Update.
10421 * cfg.c (unchecked_make_edge): Update.
10422 (make_single_succ_edge): Update.
10423 (check_bb_profile): Update.
10424 (dump_edge_info): Update.
10425 (update_bb_profile_for_threading): Update.
10426 (scale_bbs_frequencies_int): Update.
10427 (scale_bbs_frequencies_gcov_type): Update.
10428 (scale_bbs_frequencies_profile_count): Update.
10429 (scale_bbs_frequencies): Update.
10430 * cfganal.c (connect_infinite_loops_to_exit): Update.
10431 * cfgbuild.c (compute_outgoing_frequencies): Update.
10432 (find_many_sub_basic_blocks): Update.
10433 * cfgcleanup.c (try_forward_edges): Update.
10434 (try_crossjump_to_edge): Update
10435 * cfgexpand.c (expand_gimple_cond): Update
10436 (expand_gimple_tailcall): Update
10437 (construct_exit_block): Update
10438 * cfghooks.c (verify_flow_info): Update
10439 (redirect_edge_succ_nodup): Update
10440 (split_edge): Update
10441 (make_forwarder_block): Update
10442 (duplicate_block): Update
10443 (account_profile_record): Update
10444 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10445 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10446 * cfgloopmanip.c (scale_loop_profile): Update.
10448 (lv_adjust_loop_entry_edge): Update.
10449 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10450 (force_nonfallthru_and_redirect): Update.
10451 (purge_dead_edges): Update.
10452 (rtl_flow_call_edges_add): Update.
10453 * cgraphunit.c (init_lowered_empty_function): Update.
10454 (cgraph_node::expand_thunk): Update.
10455 * gimple-pretty-print.c (dump_probability): Update.
10456 (dump_edge_probability): Update.
10457 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10458 * haifa-sched.c (sched_create_recovery_edges): Update.
10459 * hsa-gen.c (convert_switch_statements): Update.
10460 * ifcvt.c (dead_or_predicable): Update.
10461 * ipa-inline-transform.c (inline_transform): Update.
10462 * ipa-split.c (split_function): Update.
10463 * ipa-utils.c (ipa_merge_profiles): Update.
10464 * loop-doloop.c (add_test): Update.
10465 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10466 * lto-streamer-in.c (input_cfg): Update.
10467 (input_function): Update.
10468 * lto-streamer-out.c (output_cfg): Update.
10469 * modulo-sched.c (sms_schedule): Update.
10470 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10471 * predict.c (maybe_hot_edge_p): Update.
10472 (unlikely_executed_edge_p): Update.
10473 (probably_never_executed_edge_p): Update.
10474 (dump_prediction): Update.
10475 (drop_profile): Update.
10476 (propagate_unlikely_bbs_forward): Update.
10477 (determine_unlikely_bbs): Update.
10478 (force_edge_cold): Update.
10479 * profile.c (compute_branch_probabilities): Update.
10480 * reg-stack.c (better_edge): Update.
10481 * shrink-wrap.c (handle_simple_exit): Update.
10482 * tracer.c (better_p): Update.
10483 * trans-mem.c (expand_transaction): Update.
10484 (split_bb_make_tm_edge): Update.
10485 * tree-call-cdce.c: Update.
10486 * tree-cfg.c (gimple_find_sub_bbs): Update.
10487 (gimple_split_edge): Update.
10488 (gimple_duplicate_sese_region): Update.
10489 (gimple_duplicate_sese_tail): Update.
10490 (gimple_flow_call_edges_add): Update.
10491 (insert_cond_bb): Update.
10492 (execute_fixup_cfg): Update.
10493 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10494 * tree-complex.c (expand_complex_div_wide): Update.
10495 * tree-eh.c (lower_resx): Update.
10496 (unsplit_eh): Update.
10497 (cleanup_empty_eh_move_lp): Update.
10498 * tree-inline.c (copy_edges_for_bb): Update.
10499 (freqs_to_counts): Update.
10500 (copy_cfg_body): Update.
10501 * tree-ssa-dce.c (remove_dead_stmt): Update.
10502 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10503 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10504 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10505 (unloop_loops): Update.
10506 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10507 * tree-ssa-loop-split.c (connect_loops): Update.
10508 (split_loop): Update.
10509 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10510 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10511 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10512 * tree-ssa-reassoc.c (branch_fixup): Update.
10513 * tree-ssa-tail-merge.c (replace_block_by): Update.
10514 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10515 (compute_path_counts): Update.
10516 (update_profile): Update.
10517 (recompute_probabilities): Update.
10518 (update_joiner_offpath_counts): Update.
10519 (estimated_freqs_path): Update.
10520 (freqs_to_counts_path): Update.
10521 (clear_counts_path): Update.
10522 (ssa_fix_duplicate_block_edges): Update.
10523 (duplicate_thread_path): Update.
10524 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10525 (case_bit_test_cmp): Update.
10526 (collect_switch_conv_info): Update.
10527 (gen_inbound_check): Update.
10528 (do_jump_if_equal): Update.
10529 (emit_cmp_and_jump_insns): Update.
10530 * tree-tailcall.c (decrease_profile): Update.
10531 (eliminate_tail_call): Update.
10532 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10533 (vect_do_peeling): Update.
10534 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10535 * ubsan.c (ubsan_expand_null_ifn): Update.
10536 (ubsan_expand_ptr_ifn): Update.
10537 * value-prof.c (gimple_divmod_fixed_value): Update.
10538 (gimple_mod_pow2): Update.
10539 (gimple_mod_subtract): Update.
10540 (gimple_ic): Update.
10541 (gimple_stringop_fixed_value): Update.
10543 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10546 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10548 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10550 PR rtl-optimization/82395
10551 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10552 based on non_spilled_static_chain_regno_p.
10554 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10556 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10557 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10558 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10560 2017-10-19 Martin Sebor <msebor@redhat.com>
10562 PR tree-optimization/82596
10563 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10565 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10567 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10568 (get_mem_refs_of_builtin_call): Likewise.
10569 * builtins.c (expand_builtin_apply): Adjust call to
10570 allocate_dynamic_stack_space.
10571 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10572 the third argument to allocate_dynamic_stack_space, otherwise -1.
10573 (expand_builtin): Deal with all alloca variants.
10574 (is_inexpensive_builtin): Likewise.
10575 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10576 * calls.c (special_function_p): Deal with all alloca variants.
10577 (initialize_argument_information): Adjust call to
10578 allocate_dynamic_stack_space.
10579 (expand_call): Likewise.
10580 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10581 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10582 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10583 use it for the stack usage computation.
10584 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10585 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10586 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10587 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10588 (in_loop_p): Remove first argument and useless check.
10589 (pass_walloca::execute): Remove useless test and adjust call to above.
10590 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10591 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10592 (gimplify_call_expr): Deal with all alloca variants.
10593 * hsa-gen.c (gen_hsa_alloca): Likewise.
10594 (gen_hsa_insns_for_call): Likewise.
10595 * ipa-pure-const.c (special_builtin_state): Likewise.
10596 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10597 * tree-object-size.c (alloc_object_size): Likewise.
10598 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10599 (call_may_clobber_ref_p_1): Likewise.
10600 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10601 (ccp_fold_stmt): Likewise.
10602 (optimize_stack_restore): Likewise.
10603 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10604 (mark_all_reaching_defs_necessary_1): Likewise.
10605 (propagate_necessity): Likewise.
10606 (eliminate_unnecessary_stmts): Likewise.
10607 * tree.c (build_common_builtin_nodes): Build
10608 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10609 (build_alloca_call_expr): New function.
10610 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10611 (CASE_BUILT_IN_ALLOCA): Likewise.
10612 (build_alloca_call_expr): Declare.
10613 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10615 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10618 * dwarf2out.c (new_die_raw): New static inline function.
10619 (new_die): Use it to create the DIE.
10620 (add_AT_external_die_ref): Likewise.
10621 (clone_die): Likewise.
10622 (clone_as_declaration): Likewise.
10623 (dwarf2out_vms_debug_main_pointer): Likewise.
10624 (base_type_die): Likewise. Remove early return for corner cases.
10625 Do not call add_pubtype on the DIE here.
10626 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10627 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10628 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10629 native order exists for base types, attach the DIE manually and call
10630 add_pubtype on it. Do not equate a reverse order DIE to the type.
10632 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10634 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10635 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10637 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10639 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10641 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10642 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10643 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10644 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10645 * opts.c (sanitizer_opts): Add builtin.
10646 * ubsan.c (instrument_builtin): New function.
10647 (pass_ubsan::execute): Call it.
10648 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10649 * doc/invoke.texi: Document -fsanitize=builtin.
10651 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10652 builtins, store max (log2 (align), 0) into uchar field instead of
10653 align into uptr field.
10654 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10655 store uchar 0 field instead of uptr 0 field.
10656 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10657 instead of passing one address of struct with 2 locations pass
10658 two addresses of structs with 1 location each.
10659 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10660 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10661 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10662 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10663 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10664 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10665 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10666 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10668 2017-10-19 Martin Liska <mliska@suse.cz>
10671 * file-find.c (remove_prefix): Remove.
10672 * file-find.h (remove_prefix): Likewise.
10673 * gcc-ar.c: Remove smartness of lookup.
10675 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10677 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10678 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10679 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10681 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10684 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10685 (setcc + and to xor + setcc): New peephole2.
10687 2017-10-19 Tom de Vries <tom@codesourcery.com>
10689 * doc/sourcebuild.texi (Test Directives, Variants of
10690 dg-require-support): Add dg-require-stack-size.
10692 2017-10-19 Martin Liska <mliska@suse.cz>
10695 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10696 that have a large alignment.
10697 (gimplify_target_expr): Likewise.
10699 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10701 PR rtl-optimization/82602
10702 * ira.c (rtx_moveable_p): Return false for volatile asm.
10704 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10707 * config/i386/i386-modes.def (CCGZ): New CC mode.
10708 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10709 * config/i386/predicates.md (ix86_comparison_operator):
10711 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10712 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10713 with double-word subtraction.
10714 (put_condition_code): Handle CCGZmode.
10716 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10718 * wide-int.cc (debug (const wide_int &)): New.
10719 (debug (const wide_int *)): New.
10720 (debug (const widest_int &)): New.
10721 (debug (const widest_int *)): New.
10723 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10725 PR middle-end/82556
10726 * lra-constraints.c (curr_insn_transform): Use non-input operand
10727 instead of output one for matched reload.
10729 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10731 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10732 (tree-ssa-loop-ivopts.h): New header file.
10733 (struct builtin_info): New fields.
10734 (classify_builtin_1): Compute and record base and offset parts for
10735 memset builtin partition by calling strip_offset.
10736 (offset_cmp, fuse_memset_builtins): New functions.
10737 (finalize_partitions): Fuse adjacent memset partitions by calling
10739 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10740 Expose the interface.
10741 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10743 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10745 PR tree-optimization/82574
10746 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10747 that data reference must be executed exactly once per iteration
10748 against the outermost loop in nest.
10749 (classify_partition): Update call to above function.
10751 2017-10-18 Richard Biener <rguenther@suse.de>
10753 PR tree-optimization/82591
10754 * graphite.c (graphite_transform_loops): Move code gen message
10756 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10757 Here. Handle scop_to_isl_ast failing.
10758 (scop_to_isl_ast): Limit the number of ISL operations.
10760 2017-10-18 Richard Biener <rguenther@suse.de>
10762 * graphite-isl-ast-to-gimple.c
10763 (translate_isl_ast_to_gimple::set_rename): Simplify.
10764 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10765 (graphite_copy_stmts_from_block): ... here.
10766 (copy_bb_and_scalar_dependences): Simplify.
10767 (add_parameters_to_ivs_params): Canonicalize.
10768 (generate_entry_out_of_ssa_copies): Simplify.
10769 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10770 by passing in ISL dimension.
10771 (parameter_index_in_region_1): Rename to ...
10772 (parameter_index_in_region): ... this.
10773 (extract_affine): Adjust assert, pass down parameter index.
10774 (add_param_constraints): Use range-info when available.
10775 (build_scop_context): Adjust.
10776 * sese.c (new_sese_info): Adjust.
10777 (free_sese_info): Likewise.
10778 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10779 Remove unused typedefs.
10780 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10782 2017-10-18 Martin Liska <mliska@suse.cz>
10784 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10786 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10788 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10789 (s390_sched_init): Do not reset s390_sched_state if we entered the
10790 current basic block via a fallthru edge and all others are unlikely.
10792 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10794 * config/s390/s390.c (NUM_SIDES): New variable.
10795 (LONGRUNNING_THRESHOLD): New variable.
10796 (LATENCY_FACTOR): New variable.
10797 (s390_sched_score): Decrease score for long-running instructions on
10799 (s390_sched_variable_issue): Perform bookkeeping for long-running
10802 2017-10-18 Richard Biener <rguenther@suse.de>
10804 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10805 Simplify with removal of the parameter rename map.
10806 (set_rename): Likewise.
10807 (should_copy_to_new_region): Likewise.
10808 (graphite_copy_stmts_from_block): Likewise.
10809 (copy_bb_and_scalar_dependences): Remove initialization of
10810 unused copied_bb_map.
10811 (copy_def): Remove.
10812 (copy_internal_parameters): Likewise.
10813 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10814 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10815 Use INTEGRAL_TYPE_P.
10816 (parameter_index_in_region_1): Rename to ...
10817 (assign_parameter_index_in_region): ... this. Assert we have
10818 a parameter we handle.
10819 (scan_tree_for_params): Adjust.
10820 * sese.h (parameter_rename_map_t): Remove.
10821 (struct sese_info_t): Remove unused parameter_rename_map and
10822 copied_bb_map members.
10823 * sese.c (new_sese_info): Adjust.
10824 (free_sese_info): Likewise.
10826 2017-10-18 Martin Liska <mliska@suse.cz>
10829 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10830 on an abnormal edge.
10832 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10834 * doc/invoke.texi (ffunction-sections and fdata-sections):
10837 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10839 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10840 the use statement can throw internally.
10842 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10844 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10845 any RTX present on the RHS of a SET.
10846 * compare-elim.c (try_eliminate_compare): Restore comment.
10848 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10850 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10851 may be also called on tcc_type nodes.
10852 * langhooks.c (lhd_tree_size): Likewise.
10854 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10856 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10857 format_warning_at_substring.
10858 (maybe_warn): Convert source_range * param to a location_t. Pass
10859 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10860 (format_directive): Remove code to extract source_ranges and
10861 source_range * in favor of just a location_t.
10862 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10864 * substring-locations.c (format_warning_va): Convert
10865 source_range * param to a location_t.
10866 (format_warning_at_substring): Likewise.
10867 * substring-locations.h (format_warning_va): Likewise.
10868 (format_warning_at_substring): Likewise.
10870 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10872 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10874 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10875 and scatter/gather ops.
10877 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10878 vec_gather_load and vec_scatter_store.
10879 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10880 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10881 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10882 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10883 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10884 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10886 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10888 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10889 Detect FTST instruction and handle its register pops. Only pop
10890 second operand if can_pop_second_op is true.
10891 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10892 set can_pop_second_op to false in the compare_for_stack_reg call.
10894 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10895 output_fp_compare for stack register operands.
10896 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10897 instructions here. Do not emit stack register pops here. Assert
10898 that FCOMPP pops next to top stack register. Rewrite function.
10900 2017-10-17 Nathan Sidwell <nathan@acm.org>
10902 PR middle-end/82577
10903 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10904 use DECL_ASSEMBLER_NAME_RAW.
10906 PR middle-end/82546
10907 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10910 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10911 Wilco Dijkstra <wilco.dijkstra@arm.com>
10913 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10914 converstion to Pmode from the buf_addr.
10916 2017-10-17 Richard Biener <rguenther@suse.de>
10918 * graphite-dependences.c (scop_get_reads_and_writes): Change
10919 output parameters to references.
10921 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10923 PR 71026/tree-optimization
10924 * fold-const.c (distribute_real_division): Removed.
10925 (fold_binary_loc): Remove calls to distribute_real_divison.
10927 2017-10-17 Richard Biener <rguenther@suse.de>
10929 * graphite-scop-detection.c
10930 (scop_detection::stmt_has_simple_data_refs_p): Always use
10931 the full nest as region.
10932 (try_generate_gimple_bb): Likewise.
10933 * sese.c (scalar_evolution_in_region): Simplify now that
10934 SCEV can handle instantiation in regions.
10935 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10936 in the non-loop part of a function if requested.
10938 2017-10-17 Richard Biener <rguenther@suse.de>
10940 PR tree-optimization/82563
10941 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10943 (graphite_regenerate_ast_isl): Call it.
10944 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10946 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10948 PR tree-optimization/82549
10949 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10950 Formatting fixes. Instead of calling make_bit_field_ref with negative
10953 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10955 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10956 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10957 _mm_maskz_reduce_ss): New.
10958 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10959 __builtin_ia32_reducess_mask): Ditto..
10960 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10961 * config/i386/sse.md (reduces<mode>): Renamed to ...
10962 (reduces<mode><mask_scalar_name>): ... this.
10963 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10965 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10966 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10968 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10970 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10971 * selftest-run-tests.c (selftest::run_tests): Call
10972 selftest::unique_ptr_tests_cc_tests.
10973 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10974 * unique-ptr-tests.cc: New file.
10976 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10979 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10981 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10982 (make_hard_regno_born, make_hard_regno_dead): Update
10983 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10985 2017-10-16 Jeff Law <law@redhat.com>
10987 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10989 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10991 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10993 2017-10-16 Olivier Hainque <hainque@adacore.com>
10995 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10996 with_cpu if we were configured for an e500v2 target cpu name.
10998 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11000 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11001 * doc/invoke.texi: Document +nodsp as a valid extension for
11004 2017-10-16 Martin Liska <mliska@suse.cz>
11006 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11007 (test_set_range): Likewise.
11008 (test_range_functions): Rename to ...
11009 (test_bit_in_range): ... this.
11010 (sbitmap_c_tests): Add new test.
11012 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11014 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11016 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11017 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11019 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11021 * config/aarch64/aarch64-builtins.c
11022 (aarch64_types_quadopu_lane_qualifiers): New.
11023 (TYPES_QUADOPU_LANE): New.
11024 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11025 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11026 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11027 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11028 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11029 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11030 (Vdottype, DOTPROD): New.
11031 (sur): Add SDOT and UDOT.
11033 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11035 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11036 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11037 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11038 Add TARGET_DOTPROD.
11039 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11040 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11041 Enable TARGET_DOTPROD.
11042 (cortex-a75.cortex-a55): Likewise.
11043 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11045 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11047 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11048 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11050 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11052 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11053 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11054 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11055 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11056 * config/arm/types.md (neon_dot, neon_dot_q): New.
11057 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11059 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11061 * config/arm/arm.h (TARGET_DOTPROD): New.
11062 * config/arm/arm.c (arm_arch_dotprod): New.
11063 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11064 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11065 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11066 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11067 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11068 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11069 * doc/invoke.texi (armv8.2-a): Document dotprod
11071 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11073 * i386.c (ix86_vec_cost): New function.
11074 (ix86_rtx_costs): Handle vector operations better.
11075 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11076 * x86-tune-costs.h: Add new costs to all tables.
11078 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11080 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11082 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11083 divsd, sqrtss and sqrtsd
11084 * x86-tune-costs.h: Add new entries to all costs.
11085 (znver1_cost): Fix to match real instruction latencies.
11087 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11088 Michael Collison <michael.collison@arm.com>
11090 * compare-elim.c: Include emit-rtl.h.
11091 (can_merge_compare_into_arith): New function.
11092 (try_validate_parallel): Likewise.
11093 (try_merge_compare): Likewise.
11094 (try_eliminate_compare): Call the above when no previous clobber
11096 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11099 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11101 PR middle-end/62263
11102 PR middle-end/82498
11103 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11104 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11106 PR middle-end/62263
11107 PR middle-end/82498
11108 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11109 to be any operand_equal_p operands. For & (B - 1) require
11110 B to be power of 2. Recognize
11111 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11113 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11116 * optabs.c (expand_memory_blockage): Fix call of
11117 targetm.have_memory_blockage.
11119 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11122 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11123 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11124 objects to extra_objs instead of overwriting it.
11126 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11128 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11129 Use any_fp_register_operand as operand[3] predicate. Simplify
11130 equality test for operands[2] and operands[4] memory location.
11131 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11132 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11133 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11134 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11135 any_fp_register_operand as operand[1] predicate. Simplify
11136 equality test for operands[0] and operands[3] memory location.
11137 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11138 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11139 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11141 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11143 * target-insns.def: Add memory_blockage.
11144 * optabs.c (expand_memory_blockage): New function.
11145 (expand_asm_memory_barrier): Rename ...
11146 (expand_asm_memory_blockage): ... to this.
11147 (expand_mem_thread_fence): Call expand_memory_blockage
11148 instead of expand_asm_memory_barrier.
11149 (expand_mem_singnal_fence): Ditto.
11150 (expand_atomic_load): Ditto.
11151 (expand_atomic_store): Ditto.
11152 * doc/md.texi (Standard Pattern Names For Generation):
11153 Document memory_blockage instruction pattern.
11155 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11157 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11158 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11159 * config/rl78/rl78.md: New define_expand "adddi3".
11161 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11163 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11166 2017-10-13 Jeff Law <law@redhat.com>
11168 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11170 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11173 * internal-fn.c (expand_mul_overflow): If both operands have
11174 the same highpart of -1 or 0 and the topmost bit of lowpart
11175 is different, overflow is if res <= 0 rather than res < 0.
11177 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11179 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11180 TARGET_P9_VECTOR code for unaligned_load case.
11182 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11184 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11186 2017-10-13 Nathan Sidwell <nathan@acm.org>
11188 * tree-core.h (tree_contains_struct): Make bool.
11189 * tree.c (tree_contains_struct): Likewise.
11190 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11191 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11192 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11193 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11194 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11196 2017-10-13 Richard Biener <rguenther@suse.de>
11198 * graphite-isl-ast-to-gimple.c
11199 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11200 parameters and dominance check.
11201 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11202 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11203 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11204 Do not update SSA form here or do intermediate IL verification.
11205 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11206 (graphite_initialize): Remove check on the number of loops in
11207 the function and inline into graphite_transform_loops.
11208 (graphite_finalize): Inline into graphite_transform_loops.
11209 (graphite_transform_loops): Perform SSA update and IL verification
11211 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11213 2017-10-13 Richard Biener <rguenther@suse.de>
11215 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11216 graphite_expression_type_precision): Avoid global constructor
11218 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11219 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11220 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11221 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11222 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11223 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11225 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11228 * config/i386/i386.h (ix86_red_zone_size): New.
11229 * config/i386/i386.md (push peephole2s): Replace
11230 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11232 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11233 Alan Hayward <alan.hayward@arm.com>
11234 David Sherwood <david.sherwood@arm.com>
11236 * combine.c (can_change_dest_mode): Reject changes in
11237 REGMODE_NATURAL_SIZE.
11239 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11240 Alan Hayward <alan.hayward@arm.com>
11241 David Sherwood <david.sherwood@arm.com>
11243 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11244 (expand_debug_source_expr): Likewise.
11245 * combine.c (combine_simplify_rtx): Likewise.
11246 * cse.c (fold_rtx): Likewise.
11247 * fwprop.c (canonicalize_address): Likewise.
11248 * targhooks.c (default_shift_truncation_mask): Likewise.
11250 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11251 Alan Hayward <alan.hayward@arm.com>
11252 David Sherwood <david.sherwood@arm.com>
11254 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11255 (widened_mode): Likewise.
11256 (expand_unop): Likewise.
11257 * ree.c (transform_ifelse): Likewise.
11258 (merge_def_and_ext): Likewise.
11259 (combine_reaching_defs): Likewise.
11260 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11262 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11263 Alan Hayward <alan.hayward@arm.com>
11264 David Sherwood <david.sherwood@arm.com>
11266 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11267 * combine.c (gen_lowpart_for_combine): Likewise.
11268 * dwarf2out.c (rtl_for_decl_location): Likewise.
11269 * final.c (alter_subreg): Likewise.
11270 * rtlhooks.c (gen_lowpart_general): Likewise.
11271 (gen_lowpart_if_possible): Likewise.
11273 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11274 Alan Hayward <alan.hayward@arm.com>
11275 David Sherwood <david.sherwood@arm.com>
11277 * calls.c (expand_call): Use subreg_lowpart_offset.
11278 * cse.c (cse_insn): Likewise.
11279 * regcprop.c (copy_value): Likewise.
11280 (copyprop_hardreg_forward_1): Likewise.
11282 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11285 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11286 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11287 =Q constraints to +Q and into insn condition add check
11288 that operands[0] and operands[1] are equal.
11289 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11290 =Q constraints to +Q and into insn condition add check
11291 that operands[0] is equal to either operands[1] or operands[2].
11294 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11295 instead of handling MINUS_EXPR twice (once for each argument),
11296 canonicalize operand order and handle just once, use rtype where
11297 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11300 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11301 any values of __C while still being pattern recognizable as a simple
11302 rotate instruction.
11304 2017-10-13 Richard Biener <rguenther@suse.de>
11306 PR tree-optimization/82451
11308 2017-10-02 Richard Biener <rguenther@suse.de>
11310 PR tree-optimization/82355
11311 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11312 a mapping for the enclosing loop but avoid generating one for
11313 the loop tree root.
11314 (copy_bb_and_scalar_dependences): Remove premature codegen
11315 error on PHIs in blocks duplicated into multiple places.
11316 * graphite-scop-detection.c
11317 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11318 in the region use it as loop and nest to analyze the DR in.
11319 (try_generate_gimple_bb): Likewise.
11320 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11321 (add_loop_constraints): For blocks in a loop not in the region
11322 create a dimension with a single iteration.
11323 * sese.h (gbb_loop_at_index): Remove assert.
11325 * cfgloop.c (loop_preheader_edge): For the loop tree root
11326 return the single successor of the entry block.
11327 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11328 Reset the SCEV hashtable and niters.
11329 * graphite-scop-detection.c
11330 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11331 assert that we only have POLYNOMIAL_CHREC that vary in loops
11332 contained in the region.
11333 (scop_detection::graphite_can_represent_expr): Adjust.
11334 (scop_detection::stmt_has_simple_data_refs_p): For loops
11335 not in the region set loop to NULL. The nest is now the
11336 entry edge to the region.
11337 (try_generate_gimple_bb): Likewise.
11338 * sese.c (scalar_evolution_in_region): Adjust for
11339 instantiate_scev change.
11340 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11341 Make nest parameter the edge into the region.
11342 (create_data_ref): Likewise.
11343 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11344 entry edge into a region and adjust instantiate_scev calls.
11345 (create_data_ref): Likewise.
11346 (graphite_find_data_references_in_stmt): Likewise.
11347 (find_data_references_in_stmt): Pass the loop preheader edge
11348 from the nest argument.
11349 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11350 parameter the edge into the region.
11351 (instantiate_parameters): Use the loop preheader edge as entry.
11352 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11354 (get_instantiated_value_entry): Make instantiate_below parameter
11355 the edge into the region.
11356 (instantiate_scev_name): Likewise. Adjust dominance checks,
11357 when we cannot use loop-based instantiation instantiate by
11358 walking use-def chains.
11359 (instantiate_scev_poly): Adjust.
11360 (instantiate_scev_binary): Likewise.
11361 (instantiate_scev_convert): Likewise.
11362 (instantiate_scev_not): Likewise.
11363 (instantiate_array_ref): Remove.
11364 (instantiate_scev_3): Likewise.
11365 (instantiate_scev_2): Likewise.
11366 (instantiate_scev_1): Likewise.
11367 (instantiate_scev_r): Do not blindly handle N-operand trees.
11368 Do not instantiate array-refs. Handle all constants and invariants.
11369 (instantiate_scev): Make instantiate_below parameter
11370 the edge into the region.
11371 (resolve_mixers): Use the loop preheader edge for the region
11372 parameter to instantiate_scev_r.
11373 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11375 2017-10-13 Richard Biener <rguenther@suse.de>
11377 PR tree-optimization/82525
11378 * graphite-isl-ast-to-gimple.c
11379 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11381 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11382 Fail code generation when we cannot represent the isl integer.
11383 (binary_op_to_tree): Elide modulo operations that are no-ops
11384 in the type we code generate. Remove now superfluous code
11387 2017-10-13 Richard Biener <rguenther@suse.de>
11389 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11390 (scop_detection::harmful_loop_in_region): Remove premature
11391 IV type restriction.
11392 (scop_detection::graphite_can_represent_scev): We can handle
11393 pointer IVs just fine.
11395 2017-10-13 Alan Modra <amodra@gmail.com>
11397 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11398 "Clobbers and Scratch Registers". Add paragraph on
11399 alternative to clobbers for scratch registers and OpenBLAS
11402 2017-10-13 Alan Modra <amodra@gmail.com>
11404 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11405 example of a memory input for a string of known length. Move
11406 commentary out of table. Add a number of new examples
11407 covering array memory inputs.
11409 2017-10-12 Martin Liska <mliska@suse.cz>
11411 PR tree-optimization/82493
11412 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11413 (test_range_functions): New function.
11414 (sbitmap_c_tests): Likewise.
11415 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11416 * selftest.h (sbitmap_c_tests): New function.
11418 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11420 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11422 * config/rs6000/amo.h: Fix spacing issue.
11424 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11427 * config/i386/i386.md (*ashl<mode>3_mask_1,
11428 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11429 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11432 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11434 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11435 (profile_probability): Set max_probability
11436 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11438 * profile-count.c (profile_probability::differs_from_p): Do not
11439 rely on max_probaiblity == 10000
11441 2017-10-12 Jeff Law <law@redhat.com>
11443 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11446 2017-10-12 Martin Sebor <msebor@redhat.com>
11450 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11451 (handle_alias_pairs): Call it.
11452 * common.opt (-Wattribute-alias): New option.
11453 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11454 * doc/invoke.texi (-Wattribute-alias): Document.
11456 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11459 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11461 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11463 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11464 (make_hard_regno_born, make_hard_regno_dead): Update
11465 bb_killed_pseudos and bb_gen_pseudos.
11467 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11469 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11471 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11473 * config/alpha/alpha.c (alpha_split_conditional_move):
11474 Use std::swap instead of manually swapping.
11475 (alpha_stdarg_optimize_hook): Ditto.
11476 (alpha_canonicalize_comparison): Ditto.
11478 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11480 * tree-loop-distribution.c (struct builtin_info): New struct.
11481 (struct partition): Refactor fields into struct builtin_info.
11482 (partition_free): Free struct builtin_info.
11483 (build_size_arg_loc, build_addr_arg_loc): Delete.
11484 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11485 information from struct builtin_info.
11486 (find_single_drs): New function refactored from classify_partition.
11487 Also moved builtin validity checks to this function.
11488 (compute_access_range, alloc_builtin): New functions.
11489 (classify_builtin_st, classify_builtin_ldst): New functions.
11490 (classify_partition): Refactor code into functions find_single_drs,
11491 classify_builtin_st and classify_builtin_ldst.
11492 (distribute_loop): Don't do runtime alias check when distributing
11494 (find_seed_stmts_for_distribution): New function.
11495 (pass_loop_distribution::execute): Refactor code finding seed
11496 stmts into above function. Support distribution for the innermost
11497 two-level loop nest. Adjust dump information.
11499 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11501 * tree-loop-distribution.c: Adjust the general comment.
11502 (NUM_PARTITION_THRESHOLD): New macro.
11503 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11504 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11505 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11506 in call to build_partition_graph.
11507 (finalize_partitions): New parameter. Make loop distribution more
11508 conservative by fusing more partitions.
11509 (distribute_loop): Don't do runtime alias check in case of loop nest
11511 (find_seed_stmts_for_distribution): New function.
11512 (prepare_perfect_loop_nest): New function.
11513 (pass_loop_distribution::execute): Refactor code finding seed stmts
11514 and loop nest into above functions. Support loop nest distribution.
11515 Adjust dump information accordingly.
11517 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11519 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11520 and set PTYPE_SEQUENTIAL for merged partition.
11522 2017-10-12 Richard Biener <rguenther@suse.de>
11524 PR tree-optimization/69728
11526 2017-09-19 Richard Biener <rguenther@suse.de>
11528 PR tree-optimization/69728
11529 * graphite-sese-to-poly.c (schedule_error): New global.
11530 (add_loop_schedule): Handle empty domain by failing the
11532 (build_original_schedule): Handle schedule_error.
11534 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11535 domain by returning an unchanged schedule.
11537 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11539 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11540 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11542 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11544 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11547 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11550 * expr.c (store_field): Don't optimize away bitsize == 0 store
11551 from CALL_EXPR with addressable return type.
11553 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11555 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11556 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11557 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11558 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11559 TARGET_ISEL instead of TARGET_ISEL<sel>.
11561 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11563 * config/rs6000/rs6000.c
11564 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11566 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11568 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11570 (reg_or_cint_operand): Fix comment.
11571 (reg_or_zero_operand): New predicate.
11572 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11573 * config/rs6000/rs6000.c (output_isel): Delete.
11574 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11575 instead of reg_or_cint_operand. Output instruction directly (not via
11577 (isel_unsigned_<mode>): Ditto.
11578 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11579 gpc_reg_operand. Add an instruction alternative for this. Output
11580 instruction directly.
11581 (*isel_reversed_unsigned_<mode>): Ditto.
11583 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11585 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11586 (TARGET_CANONICALIZE_COMPARISON): Define.
11588 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11591 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11592 Check whether the dest is REG before adding REG_EQUIV note.
11594 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11597 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11599 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11600 (make_hard_regno_born, make_hard_regno_dead): Update
11601 bb_killed_pseudos and bb_gen_pseudos.
11603 2017-10-11 Nathan Sidwell <nathan@acm.org>
11605 * incpath.h (enum incpath_kind): Name enum, prefix values.
11606 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11607 * incpath.c (heads, tails): Use INC_MAX.
11608 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11609 (merge_include_chains, split_quote_chain,
11610 register_include_chains): Update incpath_kind names.
11611 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11612 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11614 (add_framework_path, darwin_register_objc_includes): Likewise.
11615 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11617 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11619 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11620 Do not use float_operator operator predicate.
11621 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11622 * config/i386/predicates.md (float_operator): Remove predicate.
11624 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11626 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11627 (*jccxf_i387): Ditto.
11628 (*jcc<mode>_i387): Ditto.
11629 (*jccu<mode>_i387): Ditto.
11630 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11631 (*jcc_*_i387 splitters): Remove.
11632 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11633 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11634 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11637 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11639 * profile-count.h (slow_safe_scale_64bit): New function.
11640 (safe_scale_64bit): New inline.
11641 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11642 * profile-count.c: Include wide-int.h
11643 (slow_safe_scale_64bit): New.
11645 2017-10-11 Nathan Sidwell <nathan@acm.org>
11647 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11648 HAS_DECL_ASSEMBLER_NAME_P.
11649 * gimple-expr.c (gimple_decl_printable_name: Check
11650 HAS_DECL_ASSEMBLER_NAME_P too.
11651 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11652 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11653 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11654 * passes.c (rest_of_decl_compilation): Check
11655 HAS_DECL_ASSEMBLER_NAME_P too.
11656 * recog.c (verify_changes): Likewise.
11657 * tree-pretty-print.c (dump_decl_name): Likewise.
11658 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11660 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11661 (SET_DECL_ASSEMBLER_NAME): Use it.
11662 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11663 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11664 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11666 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11668 * config.gcc (i386, x86_64): Add extra objects.
11669 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11670 (ix86_min_insn_size): Declare.
11671 (ix86_issue_rate): Declare.
11672 (ix86_adjust_cost): Declare.
11673 (ia32_multipass_dfa_lookahead): Declare.
11674 (ix86_macro_fusion_p): Declare.
11675 (ix86_macro_fusion_pair_p): Declare.
11676 (ix86_bd_has_dispatch): Declare.
11677 (ix86_bd_do_dispatch): Declare.
11678 (ix86_core2i7_init_hooks): Declare.
11679 (ix86_atom_sched_reorder): Declare.
11680 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11681 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11682 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11683 (rip_relative_addr_p): Rename to ...
11684 (ix86_rip_relative_addr_p): ... this one; export.
11685 (memory_address_length): Update.
11686 (ix86_issue_rate): Move to x86-tune-sched.c.
11687 (ix86_flags_dependent): Move to x86-tune-sched.c.
11688 (ix86_agi_dependent): Move to x86-tune-sched.c.
11689 (exact_dependency_1): Move to x86-tune-sched.c.
11690 (exact_store_load_dependency): Move to x86-tune-sched.c.
11691 (ix86_adjust_cost): Move to x86-tune-sched.c.
11692 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11693 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11694 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11695 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11696 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11697 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11698 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11699 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11700 (min_insn_size): Rename to ...
11701 (ix86_min_insn_size): ... this one; export.
11702 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11703 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11704 (core2i7_first_cycle_multipass_backtrack): Move to
11705 x86-tune-sched-core.c.
11706 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11707 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11708 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11709 (ix86_avoid_jump_mispredicts): Update.
11710 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11711 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11712 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11713 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11714 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11715 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11716 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11717 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11718 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11719 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11720 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11721 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11722 (BIG): Move to ix86-tune-sched-bd.c.
11723 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11724 (enum insn_path): Move to ix86-tune-sched-bd.c.
11725 (get_mem_group): Move to ix86-tune-sched-bd.c.
11726 (is_cmp): Move to ix86-tune-sched-bd.c.
11727 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11728 (is_branch): Move to ix86-tune-sched-bd.c.
11729 (is_prefetch): Move to ix86-tune-sched-bd.c.
11730 (init_window): Move to ix86-tune-sched-bd.c.
11731 (allocate_window): Move to ix86-tune-sched-bd.c.
11732 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11733 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11734 (process_end_window): Move to ix86-tune-sched-bd.c.
11735 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11736 (find_constant): Move to ix86-tune-sched-bd.c.
11737 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11738 (has_immediate): Move to ix86-tune-sched-bd.c.
11739 (get_insn_path): Move to ix86-tune-sched-bd.c.
11740 (get_insn_group): Move to ix86-tune-sched-bd.c.
11741 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11742 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11743 (add_insn_window): Move to ix86-tune-sched-bd.c.
11744 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11745 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11746 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11747 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11748 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11749 (do_dispatch): Move to ix86-tune-sched-bd.c.
11750 (has_dispatch): Move to ix86-tune-sched-bd.c.
11751 * config/i386/t-i386: Add new object files.
11752 * config/i386/x86-tune-costs.h: New file.
11753 * config/i386/x86-tune-sched-atom.c: New file.
11754 * config/i386/x86-tune-sched-bd.c: New file.
11755 * config/i386/x86-tune-sched-core.c: New file.
11756 * config/i386/x86-tune-sched.c: New file.
11758 2017-10-11 Liu Hao <lh_mouse@126.com>
11760 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11761 the generic version below instead.
11762 (should_colorize): Recognize Windows consoles as terminals
11764 * pretty-print.c [__MINGW32__] (write_all): New function.
11765 [__MINGW32__] (find_esc_head): Likewise.
11766 [__MINGW32__] (find_esc_terminator): Likewise.
11767 [__MINGW32__] (eat_esc_sequence): Likewise.
11768 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11770 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11773 2017-10-11 Richard Biener <rguenther@suse.de>
11775 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11776 Properly call analyze_scalar_evolution with the loop of the stmt.
11778 2017-10-11 Richard Biener <rguenther@suse.de>
11780 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11781 * tree-core.h (tree_base): Add chrec_var union member.
11782 * tree.h (CHREC_VAR): Remove.
11783 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11784 * tree-chrec.h (build_polynomial_chrec): Adjust.
11785 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11786 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11788 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11790 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11791 * match.pd: ... here.
11792 ((T) X == (T) Y): Relax condition.
11794 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11796 PR tree-optimization/82472
11797 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11799 (break_alias_scc_partitions): Update postorder number.
11801 2017-10-11 Martin Liska <mliska@suse.cz>
11804 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11806 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11808 2017-10-11 Martin Liska <mliska@suse.cz>
11810 * print-rtl.c (print_insn): Move declaration of idbuf
11811 to same scope as name.
11813 2017-10-11 Martin Liska <mliska@suse.cz>
11818 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11819 volatile arguments.
11821 2017-10-11 Martin Liska <mliska@suse.cz>
11824 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11825 volatile arguments.
11827 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11829 * config.gcc (default_gnu_indirect_function): Default to yes for
11830 arm*-*-linux* with glibc.
11832 2017-10-11 Richard Biener <rguenther@suse.de>
11834 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11835 default-defs and types we do not want to analyze.
11836 (interpret_loop_phi): Replace unreachable code with an assert.
11837 (compute_scalar_evolution_in_loop): Remove and inline ...
11838 (analyze_scalar_evolution_1): ... here, replacing condition with
11839 what makes the intent clearer. Remove handling of cases
11840 get_scalar_evolution now handles.
11842 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11844 PR rtl-optimization/81434
11845 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11846 comment for main loop. In sched_group_found if, also add checks for
11847 pass and min_cost_group.
11849 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11851 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11852 (rs6000_insn_cost): New function.
11853 * config/rs6000/rs6000.md (cost): New attribute.
11855 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11856 H.J. Lu <hongjiu.lu@intel.com>
11860 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11861 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11862 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11863 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11864 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11865 OPTION_MASK_ISA_MMX and some other bit set require both
11866 mmx and the ISAs without the mmx bit.
11867 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11868 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11869 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11870 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11871 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11872 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11873 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11874 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11875 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11876 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11877 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11878 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11879 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11880 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11881 Add OPTION_MASK_ISA_MMX.
11883 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11885 * config.gcc (armv7*-*-freebsd*): New target.
11886 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11888 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11890 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11891 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11894 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11896 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11897 a template parameter.
11898 (WIDE_INT_REF_FOR): Update accordingly.
11899 * tree.h (wi::int_traits <const_tree>): Delete.
11900 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11901 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11902 (wi::tree_to_wide_ref): New typedef.
11903 (wi::to_wide): New function.
11904 * calls.c (get_size_range): Use wi::to_wide when operating on
11905 trees as wide_ints.
11906 * cgraph.c (cgraph_node::create_thunk): Likewise.
11907 * config/i386/i386.c (ix86_data_alignment): Likewise.
11908 (ix86_local_alignment): Likewise.
11909 * dbxout.c (stabstr_O): Likewise.
11910 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11911 * expr.c (const_vector_from_tree): Likewise.
11912 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11913 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11914 (fold_negate_expr_1, int_const_binop_1, const_binop)
11915 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11916 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11917 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11918 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11919 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11920 (fold_not_const, round_up_loc): Likewise.
11921 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11922 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11923 (alloca_call_type): Likewise.
11924 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11925 * godump.c (go_output_typedef): Likewise.
11926 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11927 * internal-fn.c (get_min_precision): Likewise.
11928 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11929 * ipa-polymorphic-call.c
11930 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11931 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11932 (ipa_modify_call_arguments): Likewise.
11933 * match.pd: Likewise.
11934 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11935 * print-tree.c (print_node_brief, print_node): Likewise.
11936 * stmt.c (expand_case): Likewise.
11937 * stor-layout.c (layout_type): Likewise.
11938 * tree-affine.c (tree_to_aff_combination): Likewise.
11939 * tree-cfg.c (group_case_labels_stmt): Likewise.
11940 * tree-data-ref.c (dr_analyze_indices): Likewise.
11941 (prune_runtime_alias_test_list): Likewise.
11942 * tree-dump.c (dequeue_and_dump): Likewise.
11943 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11944 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11945 * tree-pretty-print.c (dump_generic_node): Likewise.
11946 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11947 (simple_iv_with_niters): Likewise.
11948 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11949 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11950 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11951 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11952 (refine_value_range_using_guard, number_of_iterations_ne_max)
11953 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11954 (get_cst_init_from_scev, record_nonwrapping_iv)
11955 (scev_var_range_cant_overflow): Likewise.
11956 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11957 * tree-ssa-pre.c (compute_avail): Likewise.
11958 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11959 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11960 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11961 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11962 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11963 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11964 (dump_case_nodes, try_switch_expansion): Likewise.
11965 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11966 (vect_do_peeling): Likewise.
11967 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11968 * tree-vect-stmts.c (vectorizable_load): Likewise.
11969 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11970 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11971 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11972 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11973 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11974 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11975 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11976 (evrp_dom_walker::before_dom_children): Likewise.
11977 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11978 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11979 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11980 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11981 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11982 (get_range_pos_neg): Likewise.
11983 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11984 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11985 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11986 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11987 * config/avr/avr.c (avr_fold_builtin): Likewise.
11988 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11989 * config/msp430/msp430.c (msp430_attr): Likewise.
11990 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11991 * config/powerpcspe/powerpcspe-c.c
11992 (altivec_resolve_overloaded_builtin): Likewise.
11993 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11994 (rs6000_expand_ternop_builtin): Likewise.
11995 * config/rs6000/rs6000-c.c
11996 (altivec_resolve_overloaded_builtin): Likewise.
11997 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11998 (rs6000_expand_ternop_builtin): Likewise.
11999 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12001 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12003 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12004 when copying loop nest with only one inner loop.
12006 2017-10-10 Richard Biener <rguenther@suse.de>
12008 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12009 blocks if SCEV is active.
12010 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12012 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12013 (scev_initialize): Assert we are not yet initialized.
12015 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12017 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12018 inner loop's exit stmt by making it always exit the loop, otherwise
12019 we would generate an infinite empty loop.
12021 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12023 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12024 renaming variables in new preheader if it's deleted.
12026 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12028 * tree-loop-distribution.c (struct partition): Remove unused field
12029 loops of the structure.
12030 (partition_alloc, partition_free): Ditto.
12031 (build_rdg_partition_for_vertex): Ditto.
12033 2017-10-09 Jeff Law <law@redhat.com>
12035 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12036 return type to match prototype and documentation.
12038 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12040 * config/rs6000/rs6000.c (processor_costs): Move to ...
12041 * config/rs6000/rs6000.h: ... here.
12042 (rs6000_cost): Declare.
12044 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12046 * except.c (setjmp_fn): New global variable.
12047 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12048 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12049 if DONT_USE_BUILTIN_SETJMP is defined.
12051 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12053 * target.def (insn_cost): New hook.
12054 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12055 * doc/tm.texi: Regenerate.
12056 * rtlanal.c (insn_cost): Use the new hook.
12058 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12060 * combine.c (combine_validate_cost): Compute the new insn_cost,
12061 not just pattern_cost.
12062 (try_combine): Adjust comment.
12064 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12066 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12068 * combine.c (uid_insn_cost): Adjust comment.
12069 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12071 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12072 * dse.c (find_shift_sequence): Ditto.
12073 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12074 (bb_valid_for_noce_process_p): Use pattern_cost.
12075 * rtl.h (insn_rtx_cost): Delete.
12076 (pattern_cost): New prototype.
12077 (insn_cost): New prototype.
12078 * rtlanal.c (insn_rtx_cost): Rename to...
12079 (pattern_cost): ... this.
12082 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12084 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12085 (*jcc<mode>_0_r_i387): Ditto.
12086 (*jccxf_r_i387): Ditto.
12087 (*jcc<mode>_r_i387): Ditto.
12088 (*jccu<mode>_r_i387): Ditto.
12089 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12090 (*jcc): Rename from *jcc_1.
12092 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12094 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12095 deferred rescans after the lvx/stvx recombination pre-pass.
12097 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12099 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12100 memory operation instruction support.
12101 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12102 (rs6000-ibm-aix[789]*): Likewise.
12103 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12104 Document new functions.
12106 2017-10-09 Richard Biener <rguenther@suse.de>
12108 PR tree-optimization/82397
12109 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12110 equality only for semantically equal trees.
12112 2017-10-09 Richard Biener <rguenther@suse.de>
12114 PR tree-optimization/82449
12115 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12116 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12117 allow constant addresses.
12118 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12121 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12123 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12126 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12129 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12132 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12135 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12137 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12140 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12141 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12143 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12145 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12146 (WI_BINARY_PREDICATE_RESULT): Likewise.
12147 (wi::binary_traits::operator_result): New type.
12148 (wi::binary_traits::predicate_result): Likewise.
12149 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12150 (generic_wide_int::operator==, generic_wide_int::operator!=)
12151 (generic_wide_int::operator&, generic_wide_int::and_not)
12152 (generic_wide_int::operator|, generic_wide_int::or_not)
12153 (generic_wide_int::operator^, generic_wide_int::operator+
12154 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12156 (operator~, unary operator-, operator==, operator!=, operator&)
12157 (operator|, operator^, operator+, binary operator-, operator*): New
12159 * expr.c (get_inner_reference): Use wi::bit_and_not.
12160 * fold-const.c (fold_binary_loc): Likewise.
12161 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12162 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12163 (bit_value_binop): Likewise.
12164 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12165 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12166 (extract_range_from_binary_expr_1): Likewise.
12167 (masked_increment): Likewise.
12168 (simplify_bit_ops_using_ranges): Likewise.
12170 2017-10-09 Martin Jambor <mjambor@suse.cz>
12173 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12174 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12175 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12177 (hsa_fixup_mov_insn_type): New function.
12178 (hsa_op_with_type::get_in_type): Use it.
12179 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12180 immediates in an assert.
12181 (hsa_op_with_type::extend_int_to_32bit): New method.
12182 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12183 types. Convert to dest type if necessary.
12184 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12185 (reg_for_gimple_ssa): Pass false as min32int to
12186 hsa_type_for_scalar_tree_type.
12187 (gen_hsa_addr): Fixup type when creating addresable temporary.
12188 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12189 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12190 necessary. Call hsa_fixup_mov_insn_type.
12191 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12192 extend operands and convert to dest type if necessary.
12193 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12194 to dest type if necessary.
12195 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12196 if conversion nt necessary and size matches.
12197 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12198 to dest type if necessary.
12199 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12200 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12202 (gen_hsa_clrsb): Likewise.
12203 (gen_hsa_ffs): Likewise.
12204 (gen_hsa_divmod): Extend operands and convert to dest type if
12206 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12208 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12210 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12211 Remove empty default arguments. Use a brace block as output
12213 (conditional return): Ditto.
12215 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12216 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12218 (group_ending_nop): Ditto.
12219 (doloop_end): Ditto.
12220 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12221 (splitters for those): Ditto.
12223 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12225 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12226 a conditional jump (and the compare for it) so that pc_rtx is the
12228 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12229 for the deleted and renamed ctr<mode>_internal[234] patterns.
12230 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12231 Delete second conditional return pattern.
12232 (ctr<mode>_internal2): Delete this second bdnz pattern.
12233 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12234 (ctr<mode>_internal4): Delete this second bdz pattern.
12236 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12238 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12239 (always_initialized_rtx_for_ssa_name_p): New predicate.
12240 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12241 (finish_out_of_ssa): Free new field of SA.
12242 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12243 * tree-ssa-coalesce.c: Include tree-ssa.h.
12244 (get_parm_default_def_partitions): Remove extern keyword.
12245 (get_undefined_value_partitions): New function.
12246 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12247 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12248 uninitialized bits.
12249 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12251 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12253 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12255 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12257 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12258 for targets that preffer 128bit.
12260 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12262 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12264 2017-10-08 Olivier Hainque <hainque@adacore.com>
12266 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12267 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12269 (thumb_set_return_address): Likewise.
12271 2017-10-08 Olivier Hainque <hainque@adacore.com>
12273 * common/config/arm/arm-common.c (arm_except_unwind_info):
12274 Handle DWARF2_UNWIND_INFO.
12276 2017-10-07 Michael Collison <michael.collison@arm.com>
12278 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12281 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12283 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12284 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12285 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12286 defined, force the creation of a new block for a dispatch label.
12288 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12290 * invoke.texi (Wsuggest-attribute=cold): Document.
12291 * common.opt (Wsuggest-attribute=cold): New
12292 * ipa-pure-const.c (warn_function_cold): New function.
12293 * predict.c (compute_function_frequency): Use it.
12294 * predict.h (warn_function_cold): Declare.
12296 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12298 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12301 2017-10-06 Martin Liska <mliska@suse.cz>
12303 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12304 keyword for member functions.
12305 (struct sanopt_tree_couple): New struct.
12306 (struct sanopt_tree_couple_hash): New function.
12307 (struct sanopt_ctx): Add new hash_map.
12308 (has_dominating_ubsan_ptr_check): New function.
12309 (record_ubsan_ptr_check_stmt): Likewise.
12310 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12311 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12312 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12314 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12317 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12318 aarch64_simd_valid_immediate on CONST_VECTORs.
12319 (aarch64_reg_or_bic_imm): Likewise.
12321 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12323 PR rtl-optimization/82396
12324 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12326 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12328 * graphite-dependences.c (scop_get_reads): Move code to...
12329 (scop_get_must_writes): Move code to...
12330 (scop_get_may_writes): Move code to...
12331 (scop_get_reads_and_writes): ... here.
12332 (scop_get_dependences): Call scop_get_reads_and_writes.
12334 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12336 PR tree-optimization/82434
12337 * fold-const.h (can_native_encode_type_p,
12338 can_native_encode_string_p): Remove.
12339 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12340 don't encode anything, just return what would be otherwise returned.
12341 (native_encode_fixed, native_encode_complex, native_encode_vector):
12343 (native_encode_string): Likewise. Inline by hand
12344 can_native_encode_string_p.
12345 (can_native_encode_type_p): Remove.
12346 (can_native_encode_string_p): Remove.
12347 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12348 STRING_CSTs using can_native_encode_string_p, test all
12349 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12350 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12351 argument from native_encode_expr.
12352 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12353 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12356 2017-10-06 Richard Biener <rguenther@suse.de>
12358 PR tree-optimization/82397
12359 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12360 operand_equal_p but rely on data_ref_compare_tree for detecting
12362 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12363 to match up with dr_group_sort_cmp.
12365 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12368 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12370 * config/s390/s390-builtin-types.def: Regenerate.
12372 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12375 * config/s390/s390-builtin-types.def: Regenerate.
12376 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12377 Change flag from B_VXE to B_VX.
12378 (s390_vec_min_dbl): Remove B_VXE flag.
12380 2017-10-06 Richard Biener <rguenther@suse.de>
12382 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12383 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12384 translate_isl_ast_to_gimple::is_valid_rename,
12385 translate_isl_ast_to_gimple::get_rename,
12386 translate_isl_ast_to_gimple::get_def_bb_for_const,
12387 translate_isl_ast_to_gimple::get_new_name,
12388 translate_isl_ast_to_gimple::collect_all_ssa_names,
12389 translate_isl_ast_to_gimple::copy_loop_phi_args,
12390 translate_isl_ast_to_gimple::collect_all_ssa_names,
12391 translate_isl_ast_to_gimple::copy_loop_phi_args,
12392 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12393 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12394 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12395 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12396 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12397 translate_isl_ast_to_gimple::copy_cond_phi_args,
12398 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12399 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12400 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12401 translate_isl_ast_to_gimple::rename_uses,
12402 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12403 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12404 (set_rename_for_each_def): Likewise.
12405 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12406 here. Handle rewriting SCEV analyzable uses here.
12407 (copy_bb_and_scalar_dependences): Generate code for PHI
12409 (graphite_regenerate_ast_isl): Adjust.
12410 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12411 (add_write, add_read): New functions.
12412 (build_cross_bb_scalars_def): Use it and simplify.
12413 (build_cross_bb_scalars_use): Likewise.
12414 (graphite_find_cross_bb_scalar_vars): Inline into...
12415 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12416 simulating out-of-SSA. Compute liveout and add dependencies.
12417 (build_scops): Force an empty entry block.
12418 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12420 (sese_build_liveouts): Declare.
12421 (sese_trivially_empty_bb_p): Likewise.
12422 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12423 compute liveout and debug_liveout.
12424 (sese_bad_liveouts_use): Remove.
12425 (sese_reset_debug_liveouts_bb): Likewise.
12426 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12427 (sese_build_liveouts): Build liveout and debug_liveout and store
12429 (new_sese_info): Adjust.
12430 (free_sese_info): Likewise.
12431 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12432 do not build liveout here.
12433 (move_sese_in_condition): Adjust region entry.
12434 (scev_analyzable_p): Match up with chrec_apply requirements.
12435 (sese_trivially_empty_bb_p): New.
12436 * tree-into-ssa.c (get_reaching_def): Properly support generating
12437 default-defs for incremental rewrite of anonymous names.
12439 2017-10-06 Richard Biener <rguenther@suse.de>
12441 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12442 precision do not perform modulo reduction.
12444 2017-10-06 Richard Biener <rguenther@suse.de>
12446 PR tree-optimization/82436
12447 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12448 conservatively choose the vectorization factor when checking
12449 whether we can perform the required load permutation.
12450 (vect_transform_slp_perm_load): Assert when we may not fail.
12452 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12454 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12455 message for incompatible -msdata=* and -mcall-* options.
12457 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12459 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12460 rate for post-reload scheduling.
12462 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12464 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12466 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12468 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12469 to improve monte carlo in scimark.
12471 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12473 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12474 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12475 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12476 pentium4_cost, nocona_cost): Set reassociation width to 1.
12477 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12478 width to 2 for fp operations and 1 otherwise.
12479 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12481 (atom_cost): Set reassociation width to 2.
12482 (slm_cost, generic_cost): Set fp reassociation width
12483 to 2 and 1 otherwise.
12484 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12485 (core_cost): Set fp reassociation width to 4 and vector to 2.
12486 (ix86_reassociation_width): Rewrite using cost table; special case
12487 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12488 and TARGET_AVX128_OPTIMAL.
12489 * config/i386/i386.h (processor_costs): Add
12490 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12491 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12492 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12493 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12494 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12495 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12497 2017-10-05 Nathan Sidwell <nathan@acm.org>
12499 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12501 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12503 * config/arm/arm.c (arm_test_fpu_data): New.
12504 (arm_run_selftests): Call arm_test_fpu_data.
12506 2017-10-04 Nathan Sidwell <nathan@acm.org>
12508 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12510 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12512 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12514 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12515 check type for aarch64_simd_valid_immediate.
12516 (aarch64_output_simd_mov_immediate): Update prototype.
12517 (aarch64_simd_valid_immediate): Update prototype.
12518 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12519 support for ORR-immediate.
12520 (and<mode>3): modified pattern to add support for BIC-immediate.
12521 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12522 now checks for valid immediate for BIC and ORR based on new enum
12524 (aarch64_output_simd_mov_immediate): Function now used to output
12525 BIC/ORR imm as well based on new enum argument.
12526 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12528 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12529 (aarch64_reg_or_bic_imm): Likewise.
12531 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12533 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12534 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12536 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12540 PR rtl-optimization/82396
12541 * haifa-sched.c (autopref_multipass_init): Simplify
12543 (autopref_rank_data): Simplify sort order.
12544 * sched-int.h (autopref_multipass_data_): Remove
12545 multi_mem_insn_p, min_offset and max_offset.
12547 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12549 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12551 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12553 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12554 vect_doubleint_cvt.
12556 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12558 * doc/sourcebuild.texi: Document vect_long_mult.
12560 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12562 PR tree-optimization/82413
12563 * fold-const.c (build_range_check): Use widest_int when comparing
12564 the maximum ETYPE value with HIGH.
12566 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12568 PR rtl-optimization/82396
12569 * haifa-sched.c (autopref_multipass_init): Simplify
12571 (autopref_rank_data): Simplify sort order.
12572 * sched-int.h (autopref_multipass_data_): Remove
12573 multi_mem_insn_p, min_offset and max_offset.
12575 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12577 PR tree-optimization/82381
12578 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12579 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12582 PR tree-optimization/82374
12583 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12584 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12585 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12586 current_function_decl to the new decl.
12588 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12590 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12591 helper macro for IEEE float128 hardware built-in functions.
12592 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12594 (TRUNCF128_ODD): Likewise.
12595 (ADDF128_ODD): Likewise.
12596 (SUBF128_ODD): Likewise.
12597 (MULF128_ODD): Likewise.
12598 (DIVF128_ODD): Likewise.
12599 (FMAF128_ODD): Likewise.
12600 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12601 UNSPEC_TRUNC_ROUND_TO_ODD.
12602 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12603 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12604 floating point round to odd instructions.
12605 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12606 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12607 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12608 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12609 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12610 (trunc<mode>sf2_hw): Change the truncate with round to odd
12611 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12612 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12613 to odd hardware instructions.
12614 (sub<mode>3_odd): Likewise.
12615 (mul<mode>3_odd): Likewise.
12616 (div<mode>3_odd): Likewise.
12617 (sqrt<mode>2_odd): Likewise.
12618 (fma<mode>4_odd): Likewise.
12619 (fms<mode>4_odd): Likewise.
12620 (nfma<mode>4_odd): Likewise.
12621 (nfms<mode>4_odd): Likewise.
12622 (trunc<mode>df2_odd): Change the truncate with round to odd
12623 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12625 * doc/extend.texi (PowerPC built-in functions): Update documentation
12626 for existing IEEE float128-bit built-in functions. Add built-in
12627 functions that generate the IEEE 128-bit floating point round to
12630 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12632 PR rtl-optimization/77729
12633 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12634 to (X&(C1&~C2))|C2 transformations.
12636 2017-10-03 Martin Jambor <mjambor@suse.cz>
12638 PR tree-optimization/82363
12639 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12640 mismatch, mark lacc written regardless of racc.
12642 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12644 PR tree-optimization/82381
12645 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12646 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12647 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12648 fallthrough into reassoc_stmt_dominates_stmt_p.
12651 * combine.c (combine_instructions): Don't combine in unreachable
12654 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12657 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12658 function to not use the have_cpu variable. Do not set cpu_index,
12659 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12660 or the default cpu.
12661 (rs6000_valid_attribute_p): Remove duplicate initializations of
12662 old_optimize and func_optimize.
12663 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12664 (rs6000_activate_target_options): Make global.
12665 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12668 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12670 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12671 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12672 Set *poffset to 0 and *psize and *pmax_size to -1 if
12673 *poffset + *psize overflows in HOST_WIDE_INT.
12675 PR tree-optimization/82387
12676 PR tree-optimization/82388
12677 PR tree-optimization/82389
12678 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12679 instead of live_bytes non-NULL.
12681 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12684 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12687 2017-10-02 Richard Biener <rguenther@suse.de>
12689 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12690 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12691 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12693 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12695 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12696 requested precision matches the type's.
12697 * calls.c (alloc_max_size): Calculate the new candidate size as
12698 a widest_int and use wi::to_widest when comparing it with the
12699 current candidate size.
12700 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12701 zero rather than integer_zero_node.
12702 * match.pd: Check for a no-op conversion before using wi::add
12703 rather than after. Use tree_to_uhwi when summing small shift
12704 counts into an unsigned int.
12706 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12707 Alan Hayward <alan.hayward@arm.com>
12708 David Sherwood <david.sherwood@arm.com>
12711 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12712 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12713 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12714 POINTER_AND_FP_REGS.
12716 2017-10-02 Richard Biener <rguenther@suse.de>
12718 PR tree-optimization/82355
12719 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12720 a mapping for the enclosing loop but avoid generating one for
12721 the loop tree root.
12722 (copy_bb_and_scalar_dependences): Remove premature codegen
12723 error on PHIs in blocks duplicated into multiple places.
12724 * graphite-scop-detection.c
12725 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12726 in the region use it as loop and nest to analyze the DR in.
12727 (try_generate_gimple_bb): Likewise.
12728 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12729 (add_loop_constraints): For blocks in a loop not in the region
12730 create a dimension with a single iteration.
12731 * sese.h (gbb_loop_at_index): Remove assert.
12733 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12735 * omp-expand.c (adjust_context_scope): New function.
12736 (expand_parallel_call): Call adjust_context_scope.
12738 2017-10-01 Jeff Law <law@redhat.com>
12740 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12741 dom_opt_dom_walker class with direct access to private members.
12742 Add comments. Call test_for_singularity.
12743 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12744 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12745 m_dummy_cond anymore.
12746 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12748 (pass_dominator:execute): Build the dummy_cond here and pass it
12749 to the dom_opt_dom_walker ctor.
12750 (test_for_singularity): New function.
12752 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12753 Maya Rashish <coypu@sdf.org>
12755 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12756 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12757 (alpha*-*-netbsd*) Use nbsd_tm_file.
12758 (arm*-*-netbsdelf*) Likewise.
12759 (i[34567]86-*-netbsdelf*) Likewise.
12760 (x86_64-*-netbsd*) Likewise.
12761 (mips*-*-netbsd*) Likewise.
12762 (powerpc-*-netbsd*) Likewise.
12763 (sh*-*-netbsd*) Likewise.
12764 (sparc-*-netbsdelf*) Likewise.
12765 (sparc64-*-netbsd*) Likewise.
12766 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12768 (vax-*-netbsdelf*) Likewise.
12769 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12770 (UINT_FAST8_TYPE) Likewise.
12771 (INT_FAST16_TYPE) Check CHAR_FAST16.
12772 (UINT_FAST16_TYPE) Likewise.
12774 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12777 * config/i386/i386.md
12778 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12779 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12780 *divmodsi4_zext_2): New define_insn_and_split.
12781 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12782 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12783 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12784 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12785 New define_insn_and_split.
12786 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12787 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12788 operands[1] having DImode when mode is SImode.
12790 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12791 always SImode for DIV and MOD in REG_EQUAL notes.
12793 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12795 PR middle-end/82319
12796 * match.pd: Fix handling of NaNs in pattern.
12798 2017-09-29 Jeff Law <law@redhat.com>
12800 * sbitmap.c (bitmap_bit_in_range_p): New function.
12801 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12802 * tree-ssa-dse.c (live_bytes_read): New function.
12803 (dse_classify_store): Ignore reads of dead bytes.
12805 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12806 typos and whitespace errors.
12807 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12808 * config/s390/s390.c (s390_emit_prologue): Likewise.
12810 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12813 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12814 with a symbol for LRA.
12816 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12818 PR rtl-optimization/82338
12819 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12821 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12823 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12824 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12825 (qsort_chk): Declare.
12826 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12827 (qsort_chk): New function.
12829 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12831 PR tree-optimization/82337
12832 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12833 phi definition if the PHI result appears in an abnormal PHI.
12834 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12835 basis appears in an abnormal PHI.
12837 2017-09-29 Richard Biener <rguenther@suse.de>
12839 * graphite-isl-ast-to-gimple.c
12840 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12841 (binary_op_to_tree): Use it.
12842 (get_rename_from_scev): Likewise.
12843 (copy_loop_phi_nodes): Likewise.
12844 (copy_bb_and_scalar_dependences): Likewise.
12845 (translate_pending_phi_nodes): Likewise.
12847 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12850 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12851 for movabsq $(i32 << shift), r64.
12853 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12855 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12856 index when encoding %esp as %rsp to avoid 0x67 prefix.
12858 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12860 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12861 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12863 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12865 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12866 Extensions with more than 16 double VFP registers.
12867 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12868 to_clear_mask and all code related to it. Replace the remaining
12869 entry by a sbitmap and adapt code accordingly.
12871 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12873 * brig-builtins.def: Change pure attributes to const.
12875 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12877 * config.gcc (default_gnu_indirect_function): Default to yes for
12878 sparc*-*-linux* with glibc.
12880 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12882 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12883 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12884 when creating .init_array and .fini_array sections with priority
12887 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12890 * config/aarch64/aarch64.c
12891 (aarch64_builtin_support_vector_misalignment): Always return false
12892 when misalignment is unknown.
12894 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12896 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12897 this function to return false if the definition used by the swap
12898 instruction is artificial, or if the memory address from which the
12899 constant value is loaded is not represented by a base address held
12900 in a register or if the base address register is a frame or stack
12901 pointer. Additionally, return false if the base address of the
12902 loaded constant is a SYMBOL_REF but is not considered to be a
12904 (replace_swapped_load_constant): New function.
12905 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12906 loaded constant vector with a load of a swapped constant vector.
12908 2017-09-27 Carl Love <cel@us.ibm.com>
12910 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12911 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12912 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12915 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12917 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12918 first, always call autopref_rank_data otherwise.
12920 2017-09-27 Richard Biener <rguenther@suse.de>
12922 * graphite-scop-detection.c (find_scop_parameters): Move
12923 loop bound handling ...
12924 (gather_bbs::before_dom_children): ... here, avoiding the need
12925 to build scop_info->loop_nest.
12926 (record_loop_in_sese): Remove.
12927 * sese.h (sese_info_t::loop_nest): Remove.
12928 * sese.c (new_sese_info): Do not allocate loop_nest.
12929 (free_sese_info): Do not free loop_nest.
12931 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12934 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12935 lhs from calls if the lhs has addressable type.
12937 2017-09-27 Richard Biener <rguenther@suse.de>
12939 * graphite.h (scop::max_alias_set): New member.
12940 * graphite-scop-detection.c: Remove references to non-existing
12941 --param in comments.
12942 (build_alias_sets): Record the maximum alias set used for drs.
12943 (build_scops): Support zero as unlimited for
12944 --param graphite-max-arrays-per-scop.
12945 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12946 and inline into ...
12947 (build_poly_sr_1): ... here. Compute alias set based on the
12948 maximum alias set used for drs rather than
12949 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12951 2017-09-27 Richard Biener <rguenther@suse.de>
12953 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12954 --param loop-block-tile-size=0 to disable tiling.
12956 2017-09-27 Richard Biener <rguenther@suse.de>
12958 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12959 (graphite-max-nb-scop-params): Document special value zero.
12960 * domwalk.h (dom_walker::STOP): New symbolical constant.
12961 (dom_walker::dom_walker): Add optional parameter for bb to
12963 (dom_walker::~dom_walker): Declare.
12964 (dom_walker::before_dom_children): Document STOP return value.
12965 (dom_walker::m_user_bb_to_rpo): New member.
12966 (dom_walker::m_bb_to_rpo): Likewise.
12967 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12968 mapping here if not provided by the user.
12969 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12970 provided by the user.
12971 (dom_walker::STOP): Define.
12972 (dom_walker::walk): Do not compute bb to RPO mapping here.
12973 Support STOP return value from before_dom_children to stop
12975 * graphite-optimize-isl.c (optimize_isl): If the schedule
12976 is the same still generate code if -fgraphite-identity
12977 or -floop-parallelize-all are given.
12978 * graphite-scop-detection.c: Include cfganal.h.
12979 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12981 (gather_bbs::before_dom_children): Return STOP for BBs
12983 (build_scops): Compute bb to RPO mapping and pass it to
12984 the domwalk. Treat --param graphite-max-nb-scop-params=0
12985 as not limiting the number of params.
12986 * graphite.c (graphite_initialize): Remove limit on the
12987 number of basic-blocks in a function.
12988 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12989 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12990 default value of 10.
12992 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12994 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12995 Adjust code to eliminate needing to do the shift right 32-bits
12996 operation after XSCVDPSPN.
12998 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13000 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13001 ((X / Y) != 0 -> X >= Y): Likewise.
13003 2017-09-26 Carl Love <cel@us.ibm.com>
13005 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13006 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13007 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13008 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13009 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13010 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13011 definitions and overloading.
13012 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13013 statement for P9V_BUILTIN_XST_LEN_R.
13014 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13015 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13016 define_expand and define_insn for the instructions and builtins.
13017 * doc/extend.texi: Update the built-in documentation file for the new
13018 built-in functions.
13019 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13020 define_insn for the instructions
13022 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13025 * gcc/config/netbsd-protos.h: New file.
13026 * gcc/config/netbsd.c: New file.
13027 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13028 * gcc/config/t-netbsd: New file.
13029 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13030 (tmake_file) Add t-netbsd.
13031 (extra_objs) Add netbsd.o.
13033 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13037 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13039 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13041 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13042 sign extension from a vector register to a GPR by doing a 32-bit
13043 direct move and then an EXTSW.
13044 (extendsi<mode>2 splitter): Likewise.
13045 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13046 right or vector extract after doing XSCVDPSPN. Use
13047 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13049 (movdi_from_sf_zero_ext): Likewise.
13050 (reload_gpr_from_vsxsf): Likewise.
13051 (p8_mfvsrd_4_disf): Delete, no longer used.
13052 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13053 then needing to move the SFmode to a GPR to use the XSCVDPSP
13054 instruction instead of FRSP and XSCVDPSPN.
13055 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13056 it is adjacent to the other XSCVSPDP insns.
13057 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13058 SFmode to be in traditional Altivec registers.
13059 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13060 (vsx_xscvspdpn): Likewise.
13061 (vsx_xscvspdpn_scalar): Likewise.
13063 2017-09-26 Martin Jambor <mjambor@suse.cz>
13065 * tree-sra.c (compare_access_positions): Put integral types first,
13066 stabilize sorting of integral types, remove conditions putting
13067 non-full-precision integers last.
13068 (sort_and_splice_var_accesses): Disable scalarization if a
13069 non-integert would be represented by a non-full-precision integer.
13071 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13073 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13074 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13075 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13076 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13077 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13078 conditionals inside the function instead of around it. Call
13079 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13080 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13082 2017-09-26 Richard Biener <rguenther@suse.de>
13084 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13086 (scop_detection::build_scop_breadth): ... this. Removed.
13087 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13088 (scop_detection::harmful_stmt_in_bb): Likewise.
13089 (scop_detection::graphite_can_represent_stmt): Likewise.
13090 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13091 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13092 (scop_detection::can_represent_loop_1): ... this. Removed.
13093 (scop_detection::harmful_loop_in_region): Simplify after inlining
13094 the above and remove more quadraticness.
13095 (build_scops): Adjust.
13096 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13099 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13102 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13103 REGNO (base) == SP_REG if base is a REG.
13105 PR middle-end/35691
13106 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13107 if it is different SSA_NAME.
13108 (optimize_range_tests_cmp_bitwise): New function.
13109 (optimize_range_tests): Call it.
13111 2017-09-26 Richard Biener <rguenther@suse.de>
13113 PR tree-optimization/82321
13114 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13115 for the def being inside the loop.
13117 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13119 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13121 * config/s390/s390-builtins.def: Fix constraint on op4.
13123 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13125 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13126 independent expanders.
13127 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13128 ("vec_ordered", "vec_unordered"): New expanders.
13130 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13132 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13135 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13137 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13138 vec_unpacks_lo_v16qi.
13139 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13141 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13143 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13144 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13145 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13147 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13149 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13151 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13153 ("*vec_slb<mode>"): New insn pattern.
13154 ("vec_shr_<mode>"): New expander.
13155 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13156 and force the shift count operand to V16QImode.
13157 ("vec_srb<mode>"): Set shift count mode to V16QI.
13159 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13161 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13162 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13163 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13165 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13168 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13170 2017-09-26 Richard Biener <rguenther@suse.de>
13172 PR tree-optimization/82320
13173 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13176 2017-09-25 Jeff Law <law@redhat.com>
13178 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13179 prototype for new argument.
13180 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13181 mostly extracted from rs6000_emit_allocate_stack.
13182 (rs6000_emit_probe_stack_range_stack_clash): New function.
13183 (rs6000_emit_allocate_stack): Call
13184 rs6000_emit_probe_stack_range_stack_clash as needed.
13185 (rs6000_emit_probe_stack_range): Add additional argument
13186 to call to gen_probe_stack_range{si,di}.
13187 (output_probe_stack_range): New.
13188 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13189 (output_probe_stack_range_stack_clash): New.
13190 (rs6000_emit_prologue): Emit notes into dump file as requested.
13191 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13192 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13193 Add additional operand and pass it to output_probe_stack_range.
13195 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13197 PR tree-optimization/82163
13198 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13199 (checking_verify_loop_closed_ssa): New parameter.
13200 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13201 (check_loop_closed_ssa_stmt): Delete.
13202 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13203 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13204 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13207 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13209 * brig-builtins.def: Treat HSAIL barrier builtins as
13210 setjmp/longjump style functions.
13212 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13214 * target.def (constant_alignment): New hook.
13215 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13216 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13217 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13218 * doc/tm.texi: Regenerate.
13219 * targhooks.h (default_constant_alignment): Declare.
13220 (constant_alignment_word_strings): Likewise.
13221 * targhooks.c (default_constant_alignment): New function.
13222 (constant_alignment_word_strings): Likewise.
13223 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13224 instead of CONSTANT_ALIGNMENT.
13225 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13226 (force_const_mem): Likewise.
13227 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13228 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13229 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13230 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13231 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13233 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13234 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13235 constant_alignment_word_strings.
13236 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13237 (CONSTANT_ALIGNMENT): Likewise.
13238 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13239 (arm_constant_alignment): New function.
13240 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13241 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13242 constant_alignment_word_strings.
13243 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13244 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13245 constant_alignment_word_strings.
13246 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13247 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13248 (cris_constant_alignment): New function.
13249 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13250 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13251 (epiphany_constant_alignment): New function.
13252 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13253 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13254 constant_alignment_word_strings.
13255 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13256 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13257 constant_alignment_word_strings.
13258 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13259 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13260 constant_alignment_word_strings.
13261 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13262 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13263 * config/i386/i386.c (ix86_constant_alignment): Make static.
13264 Use the same interface as the target hook.
13265 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13266 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13267 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13268 constant_alignment_word_strings.
13269 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13270 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13271 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13272 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13273 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13274 constant_alignment_word_strings.
13275 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13276 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13277 constant_alignment_word_strings.
13278 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13279 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13280 constant_alignment_word_strings.
13281 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13282 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13284 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13285 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13286 * config/mips/mips.c (mips_constant_alignment): New function.
13287 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13288 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13289 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13290 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13291 (mmix_constant_alignment): Make static. Use the same interface
13292 as the target hook.
13293 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13294 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13295 constant_alignment_word_strings.
13296 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13297 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13298 constant_alignment_word_strings.
13299 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13300 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13301 constant_alignment_word_strings.
13302 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13303 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13304 (rs6000_constant_alignment): New function.
13305 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13306 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13307 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13308 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13309 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13310 (rs6000_constant_alignment): New function.
13311 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13312 * config/s390/s390.c (s390_constant_alignment): New function.
13313 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13314 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13315 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13316 constant_alignment_word_strings.
13317 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13318 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13319 (sparc_constant_alignment): New function.
13320 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13321 * config/spu/spu.c (spu_constant_alignment): New function.
13322 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13323 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13324 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13325 constant_alignment_word_strings.
13326 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13327 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13328 constant_alignment_word_strings.
13329 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13330 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13331 constant_alignment_word_strings.
13332 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13333 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13334 (visium_constant_alignment): New function.
13335 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13336 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13337 (xtensa_constant_alignment): New function.
13338 * system.h (CONSTANT_ALIGNMENT): Poison.
13340 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13342 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13343 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13344 (rs6000_builtin_valid_without_lhs): New helper function.
13345 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13346 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13348 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13350 * target.h (vec_perm_indices): Use unsigned short rather than
13352 (auto_vec_perm_indices): Likewise.
13353 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13354 Use unsigned int rather than unsigned char.
13355 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13357 2017-09-25 Richard Biener <rguenther@suse.de>
13359 * cfgloop.h (sort_sibling_loops): Declare.
13360 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13361 (sort_sibling_loops): New function sorting the sibling loop list
13363 * graphite.c (graphite_transform_loops): Sort sibling loops.
13365 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13367 * target.def (vec_perm_const_ok): Change sel parameter to
13369 * optabs-query.c (can_vec_perm_p): Update accordingly.
13370 * doc/tm.texi: Regenerate.
13371 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13372 auto_vec_perm_indices and remove separate nelt field.
13373 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13374 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13375 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13376 (aarch64_expand_vec_perm_const): Update accordingly.
13377 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13378 to vec_perm_indices.
13379 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13380 auto_vec_perm_indices and remove separate nelt field.
13381 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13382 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13383 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13385 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13386 to vec_perm_indices.
13387 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13388 sel to vec_perm_indices.
13389 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13390 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13391 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13393 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13396 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13399 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13400 on the FUNCTION_DECL function context if it has a DIE that is a
13403 2017-09-25 Richard Biener <rguenther@suse.de>
13405 PR tree-optimization/82285
13406 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13409 2017-09-25 Tom de Vries <tom@codesourcery.com>
13413 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13416 2017-09-25 Richard Biener <rguenther@suse.de>
13418 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13419 ISL errors other than isl_error_quota happen. Dump if the
13420 schedule is the same.
13421 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13422 errors instead of aborting inside ISL.
13424 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13427 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13428 of libgcc_eh for m64.
13429 * config/i386/darwin64.h: Likewise.
13431 2017-09-25 Richard Biener <rguenther@suse.de>
13433 PR middle-end/82144
13434 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13435 attribute for incomplete types nor twice for complete ones.
13437 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13440 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13441 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13444 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13447 * config/i386/i386.opt (mprefer-avx256): Use
13448 ix86_target_flags variable.
13449 * config/i386/i386.c (ix86_target_string): Move
13450 -mprefer-avx256 to flag2_opts.
13452 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13454 PR middle-end/35691
13455 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13456 and x != -1 | y != -1 into (x & y) != -1.
13458 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13460 * config.gcc: Add new case statement to set
13461 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13462 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13463 s390x-*-linux* case statements. Added aarch64 to the list of
13464 supported architectures.
13466 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13468 PR tree-optimization/82289
13469 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13470 STMT_VINFO_RELEVANT_P.
13472 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13473 Alan Hayward <alan.hayward@arm.com>
13474 David Sherwood <david.sherwood@arm.com>
13476 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13477 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13478 (extract_range_from_binary_expr_1): Don't call
13479 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13481 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13482 Alan Hayward <alan.hayward@arm.com>
13483 David Sherwood <david.sherwood@arm.com>
13485 * target.def (preferred_vector_alignment): New hook.
13486 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13488 * doc/tm.texi: Regenerate.
13489 * targhooks.h (default_preferred_vector_alignment): Declare.
13490 * targhooks.c (default_preferred_vector_alignment): New function.
13491 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13493 (DR_TARGET_ALIGNMENT): New macro.
13494 (aligned_access_p): Update commentary.
13495 (vect_known_alignment_in_bytes): New function.
13496 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13498 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13499 Calculate the misalignment based on the target alignment rather than
13501 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13502 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13503 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13504 the target alignment, rather than masking the element misalignment
13505 with the number of elements in a vector. Also use the target
13506 alignment when calculating the maximum number of peels.
13507 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13508 instead of TYPE_ALIGN_UNIT.
13509 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13510 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13511 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13512 (vect_create_data_ref_ptr): Likewise.
13513 (vect_setup_realignment): Realign by ANDing with
13514 -DR_TARGET_MISALIGNMENT.
13515 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13516 the number of peels based on DR_TARGET_ALIGNMENT.
13517 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13518 with the guaranteed alignment boundary when deciding whether
13520 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13521 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13522 (ensure_base_align): Remove stmt_info parameter. Get the
13523 target base alignment from DR_TARGET_ALIGNMENT.
13524 (vectorizable_store): Update call accordingly. Interpret
13525 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13527 (vectorizable_load): Likewise.
13529 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13530 Alan Hayward <alan.hayward@arm.com>
13531 David Sherwood <david.sherwood@arm.com>
13533 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13534 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13535 (vect_enhance_data_refs_alignment): Likewise.
13537 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13539 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13540 error. Only quit immediately if parsing is complete.
13541 (BEGIN): Initialize fatal_err and parse_done.
13542 (begin fpu, end fpu): Check number of arguments.
13543 (begin arch, end arch): Likewise.
13544 (begin cpu, end cpu): Likewise.
13545 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13546 (optalias): Likewise.
13548 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13550 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13551 * config/arm/arm-isa.h: Delete. Move definitions to ...
13552 * arm-cpus.in: ... here. Use new feature and fgroup values.
13553 * config/arm/arm.c (arm_option_override): Use lower case for feature
13555 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13556 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13557 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13559 (print_isa_bits_for): New function.
13560 (gen_isa): New function.
13561 (gen_comm_data): Use print_isa_bits_for.
13562 (define feature): New keyword.
13563 (define fgroup): New keyword.
13564 * config/arm/t-arm (TM_H): Remove.
13565 (GTM_H): Add arm-isa.h.
13566 (arm-isa.h): Add rule to generate file.
13567 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13568 case for feature bit names.
13570 2017-09-22 Richard Biener <rguenther@suse.de>
13572 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13574 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13575 print of no dependency loops ...
13576 * graphite.c (graphite_transform_loops): ... here.
13577 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13579 (same_close_phi_node, remove_duplicate_close_phi,
13580 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13581 (canonicalize_loop_closed_ssa): ... here and simplify.
13582 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13583 (optimize_isl): Use dump_printf_loc to tell when we stopped
13584 optimizing because of an ISL timeout.
13586 2017-09-22 Richard Biener <rguenther@suse.de>
13588 PR tree-optimization/82291
13589 * tree-if-conv.c (predicate_mem_writes): Make sure to
13590 remove writes in blocks predicated with false.
13592 2017-09-22 Richard Biener <rguenther@suse.de>
13594 * sese.c: Include cfganal.h.
13595 (if_region_set_false_region): Remove.
13596 (create_if_region_on_edge): Likewise.
13597 (move_sese_in_condition): Re-implement without destroying
13600 2017-09-22 Richard Biener <rguenther@suse.de>
13602 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13603 Verify both BBs contain loop PHI nodes before dispatching to
13604 copy_loop_phi_args.
13605 (graphite_regenerate_ast_isl): Do not recompute dominators,
13606 do not verify three times. Restructure for clarity.
13607 * graphite-scop-detection.c (same_close_phi_node,
13608 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13609 defined_in_loop_p, canonicalize_loop_closed_ssa,
13610 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13611 checking and SSA rewrite, move to ...
13612 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13613 (graphite_initialize): Do not pass in ctx, do not reset the
13614 SCEV cache, compute only dominators.
13615 (graphite_transform_loops): Allocate ISL ctx after
13616 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13617 Maintain post-dominators only around build_scops.
13618 * sese.c (if_region_set_false_region): Make static. Free
13619 and recompute dominators.
13620 (move_sese_in_condition): Assert we don't get called with
13621 post-dominators computed.
13622 * sese.h (if_region_set_false_region): Remove.
13624 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13626 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13627 mode attribute for TARGET_AVX512VL.
13629 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13631 * config/i386/i386.opt (mprefer-avx256): New option.
13632 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13634 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13635 for TARGET_PREFER_AVX256.
13636 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13638 2017-09-21 Jeff Law <law@redhat.com>
13640 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13641 Fix dump output if the only stack space is for pushed registers.
13643 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13645 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13648 2017-09-21 Martin Sebor <msebor@redhat.com>
13651 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13652 code (in C++) or code that triggers warnings.
13654 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13656 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13658 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13660 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13661 * sched-rgn.c: Ditto.
13662 * sel-sched-ir.c: Ditto.
13664 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13666 * toplev.h (set_random_seed): Adjust return type.
13667 * toplev.c (init_local_tick): Move eager initialization of random_seed
13668 to get_random_seed. Adjust comment.
13669 (init_random_seed): Inline to get_random_seed, delete.
13670 (get_random_seed): Initialize random_seed lazily.
13671 (set_random_seed): Do not return previous value.
13672 (print_switch_value): Do not call get_random_seed.
13674 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13676 * cgraph.c (delete_function_version): New, broken out from...
13677 (cgraph_node::delete_function_version): ...here. Rename to
13678 cgraph_node::delete_function_version_by_decl. Update all uses.
13679 (cgraph_node::remove): Call delete_function_version.
13681 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13684 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13685 VAR_DECLs to which addressable non-volatile parameters are mapped
13686 and for id->retvar after the return value assignment. Clear
13687 id->retval and id->retbnd after inlining.
13689 2017-09-21 Richard Biener <rguenther@suse.de>
13691 PR tree-optimization/82276
13692 PR tree-optimization/82244
13693 * tree-vrp.c (build_assert_expr_for): Set
13694 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13696 (remove_range_assertions): Revert earlier change.
13698 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13701 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13703 2017-09-21 Richard Biener <rguenther@suse.de>
13705 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13706 Restore valid IL after code generation errors.
13707 * graphite.c (graphite_transform_loops): Diagnose code
13708 generation issues as MSG_MISSED_OPTIMIZATION and continue
13709 with processing SCOPs.
13711 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13712 Alan Hayward <alan.hayward@arm.com>
13713 David Sherwood <david.sherwood@arm.com>
13715 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13716 rather than choosing between plus_constant and gen_rtx_<CODE>.
13717 * expr.c (emit_push_insn): Likewise.
13718 (expand_expr_real_2): Likewise.
13720 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13721 Alan Hayward <alan.hayward@arm.com>
13722 David Sherwood <david.sherwood@arm.com>
13724 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13726 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13727 Alan Hayward <alan.hayward@arm.com>
13728 David Sherwood <david.sherwood@arm.com>
13730 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13731 calling tree_to_uhwi.
13733 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13734 Alan Hayward <alan.hayward@arm.com>
13735 David Sherwood <david.sherwood@arm.com>
13737 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13738 INTEGER_CST rather than a negative test for ADDR_EXPR.
13740 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13741 Alan Hayward <alan.hayward@arm.com>
13742 David Sherwood <david.sherwood@arm.com>
13744 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13745 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13747 2017-09-21 Richard Biener <rguenther@suse.de>
13749 PR tree-optimization/71351
13750 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13751 graphite_create_new_loop_guard): Remove, fold remaining parts
13753 (translate_isl_ast_node_for): ... here and simplify.
13755 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13758 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13759 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13760 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13761 alternative always use QI mode, for -Os imov (=R,R) alternative
13762 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13765 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13766 Jeff Law <law@redhat.com>
13768 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13769 (allocate_stack_space): New function, partially extracted from
13770 s390_emit_prologue.
13771 (s390_emit_prologue): Track offset to most recent stack probe.
13772 Code to allocate space moved into allocate_stack_space.
13773 Dump actions when no stack is allocated.
13774 (s390_prologue_plus_offset): New function.
13775 (s390_emit_stack_probe): Likewise.
13777 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13779 * common.opt (Wa, Wl, Wp, g, gz=): Add
13781 (gno-column-info): Remove.
13782 (gcolumn-info): Drop RejectNegative.
13783 (gno-): New prefix.
13784 (gno-record-gcc-switches): Remove.
13785 (grecord-gcc-switches): Drop RejectNegative.
13786 (gno-split-dwarf): Remove.
13787 (gsplit-dwarf): Drop RejectNegative.
13788 (gno-strict-dwarf): Remove.
13789 (gstrict-dwarf): Drop RejectNegative.
13790 * config/darwin.opt (gfull, gused): Add RejectNegative.
13791 * dwarf2out.c (gen_producer_string): Drop
13792 gno-record-gcc-switches handler.
13793 * optc-gen.awk: Add g to prefixes with negative forms.
13794 * opts-common.c (remapping_prefix_p): New.
13795 (find_opt): Check it.
13796 (generate_canonical_option): Test g prefix.
13797 (option_map): Add -gno- mapping.
13798 (add_misspelling_candidates): Check remapping_prefix_p.
13800 2017-09-20 Jeff Law <law@redhat.com>
13802 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13803 thinko in stack clash protection support.
13805 * explow.c (compute_stack_clash_protection_loop_data): Use
13806 CONST_INT_P instead of explicit test. Verify object is a
13807 CONST_INT_P before looking at INTVAL.
13808 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13809 instead of explicit test.
13811 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13814 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13815 address instead of to r1 and r11.
13817 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13819 * config.gcc: Support "knm".
13820 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13821 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13823 * config/i386/i386.c (m_KNM): Define.
13824 (processor_target_table): Add "knm".
13826 (ix86_option_override_internal): Add "knm".
13827 (ix86_issue_rate): Add PROCESSOR_KNM.
13828 (ix86_adjust_cost): Ditto.
13829 (ia32_multipass_dfa_lookahead): Ditto.
13830 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13831 (fold_builtin_cpu): Add M_INTEL_KNM.
13832 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13833 (processor_type): Add PROCESSOR_KNM.
13834 * config/i386/x86-tune.def: Add m_KNM.
13835 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13837 2017-09-20 Richard Biener <rguenther@suse.de>
13839 PR tree-optimization/80213
13840 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13841 are allowed in empty BBs as well.
13842 (canonicalize_loop_closed_ssa): Also look for other complex
13844 (scop_detection::get_sese): Include the loop-closed PHI block
13846 (scop_detection::merge_sese): Remove code adding extra blocks.
13847 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13848 (build_scops): Assert the final returned scop is invalid.
13850 2017-09-20 Richard Biener <rguenther@suse.de>
13852 PR tree-optimization/82264
13853 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13855 (vn_phi_lookup): Likewise.
13856 (vn_phi_insert): Likewise.
13858 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13860 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13861 that fits into uhwi or shwi, add DW_AT_const_value regardless
13862 of early_dwarf without going through RTL, using add_AT_unsigned
13865 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13866 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13867 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13868 (DEBUG_MACRO_SECTION): Likewise.
13869 (DEBUG_LTO_MACRO_SECTION): Likewise.
13870 (DEBUG_STR_DWO_SECTION): Likewise.
13871 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13872 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13873 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13874 (DEBUG_LTO_LINE_STR_SECTION): Define.
13875 (init_sections_and_labels): Initialize debug_line_str_section
13876 variable. Initialize debug_loc_section for -gdwarf-5 to
13877 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13879 2017-09-20 Richard Biener <rguenther@suse.de>
13881 * graphite-sese-to-poly.c (extract_affine): Properly handle
13882 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13884 2017-09-20 Richard Biener <rguenther@suse.de>
13886 PR tree-optimization/81373
13887 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13888 Force SESE live-out defs to be handled even if they are
13891 2017-09-19 Jeff Law <law@redhat.com>
13893 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13894 nothing for stack adjustments with REG_STACK_CHECK.
13895 * sched-deps.c (parse_add_or_inc): Reject insns with
13896 REG_STACK_CHECK from dependency breaking.
13897 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13898 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13899 * reg-notes.def (STACK_CHECK): New note.
13901 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13902 (ix86_expand_prologue): Dump stack clash info as needed.
13903 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13905 * function.c (dump_stack_clash_frame_info): New function.
13906 * function.h (dump_stack_clash_frame_info): Prototype.
13907 (enum stack_clash_probes): New enum.
13909 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13910 flag_stack_clash_protection.
13911 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13912 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13913 (arm_frame_pointer_required): Likewise.
13914 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13915 (ia64_expand_prologue): Likewise.
13916 * config/mips/mips.c (mips_expand_prologue): Likewise.
13917 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13918 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13919 (sparc_flat_expand_prologue): Likewise.
13920 * config/spu/spu.c (spu_expand_prologue): Likewise.
13922 * explow.c: Include "params.h".
13923 (anti_adjust_stack_and_probe_stack_clash): New function.
13924 (get_stack_check_protect): Likewise.
13925 (compute_stack_clash_protection_loop_data): Likewise.
13926 (emit_stack_clash_protection_loop_start): Likewise.
13927 (emit_stack_clash_protection_loop_end): Likewise.
13928 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13929 Use anti_adjust_stack_and_probe_stack_clash.
13930 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13931 (emit_stack_clash_protection_loop_start): Likewise.
13932 (emit_stack_clash_protection_loop_end): Likewise.
13933 * rtl.h (get_stack_check_protect): Prototype.
13934 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13935 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13936 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13938 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13940 * doc/tm.texi: Rebuilt.
13941 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13942 get_stack_check_protect.
13943 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13944 * config/arm/arm.c (arm_expand_prologue): Likewise.
13945 (arm_frame_pointer_required): Likewise.
13946 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13947 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13948 * config/mips/mips.c (mips_expand_prologue): Likewise.
13949 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13950 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13951 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13952 (sparc_flat_expand_prologue): Likewise.
13954 * common.opt (-fstack-clash-protection): New option.
13955 * flag-types.h (enum stack_check_type): Note difference between
13956 -fstack-check= and -fstack-clash-protection.
13957 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13958 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13959 * toplev.c (process_options): Issue warnings/errors for cases
13960 not handled with -fstack-clash-protection.
13961 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13962 (-fstack-check): Note additional problem with -fstack-check=generic.
13963 Note that -fstack-check is primarily for Ada and refer users
13964 to -fstack-clash-protection for stack-clash-protection.
13965 Document new params for stack clash protection.
13967 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13969 * config/i386/i386.c (ix86_split_long_move): Do not handle
13970 address used for LEA in a special way.
13972 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13974 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13976 2017-09-19 Martin Sebor <msebor@redhat.com>
13979 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13980 of incompatible types.
13982 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13984 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13985 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13986 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13987 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13989 2017-09-19 Richard Biener <rguenther@suse.de>
13991 PR tree-optimization/82244
13992 * tree-vrp.c (remove_range_assertions): Do not propagate
13993 a constant to abnormals but replace the assert with a copy.
13995 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13997 PR rtl-optimization/57878
13998 PR rtl-optimization/68988
13999 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14000 avoidance test involving non_reload_pseudos. Move frequency test
14001 below the general fragmentation avoidance test.
14003 2017-09-19 Richard Biener <rguenther@suse.de>
14005 PR tree-optimization/69728
14006 * graphite-sese-to-poly.c (schedule_error): New global.
14007 (add_loop_schedule): Handle empty domain by failing the
14009 (build_original_schedule): Handle schedule_error.
14011 2017-09-19 Richard Biener <rguenther@suse.de>
14013 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14014 Do not iterate to sibling loops but only to siblings of inner
14017 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14020 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14021 that the registers are different.
14023 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14025 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14026 to processor_model and "amdfam17h" to arch_names_table.
14027 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14029 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14032 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14034 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14035 Alan Hayward <alan.hayward@arm.com>
14036 David Sherwood <david.sherwood@arm.com>
14038 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14040 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14042 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14043 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14044 vect_schedule_slp_instance.
14045 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14046 Update call to vect_slp_analyze_node_operations. Simplify return
14048 (vect_slp_analyze_bb_1): Update call accordingly.
14049 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14050 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14051 (vect_schedule_slp): Update call accordingly.
14053 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14054 Alan Hayward <alan.hayward@arm.com>
14055 David Sherwood <david.sherwood@arm.com>
14057 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14058 with types that aren't in fact scalar.
14060 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14062 * tree-vect-slp.c (vect_record_max_nunits): New function,
14064 (vect_build_slp_tree_1): ...here.
14065 (vect_build_slp_tree_2): Call it for phis too.
14067 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14069 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14070 to vect_get_vec_def_for_operand when getting the mask operand.
14072 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14073 Alan Hayward <alan.hayward@arm.com>
14074 David Sherwood <david.sherwood@arm.com>
14076 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14079 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14080 Alan Hayward <alan.hayward@arm.com>
14081 David Sherwood <david.sherwood@arm.com>
14083 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14084 calculation for vector booleans.
14086 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14087 Alan Hayward <alan.hayward@arm.com>
14088 David Sherwood <david.sherwood@arm.com>
14090 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14092 (vect_transform_stmt): ...here.
14093 (vect_analyze_stmt): Use it instead of calling
14094 vectorizable_live_operation directly.
14096 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14098 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14099 non-SIMT targets in acc vector loops.
14101 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14103 * configure.ac: Add arc and check if assembler supports gdwarf2.
14104 * configure: Regenerate.
14106 2017-09-18 Richard Biener <rguenther@suse.de>
14108 PR tree-optimization/82220
14109 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14110 epilogue niters from the min_profitable_iters compute.
14112 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14115 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14116 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14117 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14119 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14122 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14123 switching to a new text section.
14125 2017-09-18 Richard Biener <rguenther@suse.de>
14127 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14129 (build_alias_set): Reject aliases with no access function.
14131 2017-09-18 Richard Biener <rguenther@suse.de>
14133 PR tree-optimization/79622
14134 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14136 (build_cross_bb_scalars_use): Likewise.
14138 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14140 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14142 2017-09-18 Alan Modra <amodra@gmail.com>
14145 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14146 stack_pointer_rtx for count 0. Update comments. Break up
14147 large rtl expression.
14149 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14151 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14152 Increase to 20 bytes.
14153 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14154 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14155 or avx version of the stub.
14157 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14160 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14161 compute the minimum stack alignment. Also update preferred stack
14162 boundary for leaf functions.
14164 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14166 PR tree-optimization/82228
14167 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14170 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14172 * common/config/nds32/nds32-common.c
14173 (nds32_option_optimization_table): Refine formatting.
14174 (nds32_option_optimization_table): Use -fsched-pressure and
14175 -fomit-frame-pointer for specific optimization level.
14177 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14179 * config/nds32/nds32.c: Refine formatting and comments.
14180 * config/nds32/nds32.h: Likewise.
14181 * config/nds32/nds32.md: Likewise.
14182 * config/nds32/nds32-cost.c: Likewise.
14183 * config/nds32/nds32-isr.c: Likewise.
14184 * config/nds32/nds32-md-auxiliary.c: Likewise.
14185 * config/nds32/nds32-multiple.md: Likewise.
14186 * config/nds32/nds32-predicates.c: Likewise.
14188 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14189 Jakub Jelinek <jakub@redhat.com>
14191 Add support for -std=c++2a.
14192 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14194 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14196 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14199 * doc/extend.texi (Common Function Attributes): Add
14200 references to ARM, AArch64, and S/390 specific attributes.
14201 (Function Specific Option Pragmas): Add AArch64 and S/390
14202 to list of back ends that support the target pragma.
14204 2017-09-15 Nathan Sidwell <nathan@acm.org>
14206 * doc/standards.texi: Fix C++17 description. Update URLs for
14209 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14211 * common.opt (Wcast-align=strict): New warning option.
14212 * doc/invoke.texi: Document -Wcast-align=strict.
14214 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14216 * cgraph.h (cgraph_thunk_info): Add comments.
14217 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14218 assert for VIRTUAL_* arguments stricter.
14220 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14222 PR tree-optimization/71026
14223 * match.pd: Move RDIV patterns from fold-const.c
14224 * fold-const.c (distribute_real_division): Removed.
14225 (fold_binary_loc): Remove calls to distribute_real_divison.
14227 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14229 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14230 c++1z and gnu++1z as deprecated. Change other references to
14231 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14232 Change -Wc++1z-compat to -Wc++17-compat.
14233 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14234 * dwarf2out.c (highest_c_language): Handle C++17.
14235 (gen_compile_unit_die): Likewise.
14237 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14239 PR rtl-optimization/82192
14240 * combine.c (make_extraction): Don't look through non-paradoxical
14241 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14244 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14245 Alan Hayward <alan.hayward@arm.com>
14246 David Sherwood <david.sherwood@arm.com>
14248 * target.def (function_arg_offset): New hook.
14249 * targhooks.h (default_function_arg_offset): Declare.
14250 * targhooks.c (default_function_arg_offset): New function.
14251 * function.c (locate_and_pad_parm): Use
14252 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14253 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14254 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14255 * doc/tm.texi: Regenerate.
14256 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14257 * config/spu/spu.c (spu_function_arg_offset): New function.
14258 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14259 * system.h (FUNCTION_ARG_OFFSET): Poison.
14261 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14262 Alan Hayard <alan.hayward@arm.com>
14263 David Sherwood <david.sherwood@arm.com>
14265 * target.def (truly_noop_truncation): New hook.
14266 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14267 than TRULY_NOOP_TRUNCATION.
14268 * hooks.h (hook_bool_uint_uint_true): Declare.
14269 * hooks.c (hook_bool_uint_uint_true): New function.
14270 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14271 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14272 * doc/tm.texi: Regenerate.
14273 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14274 rather than TRULY_NOOP_TRUNCATION in comments.
14275 (simplify_comparison): Likewise.
14276 (record_truncated_value): Likewise.
14277 * expmed.c (extract_bit_field_1): Likewise.
14278 (extract_split_bit_field): Likewise.
14279 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14280 instead of TRULY_NOOP_TRUNCATION.
14281 * function.c (assign_parm_setup_block): Likewise.
14282 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14283 * rtlhooks.c: Include target.h.
14284 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14285 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14286 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14287 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14288 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14289 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14290 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14291 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14292 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14293 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14294 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14295 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14296 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14297 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14298 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14299 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14300 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14301 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14302 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14303 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14304 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14305 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14306 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14307 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14308 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14309 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14310 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14311 rather than TRULY_NOOP_TRUNCATION in comments.
14312 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14313 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14314 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14315 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14316 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14317 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14318 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14319 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14320 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14321 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14322 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14323 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14324 rather than TRULY_NOOP_TRUNCATION in comments.
14325 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14326 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14327 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14328 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14329 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14330 TRULY_NOOP_TRUNCATION condition.
14331 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14332 (TRULY_NOOP_TRUNCATION): Delete.
14333 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14334 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14335 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14336 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14337 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14338 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14339 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14340 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14341 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14342 rather than TRULY_NOOP_TRUNCATION in comments.
14343 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14344 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14345 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14346 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14347 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14348 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14350 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14353 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14354 (*cmp_ior): Likewise.
14355 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14356 (*ior_scc_scc_cmp): Likewise.
14357 (*and_scc_scc): Likewise.
14358 (*and_scc_scc_cmp): Likewise.
14360 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14361 Alan Hayard <alan.hayward@arm.com>
14362 David Sherwood <david.sherwood@arm.com>
14364 * target.def (can_change_mode_class): New hook.
14365 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14366 (hard_regno_nregs): Likewise.
14367 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14368 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14369 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14370 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14371 (LOAD_EXTEND_OP): Update accordingly.
14372 * doc/tm.texi: Regenerate.
14373 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14374 CANNOT_CHANGE_MODE_CLASS.
14375 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14376 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14377 * combine.c (simplify_set): Update accordingly.
14378 * emit-rtl.c (validate_subreg): Likewise.
14379 * recog.c (general_operand): Likewise.
14380 * regcprop.c (mode_change_ok): Likewise.
14381 * reload1.c (choose_reload_regs): Likewise.
14382 (inherit_piecemeal_p): Likewise.
14383 * rtlanal.c (simplify_subreg_regno): Likewise.
14384 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14385 instead of CANNOT_CHANGE_MODE_CLASS.
14386 (reload_cse_simplify_operands): Likewise.
14387 * reload.c (push_reload): Use targetm.can_change_mode_class
14388 instead of CANNOT_CHANGE_MODE_CLASS.
14389 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14390 REG_CANNOT_CHANGE_MODE_P.
14391 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14392 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14393 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14394 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14395 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14396 (arm_can_change_mode_class): New function.
14397 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14398 than CANNOT_CHANGE_MODE_CLASS in comments.
14399 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14400 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14401 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14402 (ix86_can_change_mode_class): ...this new function, inverting the
14403 sense of the return value.
14404 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14405 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14406 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14407 (ia64_can_change_mode_class): New function.
14408 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14409 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14410 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14411 (m32c_can_change_mode_class): ...this new function, inverting the
14412 sense of the return value.
14413 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14414 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14415 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14416 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14417 (mips_can_change_mode_class): ...this new function, inverting the
14418 sense of the return value.
14419 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14420 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14421 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14422 (msp430_can_change_mode_class): New function.
14423 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14424 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14425 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14426 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14427 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14428 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14429 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14430 (pa_cannot_change_mode_class): Replace with...
14431 (pa_can_change_mode_class): ...this new function, inverting the
14432 sense of the return value.
14433 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14434 than CANNOT_CHANGE_MODE_CLASS in comments.
14435 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14436 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14437 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14438 (pdp11_cannot_change_mode_class): Replace with...
14439 (pdp11_can_change_mode_class): ...this new function, inverting the
14440 sense of the return value.
14441 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14442 * config/powerpcspe/powerpcspe-protos.h
14443 (rs6000_cannot_change_mode_class_ptr): Delete.
14444 * config/powerpcspe/powerpcspe.c
14445 (rs6000_cannot_change_mode_class_ptr): Delete.
14446 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14447 (rs6000_option_override_internal): Assign to
14448 targetm.can_change_mode_class instead of
14449 rs6000_cannot_change_mode_class_ptr.
14450 (rs6000_cannot_change_mode_class): Replace with...
14451 (rs6000_can_change_mode_class): ...this new function, inverting the
14452 sense of the return value.
14453 (rs6000_debug_cannot_change_mode_class): Replace with...
14454 (rs6000_debug_can_change_mode_class): ...this new function.
14455 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14456 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14457 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14458 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14459 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14461 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14462 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14463 (rs6000_option_override_internal): Assign to
14464 targetm.can_change_mode_class instead of
14465 rs6000_cannot_change_mode_class_ptr.
14466 (rs6000_cannot_change_mode_class): Replace with...
14467 (rs6000_can_change_mode_class): ...this new function, inverting the
14468 sense of the return value.
14469 (rs6000_debug_cannot_change_mode_class): Replace with...
14470 (rs6000_debug_can_change_mode_class): ...this new function.
14471 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14472 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14473 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14474 (s390_can_change_mode_class): ...this new function, inverting the
14475 sense of the return value.
14476 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14477 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14478 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14479 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14480 (sh_cannot_change_mode_class): Replace with...
14481 (sh_can_change_mode_class): ...this new function, inverting the
14482 sense of the return value.
14483 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14484 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14485 (sparc_can_change_mode_class): New function.
14486 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14487 * config/spu/spu.c (spu_can_change_mode_class): New function.
14488 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14489 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14490 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14491 (visium_can_change_mode_class): New function.
14492 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14494 2017-09-15 Richard Biener <rguenther@suse.de>
14496 PR tree-optimization/82217
14497 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14498 but not undefined case.
14500 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14503 * postreload.c (reload_cse_simplify_operands): Skip
14504 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14506 2017-09-15 Richard Biener <rguenther@suse.de>
14508 PR tree-optimization/68823
14509 * graphite-scop-detection.c (build_alias_set): If we have a
14510 possible dependence check whether we can handle them by just
14511 looking at the DRs DR_ACCESS_FNs.
14512 (build_scops): If build_alias_set fails, fail the SCOP.
14514 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14516 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14517 to support float128 built-in functions that require the ISA 3.0
14519 (BU_FLOAT128_3_HW): Likewise.
14520 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14521 built-in functions.
14522 (FMAF128): Likewise.
14524 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14525 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14526 floating point instructions.
14527 (rs6000_invalid_builtin): Likewise.
14528 (rs6000_builtin_mask_names): Likewise.
14529 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14530 (RS6000_BTM_FLOAT128_HW): Likewise.
14531 (RS6000_BTM_COMMON): Likewise.
14532 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14534 * doc/extend.texi (RS/6000 built-in functions): Document the
14535 IEEE 128-bit floating point square root and fused multiply-add
14536 built-in functions.
14538 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14540 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14541 reg (r2) isn't in the set of registers defined in the prologue.
14543 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14544 Alan Hayward <alan.hayward@arm.com>
14545 David Sherwood <david.sherwood@arm.com>
14547 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14548 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14549 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14550 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14551 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14553 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14554 max_vectorization_factor.
14555 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14557 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14558 Alan Hayward <alan.hayward@arm.com>
14559 David Sherwood <david.sherwood@arm.com>
14561 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14562 (vect_worthwhile_without_simd_p): Declare.
14563 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14564 (vectorizable_reduction): Use it.
14565 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14566 (vectorizable_operation): Likewise.
14568 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14569 Alan Hayward <alan.hayward@arm.com>
14570 David Sherwood <david.sherwood@arm.com>
14572 * tree-vectorizer.h (vect_get_num_copies): New function.
14573 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14574 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14575 (vectorizable_induction): Likewise.
14576 (vectorizable_live_operation): Likewise.
14577 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14578 (vectorizable_bswap): Likewise.
14579 (vectorizable_call): Likewise.
14580 (vectorizable_conversion): Likewise.
14581 (vectorizable_assignment): Likewise.
14582 (vectorizable_shift): Likewise.
14583 (vectorizable_operation): Likewise.
14584 (vectorizable_store): Likewise.
14585 (vectorizable_load): Likewise.
14586 (vectorizable_condition): Likewise.
14587 (vectorizable_comparison): Likewise.
14588 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14590 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14591 Alan Hayward <alan.hayward@arm.com>
14592 David Sherwood <david.sherwood@arm.com>
14594 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14595 of vect_init_vector.
14597 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14598 Alan Hayward <alan.hayward@arm.com>
14599 David Sherwood <david.sherwood@arm.com>
14601 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14602 an inline wrapper that provides a location.
14603 (gimple_build_vector): Likewise.
14604 * gimple-fold.c (gimple_build_vector_from_val): New function.
14605 (gimple_build_vector): Likewise.
14606 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14607 functions to build the initial value. Always return a gimple value.
14608 (get_initial_defs_for_reduction): Likewise. Only compute
14610 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14611 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14612 (vectorizable_induction): Use gimple_build_vector rather than
14615 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14616 Alan Hayward <alan.hayward@arm.com>
14617 David Sherwood <david.sherwood@arm.com>
14619 * target.h (vec_perm_indices): New typedef.
14620 (auto_vec_perm_indices): Likewise.
14621 * optabs-query.h: Include target.h
14622 (can_vec_perm_p): Take a vec_perm_indices *.
14623 * optabs-query.c (can_vec_perm_p): Likewise.
14624 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14625 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14626 * tree-vect-generic.c (lower_vec_perm): Likewise.
14627 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14628 (vect_grouped_load_supported): Likewise.
14629 (vect_shift_permute_load_chain): Likewise.
14630 (vect_permute_store_chain): Use auto_vec_perm_indices.
14631 (vect_permute_load_chain): Likewise.
14632 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14633 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14634 Update uses of can_vec_perm_p.
14635 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14636 mode with a number of elements. Take a vec_perm_indices *.
14637 (vect_create_epilog_for_reduction): Update accordingly.
14638 Use auto_vec_perm_indices.
14639 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14640 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14641 (vect_transform_slp_perm_load): Likewise.
14642 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14643 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14644 (vect_gen_perm_mask_checked): Likewise.
14645 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14646 (vect_gen_perm_mask_checked): Likewise.
14647 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14648 (vectorizable_store): Likewise.
14649 (vectorizable_load): Likewise.
14650 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14651 (vectorizable_bswap): Likewise.
14653 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14654 Alan Hayward <alan.hayward@arm.com>
14655 David Sherwood <david.sherwood@arm.com>
14657 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14658 * tree.c (build_vector): Likewise.
14659 (build_vector_from_ctor): Update accordingly.
14660 (build_vector_from_val): Likewise.
14661 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14662 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14663 * tree-vect-generic.c (add_rshift): Likewise.
14664 (expand_vector_divmod): Likewise.
14665 (optimize_vector_constructor): Likewise.
14666 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14667 (vect_transform_slp_perm_load): Likewise.
14668 (vect_schedule_slp_instance): Likewise.
14669 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14670 (vectorizable_call): Likewise.
14671 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14672 * expmed.c (make_tree): Likewise.
14673 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14674 a vector passed to build_vector.
14675 (fold_convert_const): Likewise.
14676 (exact_inverse): Likewise.
14677 (fold_ternary_loc): Likewise.
14678 (fold_relational_const): Likewise.
14679 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14680 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14681 (const_unop): Likewise. Store the reduction accumulator in a
14682 variable rather than an array.
14683 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14684 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14685 the new vector, rather than constructing it after the input arrays.
14686 (native_interpret_vector): Use auto_vec<tree> when building
14687 a vector passed to build_vector. Add elements in order.
14688 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14689 auto_vec<tree> when building a vector passed to build_vector.
14690 (vect_create_epilog_for_reduction): Likewise.
14691 (vectorizable_induction): Likewise.
14692 (get_initial_def_for_reduction): Likewise. Fix indentation of
14694 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14696 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14697 passed to build_vector.
14699 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14700 Alan Hayward <alan.hayward@arm.com>
14701 David Sherwood <david.sherwood@arm.com>
14703 * tree-core.h (tree_base::u): Add an "nelts" field.
14704 (tree_vector): Use VECTOR_CST_NELTS as the length.
14705 * tree.c (tree_size): Likewise.
14706 (make_vector): Initialize VECTOR_CST_NELTS.
14707 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14708 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14709 TYPE_VECTOR_SUBPARTS.
14710 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14711 as the number of units, setting it from VECTOR_CST_NELTS.
14712 (const_vector_from_tree): Likewise.
14713 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14714 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14715 (fold_negate_expr_1): Likewise.
14716 (fold_convert_const): Likewise.
14717 (const_binop): Likewise. Differentiate the number of output and
14719 (const_unop): Likewise.
14720 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14721 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14722 in cases that did the opposite.
14724 2017-09-14 Richard Biener <rguenther@suse.de>
14726 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14729 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14731 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14733 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14736 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14737 if and where to split a bb, except for splitting before debug insn
14738 sequences followed by non-label real insn. Delete debug insns
14739 in between basic blocks.
14741 * combine.c (make_compound_operation_int): Formatting fixes.
14743 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14744 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14745 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14746 * config/sol2.h (LINK_EH_SPEC): Likewise.
14747 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14748 * config/s390/linux.h (LINK_SPEC): Likewise.
14749 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14750 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14751 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14752 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14753 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14754 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14755 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14756 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14758 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14760 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14762 (ENDFILE_LINUX_SPEC): Likewise.
14763 (LINK_EH_SPEC): Likewise.
14764 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14765 (LINK_OS_LINUX_SPEC32): Likewise.
14766 (LINK_OS_LINUX_SPEC64): Likewise.
14767 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14768 (LINK_OS_LINUX_SPEC): Likewise.
14770 2017-09-13 Martin Liska <mliska@suse.cz>
14772 PR middle-end/82154
14773 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14774 CASE_HIGH is NULL_TREE.
14776 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14777 Alan Hayward <alan.hayward@arm.com>
14778 David Sherwood <david.sherwood@arm.com>
14780 * target.def (secondary_memory_needed): New hook.
14781 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14782 instead of SECONDARY_MEMORY_NEEDED.
14783 (secondary_memory_needed_mode): Likewise.
14784 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14785 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14786 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14787 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14788 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14789 * doc/tm.texi: Regenerate.
14790 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14791 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14792 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14793 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14794 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14795 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14796 mode argument first and change the reg_class arguments to reg_class_t.
14797 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14798 Make static. Update the call to inline_secondary_memory_needed.
14799 (ix86_register_move_cost): Update the call to
14800 inline_secondary_memory_needed.
14801 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14802 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14804 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14805 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14807 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14808 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14809 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14810 and match hook interface. Add comment from mips.h.
14811 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14812 * config/mmix/mmix.md (truncdfsf2): Refer to
14813 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14815 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14816 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14817 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14818 (pa_secondary_memory_needed): New function.
14819 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14820 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14821 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14822 (pdp11_secondary_memory_needed): Make static and match hook interface.
14823 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14824 * config/powerpcspe/powerpcspe-protos.h
14825 (rs6000_secondary_memory_needed_ptr): Delete.
14826 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14828 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14829 (rs6000_option_override_internal): Assign to
14830 targetm.secondary_memory_needed rather than
14831 rs6000_secondary_memory_needed_ptr.
14832 (rs6000_secondary_memory_needed): Match hook interface.
14833 (rs6000_debug_secondary_memory_needed): Likewise.
14834 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14835 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14836 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14837 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14838 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14839 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14841 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14842 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14843 (rs6000_option_override_internal): Assign to
14844 targetm.secondary_memory_needed rather than
14845 rs6000_secondary_memory_needed_ptr.
14846 (rs6000_secondary_memory_needed): Match hook interface.
14847 (rs6000_debug_secondary_memory_needed): Likewise.
14848 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14849 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14850 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14851 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14852 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14853 (sparc_secondary_memory_needed): New function.
14854 * lra-constraints.c (check_and_process_move): Refer to
14855 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14857 (curr_insn_transform): Likewise.
14858 (process_alt_operands): Use targetm.secondary_memory_needed
14859 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14860 (check_secondary_memory_needed_p): Likewise.
14861 (choose_split_class): Likewise.
14862 * reload.c: Unconditionally include code that was previously
14863 conditional on SECONDARY_MEMORY_NEEDED.
14864 (push_secondary_reload): Use targetm.secondary_memory_needed
14865 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14866 (push_reload): Likewise.
14867 * reload1.c: Unconditionally include code that was previously
14868 conditional on SECONDARY_MEMORY_NEEDED.
14869 (choose_reload_regs): Use targetm.secondary_memory_needed
14870 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14871 (gen_reload): Likewise.
14872 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14874 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14875 Alan Hayward <alan.hayward@arm.com>
14876 David Sherwood <david.sherwood@arm.com>
14878 * target.def (secondary_memory_needed_mode): New hook:
14879 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14880 * targhooks.h (default_secondary_memory_needed_mode): New function.
14881 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14882 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14883 * doc/tm.texi: Regenerate.
14884 * lra-constraints.c (check_and_process_move): Use
14885 targetm.secondary_memory_needed_mode instead of
14886 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14887 (curr_insn_transform): Likewise.
14888 * reload.c (get_secondary_mem): Likewise.
14889 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14890 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14892 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14893 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14894 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14895 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14896 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14898 * config/powerpcspe/powerpcspe-protos.h
14899 (rs6000_secondary_memory_needed_mode): Delete.
14900 * config/powerpcspe/powerpcspe.c
14901 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14902 (rs6000_secondary_memory_needed_mode): Make static.
14903 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14904 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14906 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14908 (rs6000_secondary_memory_needed_mode): Make static.
14909 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14910 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14911 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14912 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14913 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14915 (sparc_secondary_memory_needed_mode): New function.
14916 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14918 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14920 * config/aarch64/constraints.md (Umq): New constraint.
14921 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14923 (mov<mode>): Update condition.
14925 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14927 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14928 when bitposition is the same.
14930 2017-09-13 Richard Biener <rguenther@suse.de>
14932 * dwarf2out.c (output_die_symbol): Remove.
14933 (output_die): Do not output a DIEs symbol.
14935 2017-09-13 Richard Biener <rguenther@suse.de>
14937 PR middle-end/82128
14938 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14939 default-def to avoid breaking iterator update with the weird
14940 interaction with cgraph_update_edges_for_call_stmt_node.
14942 2017-09-13 Richard Biener <rguenther@suse.de>
14944 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14945 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14946 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14947 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14949 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14951 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14952 Disable pc relative literal load irrespective of
14953 TARGET_FIX_ERR_A53_84341 for default.
14955 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14957 * config/sparc/sparc.c (output_return): Output the source location of
14958 the insn in the delay slot, if any.
14959 (output_sibcall): Likewise.
14961 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14964 * common.opt (-static-pie): New alias.
14965 (shared): Negate static-pie.
14966 (-no-pie): Update help text.
14968 (static-pie): New option.
14969 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14970 -static-pie support.
14971 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14972 (LINK_EH_SPEC): Likewise.
14973 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14974 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14975 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14976 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14977 (init_gcc_specs): Likewise.
14978 (init_spec): Likewise.
14979 (display_help): Update help message for -pie.
14980 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14983 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14985 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14986 (movdi_aarch64): Likewise.
14987 (movti_aarch64): Likewise.
14989 2017-09-12 Simon Wright <simon@pushface.org>
14992 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14993 calculation of the minor version, always output as 0.
14995 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14998 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14999 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15000 on it early, rather than manual conversion late. For
15001 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15002 instead of performing manual conversion.
15004 2017-09-12 Carl Love <cel@us.ibm.com>
15006 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15007 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15009 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15010 VMULOSW): Add definitions.
15011 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15012 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15013 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15014 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15015 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15017 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15019 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15021 (movti_aarch64): Likewise.
15022 (movdf_aarch64): Likewise.
15023 (movtf_aarch64): Likewise.
15024 (load_pairdi): Likewise.
15025 (store_pairdi): Likewise.
15026 (load_pairdf): Likewise.
15027 (store_pairdf): Likewise.
15028 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15029 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15030 (ldr_got_small_<mode>): Likewise.
15031 (ldr_got_small_28k_<mode>): Likewise.
15032 (ldr_got_tiny): Likewise.
15033 * config/aarch64/iterators.md (ldst_sz): New.
15034 (ldpstp_sz): Likewise.
15035 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15037 (thunderx_load): Split load_8 to load_16.
15038 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15040 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15041 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15042 (xgene1_store_pair): Split store_8 to store_16.
15043 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15044 (falkor_st_0_st_sd): Split store_8 to store_16.
15046 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15048 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15049 and store1/2/3/4 to store_4/8/12/16.
15050 * config/aarch64/aarch64.md: Update for rename.
15051 * config/arm/arm.md: Likewise.: Likewise.
15052 * config/arm/arm.c: Likewise.
15053 * config/arm/thumb1.md: Likewise.
15054 * config/arm/thumb2.md: Likewise.
15055 * config/arm/vfp.md: Likewise.
15056 * config/arm/arm-generic.md: Likewise.
15057 * config/arm/arm1020e.md: Likewise.
15058 * config/arm/arm1026ejs.md: Likewise.
15059 * config/arm/arm1136jfs.md: Likewise.
15060 * config/arm/arm926ejs.md: Likewise.
15061 * config/arm/cortex-a15.md: Likewise.
15062 * config/arm/cortex-a17.md: Likewise.
15063 * config/arm/cortex-a5.md: Likewise.
15064 * config/arm/cortex-a53.md: Likewise.
15065 * config/arm/cortex-a57.md: Likewise.
15066 * config/arm/cortex-a7.md: Likewise.
15067 * config/arm/cortex-a8.md: Likewise.
15068 * config/arm/cortex-a9.md: Likewise.
15069 * config/arm/cortex-m4.md: Likewise.
15070 * config/arm/cortex-m7.md: Likewise.
15071 * config/arm/cortex-r4.md: Likewise.
15072 * config/arm/exynos-m1.md: Likewise.
15073 * config/arm/fa526.md: Likewise.
15074 * config/arm/fa606te.md: Likewise.
15075 * config/arm/fa626te.md: Likewise.
15076 * config/arm/fa726te.md: Likewise.
15077 * config/arm/fmp626.md: Likewise.
15078 * config/arm/iwmmxt.md: Likewise.
15079 * config/arm/ldmstm.md: Likewise.
15080 * config/arm/marvell-pj4.md: Likewise.
15081 * config/arm/xgene1.md: Likewise.
15082 * config/aarch64/thunderx.md: Likewise.
15083 * config/aarch64/thunderx2t99.md: Likewise.
15084 * config/aarch64/falkor.md: Likewise.
15086 2017-09-12 Martin Liska <mliska@suse.cz>
15088 * attribs.c (private_lookup_attribute): New function.
15089 * attribs.h (private_lookup_attribute): Declared here.
15090 (lookup_attribute): Called from this place.
15092 2017-09-12 Richard Biener <rguenther@suse.de>
15094 PR tree-optimization/82157
15095 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15096 stmts with side-effects.
15098 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15099 Alan Hayward <alan.hayward@arm.com>
15100 David Sherwood <david.sherwood@arm.com>
15102 * target.def (hard_regno_nregs): New hook.
15103 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15104 * targhooks.h (default_hard_regno_nregs): Declare.
15105 * targhooks.c (default_hard_regno_nregs): New function.
15106 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15107 (TARGET_HARD_REGNO_NREGS): ...this hook.
15108 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15109 (CLASS_MAX_NREGS): Likewise.
15110 * doc/tm.texi: Regenerate.
15111 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15112 instead of HARD_REGNO_NREGS.
15113 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15114 HARD_REGNO_NREGS in the comment.
15115 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15116 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15117 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15118 Return an unsigned int.
15119 (TARGET_HARD_REGNO_NREGS): Redefine.
15120 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15121 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15122 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15123 (arc_hard_regno_nregs): New function.
15124 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15125 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15126 (arm_hard_regno_nregs): New function.
15127 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15128 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15129 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15130 (TARGET_HARD_REGNO_NREGS): Redefine.
15131 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15132 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15133 (HARD_REGNO_NREGS): Delete.
15134 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15135 (cr16_hard_regno_nregs): New function.
15136 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15137 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15138 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15139 (cris_hard_regno_nregs): New function.
15140 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15141 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15142 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15143 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15144 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15145 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15146 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15147 (frv_hard_regno_nregs): Make static. Take and return an
15149 (frv_class_max_nregs): Remove outdated copy of documentation.
15150 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15151 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15152 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15153 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15154 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15155 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15156 (TARGET_HARD_REGNO_NREGS): Redefine.
15157 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15158 (CLASS_MAX_NREGS): Update comment.
15159 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15160 (ia64_hard_regno_nregs): New function.
15161 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15162 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15163 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15164 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15165 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15167 (m32c_hard_regno_nregs): Likewise. Make static.
15168 (TARGET_HARD_REGNO_NREGS): Redefine.
15169 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15170 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15171 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15172 (m68k_hard_regno_nregs): New function.
15173 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15174 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15175 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15176 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15177 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15178 Take and return an unsigned int.
15179 (TARGET_HARD_REGNO_NREGS): Redefine.
15180 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15181 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15182 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15183 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15184 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15185 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15186 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15187 (msp430_hard_regno_nregs): Make static. Take and return an
15189 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15190 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15191 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15192 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15193 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15194 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15195 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15196 (TARGET_HARD_REGNO_NREGS): Redefine.
15197 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15198 (PA_HARD_REGNO_NREGS): ...this.
15199 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15200 (PA_HARD_REGNO_NREGS): ...this.
15201 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15202 (pa_hard_regno_nregs): New function.
15203 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15204 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15205 (pdp11_hard_regno_nregs): New function.
15206 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15207 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15208 (rs6000_hard_regno_nregs_hook): New function.
15209 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15210 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15211 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15212 Take and return an unsigned int. Move earlier in file.
15213 (TARGET_HARD_REGNO_NREGS): Redefine.
15214 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15215 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15216 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15217 (rl78_hard_regno_nregs): Make static. Take and return an
15219 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15220 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15221 (rs6000_hard_regno_nregs_hook): New function.
15222 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15223 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15224 (TARGET_HARD_REGNO_NREGS): Redefine.
15225 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15226 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15227 instead of HARD_REGNO_NREGS.
15228 (s390_hard_regno_nregs): New function.
15229 (s390_hard_regno_mode_ok): Add comment from s390.h.
15230 (TARGET_HARD_REGNO_NREGS): Redefine.
15231 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15232 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15233 (sh_hard_regno_nregs): New function.
15234 (sh_pass_in_reg_p): Use it.
15235 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15236 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15237 (sparc_hard_regno_nregs): New function.
15238 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15239 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15240 (spu_function_arg_advance): Use it, supplying a valid register number.
15241 (TARGET_HARD_REGNO_NREGS): Redefine.
15242 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15243 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15244 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15245 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15246 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15247 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15248 (CLASS_MAX_NREGS): Remove copy of old documentation.
15249 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15250 (visium_hard_regno_nregs): New function.
15251 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15252 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15253 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15254 xtensa_hard_regno_nregs): New function.
15255 * system.h (HARD_REGNO_NREGS): Poison.
15257 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15259 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15260 hard_regno_nregs instead of HARD_REGNO_NREGS.
15261 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15262 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15263 (c6x_expand_epilogue): Likewise.
15264 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15265 (frv_read_iacc_argument): Likewise.
15266 * config/sh/sh.c: Include regs.h.
15267 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15268 (regs_used): Likewise.
15269 (output_stack_adjust): Likewise.
15270 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15271 * expmed.c: Include regs.h.
15272 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15273 * ree.c: Include regs.h.
15274 (combine_reaching_defs): Use hard_regno_nregs instead of
15276 (add_removable_extension): Likewise.
15278 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15280 * regs.h (hard_regno_nregs): Turn into a function.
15281 (end_hard_regno): Update accordingly.
15282 * caller-save.c (setup_save_areas): Likewise.
15283 (save_call_clobbered_regs): Likewise.
15284 (replace_reg_with_saved_mem): Likewise.
15285 (insert_restore): Likewise.
15286 (insert_save): Likewise.
15287 * combine.c (can_change_dest_mode): Likewise.
15288 (move_deaths): Likewise.
15289 (distribute_notes): Likewise.
15290 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15291 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15292 (rs6000_split_multireg_move): Likewise.
15293 (rs6000_register_move_cost): Likewise.
15294 (rs6000_memory_move_cost): Likewise.
15295 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15296 (rs6000_split_multireg_move): Likewise.
15297 (rs6000_register_move_cost): Likewise.
15298 (rs6000_memory_move_cost): Likewise.
15299 * cselib.c (cselib_reset_table): Likewise.
15300 (cselib_lookup_1): Likewise.
15301 * emit-rtl.c (set_mode_and_regno): Likewise.
15302 * function.c (aggregate_value_p): Likewise.
15303 * ira-color.c (setup_profitable_hard_regs): Likewise.
15304 (check_hard_reg_p): Likewise.
15305 (calculate_saved_nregs): Likewise.
15306 (assign_hard_reg): Likewise.
15307 (improve_allocation): Likewise.
15308 (calculate_spill_cost): Likewise.
15309 * ira-emit.c (modify_move_list): Likewise.
15310 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15311 (ira_hard_reg_in_set_p): Likewise.
15312 * ira.c (setup_reg_mode_hard_regset): Likewise.
15313 (clarify_prohibited_class_mode_regs): Likewise.
15314 (check_allocation): Likewise.
15315 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15316 (lra_setup_reg_renumber): Likewise.
15317 (setup_try_hard_regno_pseudos): Likewise.
15318 (spill_for): Likewise.
15319 (assign_hard_regno): Likewise.
15320 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15321 * lra-constraints.c (in_class_p): Likewise.
15322 (lra_constraint_offset): Likewise.
15323 (simplify_operand_subreg): Likewise.
15324 (lra_constraints): Likewise.
15325 (split_reg): Likewise.
15326 (split_if_necessary): Likewise.
15327 (invariant_p): Likewise.
15328 (inherit_in_ebb): Likewise.
15329 * lra-lives.c (process_bb_lives): Likewise.
15330 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15331 (get_hard_regs): Likewise.
15332 (do_remat): Likewise.
15333 * lra-spills.c (assign_spill_hard_regs): Likewise.
15334 * mode-switching.c (create_pre_exit): Likewise.
15335 * postreload.c (reload_combine_recognize_pattern): Likewise.
15336 * recog.c (peep2_find_free_register): Likewise.
15337 * regcprop.c (kill_value_regno): Likewise.
15338 (set_value_regno): Likewise.
15339 (copy_value): Likewise.
15340 (maybe_mode_change): Likewise.
15341 (find_oldest_value_reg): Likewise.
15342 (copyprop_hardreg_forward_1): Likewise.
15343 * regrename.c (check_new_reg_p): Likewise.
15344 (regrename_do_replace): Likewise.
15345 * reload.c (push_reload): Likewise.
15346 (combine_reloads): Likewise.
15347 (find_dummy_reload): Likewise.
15348 (operands_match_p): Likewise.
15349 (find_reloads): Likewise.
15350 (find_equiv_reg): Likewise.
15351 (reload_adjust_reg_for_mode): Likewise.
15352 * reload1.c (count_pseudo): Likewise.
15353 (count_spilled_pseudo): Likewise.
15354 (find_reg): Likewise.
15355 (clear_reload_reg_in_use): Likewise.
15356 (free_for_value_p): Likewise.
15357 (allocate_reload_reg): Likewise.
15358 (choose_reload_regs): Likewise.
15359 (reload_adjust_reg_for_temp): Likewise.
15360 (emit_reload_insns): Likewise.
15361 (delete_output_reload): Likewise.
15362 * rtlanal.c (subreg_get_info): Likewise.
15363 * sched-deps.c (sched_analyze_reg): Likewise.
15364 * sel-sched.c (init_regs_for_mode): Likewise.
15365 (mark_unavailable_hard_regs): Likewise.
15366 (choose_best_reg_1): Likewise.
15367 (verify_target_availability): Likewise.
15368 * valtrack.c (dead_debug_insert_temp): Likewise.
15369 * var-tracking.c (track_loc_p): Likewise.
15370 (emit_note_insn_var_location): Likewise.
15371 * varasm.c (make_decl_rtl): Likewise.
15372 * reginfo.c (choose_hard_reg_mode): Likewise.
15373 (init_reg_modes_target): Refer directly to
15374 this_target_regs->x_hard_regno_nregs.
15376 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15378 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15379 instead of hard_regno_nregs.
15381 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15383 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15384 end_hard_regno instead of hard_regno_nregs.
15385 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15386 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15387 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15388 * ira-color.c (improve_allocation): Likewise.
15389 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15390 * lra-lives.c (mark_regno_live): Likewise.
15391 (mark_regno_dead): Likewise.
15392 * lra-remat.c (operand_to_remat): Likewise.
15393 * lra.c (collect_non_operand_hard_regs): Likewise.
15394 * postreload.c (reload_combine_note_store): Likewise.
15395 (move2add_valid_value_p): Likewise.
15396 * reload.c (regno_clobbered_p): Likewise.
15398 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15400 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15402 * config/v850/v850.c (v850_reorg): Likewise.
15403 * reload.c (refers_to_regno_for_reload_p): Likewise.
15404 (find_equiv_reg): Likewise.
15405 * reload1.c (reload_reg_reaches_end_p): Likewise.
15407 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15409 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15411 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15412 * config/arm/arm.c (output_move_neon): Likewise.
15413 (arm_attr_length_move_neon): Likewise.
15414 (neon_split_vcombine): Likewise.
15415 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15416 (c6x_mark_reg_written): Likewise.
15417 (c6x_dwarf_register_span): Likewise.
15418 * config/i386/i386.c (ix86_save_reg): Likewise.
15419 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15420 (rws_access_reg): Likewise.
15421 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15422 * mode-switching.c (create_pre_exit): Likewise.
15423 * ree.c (combine_reaching_defs): Likewise.
15424 (add_removable_extension): Likewise.
15425 * regcprop.c (find_oldest_value_reg): Likewise.
15426 (copyprop_hardreg_forward_1): Likewise.
15427 * reload.c (reload_inner_reg_of_subreg): Likewise.
15428 (push_reload): Likewise.
15429 (combine_reloads): Likewise.
15430 (find_dummy_reload): Likewise.
15431 (reload_adjust_reg_for_mode): Likewise.
15432 * reload1.c (find_reload_regs): Likewise.
15433 (forget_old_reloads_1): Likewise.
15434 (reload_reg_free_for_value_p): Likewise.
15435 (reload_adjust_reg_for_temp): Likewise.
15436 (emit_reload_insns): Likewise.
15437 (delete_output_reload): Likewise.
15438 * sel-sched.c (choose_best_reg_1): Likewise.
15439 (choose_best_pseudo_reg): Likewise.
15441 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15442 Alan Hayward <alan.hayward@arm.com>
15443 David Sherwood <david.sherwood@arm.com>
15445 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15446 * target.def (slow_unaligned_access): New hook.
15447 * targhooks.h (default_slow_unaligned_access): Declare.
15448 * targhooks.c (default_slow_unaligned_access): New function.
15449 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15450 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15451 * doc/tm.texi: Regenerate.
15452 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15453 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15454 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15456 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15457 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15459 (rs6000_slow_unaligned_access): New function.
15460 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15461 (expand_block_compare): Likewise.
15462 (expand_strn_compare): Likewise.
15463 (rs6000_rtx_costs): Likewise.
15464 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15465 (riscv_slow_unaligned_access): Likewise.
15466 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15467 (riscv_slow_unaligned_access_p): ...this and make static.
15468 (riscv_option_override): Update accordingly.
15469 (riscv_slow_unaligned_access): New function.
15470 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15471 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15472 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15473 (rs6000_slow_unaligned_access): New function.
15474 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15475 (rs6000_rtx_costs): Likewise.
15476 * config/rs6000/rs6000-string.c (expand_block_compare)
15477 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15478 of SLOW_UNALIGNED_ACCESS.
15479 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15480 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15481 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15482 of SLOW_UNALIGNED_ACCESS.
15483 * expmed.c (simple_mem_bitfield_p): Likewise.
15484 * expr.c (alignment_for_piecewise_move): Likewise.
15485 (emit_group_load_1): Likewise.
15486 (emit_group_store): Likewise.
15487 (copy_blkmode_from_reg): Likewise.
15488 (emit_push_insn): Likewise.
15489 (expand_assignment): Likewise.
15490 (store_field): Likewise.
15491 (expand_expr_real_1): Likewise.
15492 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15493 * lra-constraints.c (simplify_operand_subreg): Likewise.
15494 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15495 * gimple-ssa-store-merging.c: Likewise in block comment at start
15497 * tree-ssa-strlen.c: Include target.h.
15498 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15499 of SLOW_UNALIGNED_ACCESS.
15500 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15502 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15504 PR rtl-optimization/82185
15505 * expmed.c (emit_store_flag_int): Only test tem if it has been
15508 2017-09-12 Richard Biener <rguenther@suse.de>
15510 PR middle-end/82149
15511 * match.pd ((FTYPE) N CMP CST): Fix typo.
15513 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15515 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15517 (mips_near_type_p): Add 'short_call' attribute as a synonym
15519 * doc/extend.texi (short_call): Document new function attribute.
15521 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15524 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15525 assertion check that in the condition.
15526 (get_atomic_generic_size): Likewise. Before testing if parameter
15527 has pointer type, if it has array type, call for C++
15528 default_conversion to perform array-to-pointer conversion.
15530 2017-09-12 Richard Biener <rguenther@suse.de>
15532 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15533 for operations we cannot scalarize.
15535 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15537 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15538 vectors heap vectors. Clean up comments.
15539 Make visited_bbs a reference.
15540 (profitable_jump_thread_path): Make GC
15541 vectors heap vectors. Clean up comments.
15543 (convert_and_register_jump_thread_path): Make GC vectors heap
15545 (check_subpath_and_update_thread_path): Same. Clean up comments.
15546 Make visited_bbs a reference.
15547 (handle_phi): Abstract common code to to
15548 register_jump_thread_path_if_profitable.
15549 Rename VAR_BB to DEF_BB.
15551 Make GC vectors heap vectors.
15552 Make visited_bbs a reference.
15553 (handle_assignment): Same.
15554 (register_jump_thread_path_if_profitable): New.
15555 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15557 Make GC vectors heap vectors. Clean up comments.
15558 Make visited_bbs a reference.
15559 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15560 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15563 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15566 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15567 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15569 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15571 Revert r251800 and r251799.
15573 2017-09-11 Martin Jambor <mjambor@suse.cz>
15576 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15577 arguments in advance.
15578 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15579 use it to find predecessor edges.
15580 (naive_outof_ssa): Collect vector of predecessors.
15582 2017-09-08 Jason Merrill <jason@redhat.com>
15584 PR c++/70029 - ICE with ref-qualifier and -flto
15585 * langhooks.h (struct lang_hooks_for_types): Add
15586 copy_lang_qualifiers.
15587 * attribs.c (build_type_attribute_qual_variant): Use it.
15588 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15590 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15591 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15593 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15596 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15597 (*mulsi3_sp64): New instruction.
15598 (mulsi3): New expander.
15600 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15602 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15604 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15606 * sancov.c: Include memmodel.h.
15608 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15611 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15614 2017-09-07 Carl Love <cel@us.ibm.com>
15616 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15617 the sldi instruction.
15619 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15621 * sancov.c: Include tm_p.h.
15623 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15626 * output.h (switch_to_other_text_partition): New declaration.
15627 * varasm.c (switch_to_other_text_partition): New function.
15628 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15629 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15630 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15631 to the other text partition before emitting LCL label and switch back
15632 after emitting the word after it.
15634 2017-09-07 Richard Biener <rguenther@suse.de>
15636 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15637 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15638 critical edges here, after loop init.
15639 (pass_data_pre): Remove PROP_no_crit_edges flags.
15640 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15641 for valueization of call args to avoid leaking VN_TOP.
15642 (visit_use): Assert we do not visit default defs.
15643 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15644 Use error_mark_node to more easily detect leaking VN_TOP.
15645 All default-defs are varying, not VN_TOP. Mark them visited.
15646 (run_scc_vn): Make code match comment.
15648 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15650 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15651 OPTION_MASK_FLOAT128_KEYWORD.
15652 (POWERPC_MASKS): Likewise.
15653 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15654 support for the -mfloat128-type option, and make -mfloat128
15655 default on PowerPC Linux systems. Define or undefine
15656 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15657 Define __float128 to be __ieee128 if IEEE 128-bit support is
15658 enabled, or undefine it.
15659 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15660 Delete defining __FLOAT128_TYPE__.
15661 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15662 -mfloat128-type option and make -mfloat128 default on PowerPC
15664 (TARGET_FLOAT128_TYPE): Likewise.
15665 (-mfloat128-type): Likewise.
15666 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15667 Delete the -mfloat128-type option and make -mfloat128 default on
15668 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15669 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15670 128-bit floating point is enabled. Change tests from using
15671 -mfloat128-type to -mfloat128.
15672 (rs6000_mangle_type): Use the correct mangling for the __float128
15673 type even if normal long double is restricted to 64-bits.
15674 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15676 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15677 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15678 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15680 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15681 documentation for -mfloat128.
15683 2017-09-06 Olivier Hainque <hainque@adacore.com>
15685 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15687 2017-09-06 Wish Wu <wishwu007@gmail.com>
15688 Jakub Jelinek <jakub@redhat.com>
15690 * asan.c (initialize_sanitizer_builtins): Add
15691 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15692 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15693 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15694 BT_FN_VOID_UINT64_PTR variables.
15695 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15696 (BT_FN_VOID_UINT16_UINT16): Likewise.
15697 (BT_FN_VOID_UINT32_UINT32): Likewise.
15698 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15699 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15700 (BT_FN_VOID_UINT64_PTR): Likewise.
15701 * common.opt (flag_sanitize_coverage): New variable.
15702 (fsanitize-coverage=trace-pc): Remove.
15703 (fsanitize-coverage=): Add.
15704 * flag-types.h (enum sanitize_coverage_code): New enum.
15705 * fold-const.c (fold_range_test): Disable non-short-circuit
15706 optimization if flag_sanitize_coverage.
15707 (fold_truth_andor): Likewise.
15708 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15709 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15710 (coverage_sanitizer_opts): New array.
15711 (get_closest_sanitizer_option): Add OPTS argument, handle also
15712 OPT_fsanitize_coverage_.
15713 (parse_sanitizer_options): Adjusted to also handle
15714 OPT_fsanitize_coverage_.
15715 (common_handle_option): Add OPT_fsanitize_coverage_.
15716 * sancov.c (instrument_comparison, instrument_switch): New function.
15717 (sancov_pass): Add trace-cmp support.
15718 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15719 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15720 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15721 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15722 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15723 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15724 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15725 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15726 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15727 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15729 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15731 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15732 error. Only quit immediately if parsing is complete.
15733 (BEGIN): Initialize fatal_err and parse_done.
15734 (begin fpu, end fpu): Check number of arguments.
15735 (begin arch, end arch): Likewise.
15736 (begin cpu, end cpu): Likewise.
15737 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15738 (optalias): Likewise.
15740 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15742 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15743 * config/arm/arm-isa.h: Delete. Move definitions to ...
15744 * arm-cpus.in: ... here. Use new feature and fgroup values.
15745 * config/arm/arm.c (arm_option_override): Use lower case for feature
15747 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15748 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15749 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15751 (print_isa_bits_for): New function.
15752 (gen_isa): New function.
15753 (gen_comm_data): Use print_isa_bits_for.
15754 (define feature): New keyword.
15755 (define fgroup): New keyword.
15756 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15757 (arm-isa.h): Add rule to generate file.
15758 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15759 case for feature bit names.
15761 2017-09-06 Richard Biener <rguenther@suse.de>
15763 * tree-ssa-pre.c (NECESSARY): Remove.
15764 (create_expression_by_pieces): Do not touch pass-local flags.
15765 (insert_into_preds_of_block): Likewise.
15766 (do_pre_regular_insertion): Likewise.
15767 (eliminate_insert): Likewise.
15768 (eliminate_dom_walker::before_dom_children): Likewise.
15769 (fini_eliminate): Do not look at inserted_exprs.
15770 (mark_operand_necessary): Remove.
15771 (remove_dead_inserted_code): Replace with simple work-list
15772 algorithm based on inserted_exprs and SSA uses.
15773 (pass_pre::execute): Re-order fini_eliminate and
15774 remove_dead_inserted_code.
15776 2017-09-06 Olivier Hainque <hainque@adacore.com>
15778 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15779 for VxWorks 7. Adjust surrounding comments.
15781 2017-09-06 Richard Biener <rguenther@suse.de>
15783 * gimple-ssa-strength-reduction.c
15784 (find_candidates_dom_walker::before_dom_children): Also allow
15787 2017-09-06 Richard Biener <rguenther@suse.de>
15789 PR tree-optimization/82108
15790 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15791 for gap in the non-permutation SLP case.
15793 2017-09-06 Martin Jambor <mjambor@suse.cz>
15795 PR tree-optimization/82078
15796 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15797 add_access_to_work_queue...
15798 (build_accesses_from_assign): ...here.
15799 (propagate_all_subaccesses): Make sure racc is the group
15800 representative, if there is one.
15802 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15804 PR middle-end/82095
15805 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15808 2017-09-06 Richard Biener <rguenther@suse.de>
15810 * gimple-ssa-strength-reduction.c
15811 (find_candidates_dom_walker::before_doom_children): Use a
15812 type and not a mode check.
15814 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15817 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15819 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15821 2017-09-05 Jeff Law <law@redhat.com>
15823 PR tree-optimization/64910
15824 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15825 cases where we have 3 or more operands.
15827 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15829 PR middle-end/81768
15830 * omp-low.c (lower_omp_for): Recompute tree invariant if
15831 gimple_omp_for_initial/final is ADDR_EXPR.
15833 PR middle-end/81768
15834 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15835 into gimple val before gimplification fo the COND_EXPR.
15837 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15839 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15840 REGION_COPY argument.
15841 (thread_through_all_blocks): Remove unused argument to
15842 duplicate_thread_path.
15844 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15845 Alan Hayward <alan.hayward@arm.com>
15846 David Sherwood <david.sherwood@arm.com>
15848 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15849 Take a scalar_mode rather than a machine_mode.
15850 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15851 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15852 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15853 (aarch64_gen_adjusted_ldpstp): Likewise.
15854 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15856 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15857 Alan Hayward <alan.hayward@arm.com>
15858 David Sherwood <david.sherwood@arm.com>
15860 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15861 Take a scalar_int_mode instead of a machine_mode.
15862 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15863 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15864 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15865 (aarch64_simd_attr_length_rglist): Delete.
15866 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15867 a scalar_int_mode instead of a machine_mode.
15868 (aarch64_add_offset): Likewise.
15869 (aarch64_internal_mov_immediate): Likewise
15870 (aarch64_add_constant_internal): Likewise.
15871 (aarch64_add_constant): Likewise.
15872 (aarch64_movw_imm): Likewise.
15873 (aarch64_rtx_arith_op_extract_p): Likewise.
15874 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15875 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15876 Remove assert that the mode isn't a vector.
15877 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15878 (aarch64_expand_mov_immediate): Update calls after above changes.
15879 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15880 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15881 (aarch64_move_imm): Likewise.
15882 (aarch64_can_const_movi_rtx_p): Likewise.
15883 (aarch64_strip_extend): Likewise.
15884 (aarch64_extr_rtx_p): Likewise.
15885 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15886 a CONST_INT when the mode parameter is VOIDmode.
15887 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15889 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15891 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15892 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15893 (bitwise_type_for_mode): Update accordingly.
15895 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15897 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15898 * stor-layout.c (mode_for_size_tree): Likewise.
15899 (mode_for_array): Update accordingly.
15900 (layout_decl): Likewise.
15901 (compute_record_mode): Likewise. Only set the mode once.
15903 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15905 * target.def (get_mask_mode): Change return type to opt_mode.
15907 * doc/tm.texi: Regenerate.
15908 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15909 * targhooks.c (default_get_mask_mode): Likewise.
15910 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15911 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15912 targetm.get_mask_mode.
15913 * tree.c (build_truth_vector_type): Likewise.
15915 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15917 * machmode.h (mode_for_vector): Return an opt_mode.
15918 * stor-layout.c (mode_for_vector): Likewise.
15919 (mode_for_int_vector): Update accordingly.
15920 (layout_type): Likewise.
15921 * config/i386/i386.c (emit_memmov): Likewise.
15922 (ix86_expand_set_or_movmem): Likewise.
15923 (ix86_expand_vector_init): Likewise.
15924 (ix86_get_mask_mode): Likewise.
15925 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15927 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15928 * expmed.c (extract_bit_field_1): Likewise.
15929 * expr.c (expand_expr_real_2): Likewise.
15930 * optabs-query.c (can_vec_perm_p): Likewise.
15931 (can_vec_mask_load_store_p): Likewise.
15932 * optabs.c (expand_vec_perm): Likewise.
15933 * targhooks.c (default_get_mask_mode): Likewise.
15934 * tree-vect-stmts.c (vectorizable_store): Likewise.
15935 (vectorizable_load): Likewise.
15936 (get_vectype_for_scalar_type_and_size): Likewise.
15938 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15940 * machmode.h (mode_for_int_vector): New function.
15941 * stor-layout.c (mode_for_int_vector): Likewise.
15942 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15943 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15944 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15945 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15946 (s390_expand_vcond): Likewise.
15948 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15950 * machmode.h (opt_machine_mode): New type.
15951 (opt_mode<T>): Allow construction from anything that can be
15953 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15954 (mode_for_size): Return an opt_machine_mode.
15955 * stor-layout.c (mode_for_size): Likewise.
15956 (mode_for_size_tree): Update call accordingly.
15957 (bitwise_mode_for_mode): Likewise.
15958 (make_fract_type): Likewise.
15959 (make_accum_type): Likewise.
15960 * caller-save.c (replace_reg_with_saved_mem): Update call
15962 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15963 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15964 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15965 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15966 * expmed.c (extract_bit_field_1): Likewise.
15967 * reload.c (get_secondary_mem): Likewise.
15968 * varasm.c (assemble_integer): Likewise.
15969 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15972 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15974 * machmode.h (decimal_float_mode_for_size): New function.
15975 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15976 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15977 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15978 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15979 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15980 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15982 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15984 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15985 (get_builtin_sync_mode): Likewise.
15986 (expand_ifn_atomic_compare_exchange): Likewise.
15987 (expand_builtin_atomic_clear): Likewise.
15988 (expand_builtin_atomic_test_and_set): Likewise.
15989 (fold_builtin_atomic_always_lock_free): Likewise.
15990 * calls.c (compute_argument_addresses): Likewise.
15991 (emit_library_call_value_1): Likewise.
15992 (store_one_arg): Likewise.
15993 * combine.c (combine_instructions): Likewise.
15994 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15995 * config/arm/arm.c (arm_function_value): Likewise.
15996 (aapcs_allocate_return_reg): Likewise.
15997 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15998 * config/i386/i386.c (construct_container): Likewise.
15999 (ix86_gimplify_va_arg): Likewise.
16000 (ix86_expand_sse_cmp): Likewise.
16001 (emit_memmov): Likewise.
16002 (emit_memset): Likewise.
16003 (expand_small_movmem_or_setmem): Likewise.
16004 (ix86_expand_pextr): Likewise.
16005 (ix86_expand_pinsr): Likewise.
16006 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16007 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16009 * config/mips/mips.c (mips_function_value_1) Likewise.
16010 (mips_block_move_straight): Likewise.
16011 (mips_expand_ins_as_unaligned_store): Likewise.
16012 * config/powerpcspe/powerpcspe.c
16013 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16014 (rs6000_darwin64_record_arg_flush): Likewise.
16015 * config/rs6000/rs6000.c
16016 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16017 (rs6000_darwin64_record_arg_flush): Likewise.
16018 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16019 (sparc_function_value_1): Likewise.
16020 * config/spu/spu.c (adjust_operand): Likewise.
16021 (spu_emit_branch_or_set): Likewise.
16022 (arith_immediate_p): Likewise.
16023 * emit-rtl.c (gen_lowpart_common): Likewise.
16024 * expr.c (expand_expr_real_1): Likewise.
16025 * function.c (assign_parm_setup_block): Likewise.
16026 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16027 * reload1.c (alter_reg): Likewise.
16028 * stor-layout.c (mode_for_vector): Likewise.
16029 (layout_type): Likewise.
16031 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16033 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16034 (spu_convert_move): Likewise.
16035 * lower-subreg.c (resolve_simple_move): Likewise.
16037 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16040 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16041 define_insn to a define_expand.
16042 (altivec_vsum2sws_direct): New define_insn.
16043 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16045 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16047 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16050 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16055 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16056 realign stack if stack alignment needed is less than incoming
16059 2017-09-05 Marek Polacek <polacek@redhat.com>
16062 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16065 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16067 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16069 2017-09-05 Richard Biener <rguenther@suse.de>
16071 PR tree-optimization/82084
16072 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16074 2017-09-05 Richard Biener <rguenther@suse.de>
16076 PR tree-optimization/82102
16077 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16079 2017-09-05 Martin Liska <mliska@suse.cz>
16081 PR tree-optimization/82032
16082 * tree-cfg.c (generate_range_test): New function.
16083 * tree-cfg.h (generate_range_test): Declared here.
16084 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16085 (cleanup_control_expr_graph): Use it.
16086 * tree-switch-conversion.c (try_switch_expansion): Remove
16088 (emit_case_nodes): Use generate_range_test.
16090 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16093 * config/i386/i386.md (*<btsc><mode>_mask): Add
16094 TARGET_USE_BT to insn constraint.
16095 (*btr<mode>_mask): Ditto.
16097 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16099 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16100 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16102 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16105 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16106 TARGET_NEON and TARGET_IWMMXT.
16107 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16108 TARGET_NEON and TARGET_IWMMXT.
16109 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16111 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16113 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16114 (ix86_rewrite_tls_address): Ditto.
16115 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16116 (ix86_rewrite_tls_address_1): Ditto.
16117 (ix86_rewrite_tls_address): Ditto.
16118 * config/i386/predicates.md (tls_address_pattern): New predicate.
16119 * config/i386/i386.md (TLS address splitter): New splitter.
16121 2017-09-04 Richard Biener <rguenther@suse.de>
16123 PR tree-optimization/82084
16124 * fold-const.h (can_native_encode_string_p): Declare.
16125 * fold-const.c (can_native_encode_string_p): Factor out from ...
16126 (native_encode_string): ... here.
16127 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16128 vectorizing stores from constants we later cannot handle.
16130 2017-09-04 Marek Polacek <polacek@redhat.com>
16133 * doc/invoke.texi: Update -Wtautological-compare documentation.
16135 2017-09-04 Jeff Law <law@redhat.com>
16137 PR tree-optimization/64910
16138 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16139 swap the first and last operand if the last is a constant.
16141 2017-09-04 Marek Polacek <polacek@redhat.com>
16144 * convert.c (do_narrow): When sanitizing signed integer overflows,
16145 bail out for signed types.
16146 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16148 2017-09-04 Richard Biener <rguenther@suse.de>
16150 PR tree-optimization/82060
16151 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16152 Move devirtualization after stmt folding and before EH/AB/noreturn
16153 cleanup to get the stmt refs canonicalized. Use a bool instead
16154 of gimple_modified_p since that doesn't work for NOPs. Schedule
16155 NOPs generated by folding for removal.
16157 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16158 Alan Hayward <alan.hayward@arm.com>
16159 David Sherwood <david.sherwood@arm.com>
16161 * coretypes.h (pad_direction): New enum.
16162 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16163 (FUNCTION_ARG_PADDING): Likewise.
16164 * target.def (function_arg_padding): New hook.
16165 * targhooks.h (default_function_arg_padding): Declare.
16166 * targhooks.c (default_function_arg_padding): New function.
16167 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16168 (TARGET_FUNCTION_ARG_PADDING): ...this.
16169 * doc/tm.texi: Regenerate.
16170 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16171 instead of direction.
16172 (compute_argument_addresses): Likewise.
16173 (load_register_parameters): Likewise.
16174 (emit_library_call_value_1): Likewise.
16175 (store_one_arg): Use targetm.calls.function_arg_padding instead
16176 of FUNCTION_ARG_PADDING.
16177 (must_pass_in_stack_var_size_or_pad): Likewise.
16178 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16179 (emit_group_store): Likewise.
16180 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16181 instead of FUNCTION_ARG_PADDING.
16182 (emit_push_insn): Likewise, and propagate enum change throughout
16184 * function.h (direction): Delete.
16185 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16187 * function.c (assign_parm_find_stack_rtl): Likewise.
16188 (assign_parm_setup_block_p): Likewise.
16189 (assign_parm_setup_block): Likewise.
16190 (gimplify_parameters): Likewise.
16191 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16192 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16194 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16195 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16196 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16197 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16198 (aarch64_function_arg_padding): ...this new function.
16199 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16200 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16201 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16202 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16203 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16204 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16205 (arm_pad_arg_upward): Replace with...
16206 (arm_function_arg_padding): ...this new function.
16207 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16209 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16210 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16211 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16212 (ia64_hpux_function_arg_padding): Replace with...
16213 (ia64_function_arg_padding): ...this new function. Use pad_direction
16214 instead of direction. Check for TARGET_HPUX.
16215 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16216 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16217 (iq2000_function_arg_padding): New function.
16218 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16219 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16220 (mips_function_arg_padding): ...this new function.
16221 (mips_pad_reg_upward): Update accordingly.
16222 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16223 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16224 targetm.calls.function_arg_padding.
16225 (FUNCTION_ARG_PADDING): Delete.
16226 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16227 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16228 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16229 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16230 (nios2_block_reg_padding): Return pad_direction instead of direction.
16231 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16232 instead of direction.
16233 (nios2_function_arg_padding): Likewise. Make static.
16234 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16235 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16236 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16237 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16238 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16239 (pa_function_arg_padding): Make static. Return pad_direction instead
16241 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16242 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16243 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16244 instead of direction. Use targetm.calls.function_arg_padding.
16245 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16246 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16247 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16248 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16249 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16251 (function_arg_padding): Rename to...
16252 (rs6000_function_arg_padding): ...this. Make static. Return
16253 pad_direction instead of direction.
16254 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16255 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16256 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16257 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16258 instead of direction. Use targetm.calls.function_arg_padding.
16259 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16260 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16261 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16262 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16263 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16264 (function_arg_padding): Rename to...
16265 (rs6000_function_arg_padding): ...this. Make static. Return
16266 pad_direction instead of direction.
16267 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16268 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16269 * config/s390/s390.c (s390_function_arg_padding): New function.
16270 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16271 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16272 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16273 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16274 (function_arg_padding): Rename to...
16275 (sparc_function_arg_padding): ...this. Make static. Return
16276 pad_direction instead of direction.
16277 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16278 * config/spu/spu.c (spu_function_arg_padding): New function.
16279 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16280 * system.h (FUNCTION_ARG_PADDING): Poison.
16282 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16283 Alan Hayward <alan.hayward@arm.com>
16284 David Sherwood <david.sherwood@arm.com>
16286 * target.def (modes_tieable_p): New hook.
16287 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16288 (TARGET_MODES_TIEABLE_P): ...this.
16289 * doc/tm.texi.in: Regenerate.
16290 * hooks.h (hook_bool_mode_mode_true): Declare.
16291 * hooks.c (hook_bool_mode_mode_true): New function.
16292 * combine.c (subst): Use targetm.modes_tieable_p instead of
16294 * dse.c (find_shift_sequence): Likewise.
16295 * expmed.c (extract_low_bits): Likewise.
16296 * lower-subreg.c: Include target.h.
16297 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16299 * rtlanal.c (rtx_cost): Likewise.
16300 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16301 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16302 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16303 (TARGET_MODES_TIEABLE_P): Redefine.
16304 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16305 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16306 (TARGET_MODES_TIEABLE_P): Redefine.
16307 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16308 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16309 (arc_modes_tieable_p): New function.
16310 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16311 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16312 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16313 (arm_modes_tieable_p): Make static.
16314 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16315 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16316 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16317 (TARGET_MODES_TIEABLE_P): Redefine.
16318 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16319 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16320 (TARGET_MODES_TIEABLE_P): Redefine.
16321 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16322 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16323 (cr16_modes_tieable_p): New function.
16324 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16325 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16326 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16327 (TRULY_NOOP_TRUNCATION): Update comment.
16328 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16329 (TRULY_NOOP_TRUNCATION): Update comment.
16330 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16331 (frv_modes_tieable_p): New function.
16332 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16333 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16334 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16335 (TARGET_MODES_TIEABLE_P): Redefine.
16336 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16337 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16338 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16339 (TARGET_MODES_TIEABLE_P): Redefine.
16340 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16341 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16342 (ia64_modes_tieable_p): New function.
16343 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16344 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16345 (iq2000_modes_tieable_p): New function.
16346 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16347 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16348 (lm32_modes_tieable_p): New function.
16349 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16350 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16351 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16352 (TARGET_MODES_TIEABLE_P): Redefine.
16353 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16354 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16355 (m32r_modes_tieable_p): New function.
16356 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16357 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16358 (m68k_modes_tieable_p): New function.
16359 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16360 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16361 (mcore_modes_tieable_p): New function.
16362 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16363 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16365 (TARGET_MODES_TIEABLE_P): Redefine.
16366 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16367 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16368 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16369 (TARGET_MODES_TIEABLE_P): Redefine.
16370 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16371 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16372 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16373 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16374 (mn10300_modes_tieable_p): ...this and make static.
16375 (TARGET_MODES_TIEABLE_P): Redefine.
16376 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16377 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16378 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16379 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16380 (msp430_modes_tieable_p): Make static.
16381 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16382 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16383 (TARGET_MODES_TIEABLE_P): Redefine.
16384 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16385 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16386 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16387 (TARGET_MODES_TIEABLE_P): Redefine.
16388 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16389 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16390 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16391 (TARGET_MODES_TIEABLE_P): Redefine.
16392 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16393 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16394 (pdp11_modes_tieable_p): New function.
16395 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16396 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16397 (rs6000_modes_tieable_p): New function.
16398 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16399 * config/powerpcspe/powerpcspe.md: Update comment.
16400 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16401 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16402 (TARGET_MODES_TIEABLE_P): Redefine.
16403 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16404 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16405 (rl78_modes_tieable_p): New function.
16406 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16407 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16408 (rs6000_modes_tieable_p): New function.
16409 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16410 * config/rs6000/rs6000.md: Update comment.
16411 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16412 * config/rx/rx.c (rx_modes_tieable_p): New function.
16413 (TARGET_MODES_TIEABLE_P): Redefine.
16414 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16415 * config/s390/s390.c (s390_modes_tieable_p): New function.
16416 (TARGET_MODES_TIEABLE_P): Redefine.
16417 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16418 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16419 (sh_modes_tieable_p): New function.
16420 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16421 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16422 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16423 (sparc_modes_tieable_p): Make static.
16424 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16425 * config/spu/spu.c (spu_modes_tieable_p): New function.
16426 (TARGET_MODES_TIEABLE_P): Redefine.
16427 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16428 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16429 (TARGET_MODES_TIEABLE_P): Redefine.
16430 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16431 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16432 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16433 * config/v850/v850.c (v850_modes_tieable_p): New function.
16434 (TARGET_MODES_TIEABLE_P): Redefine.
16435 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16436 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16437 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16438 (visium_modes_tieable_p): New function.
16439 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16440 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16441 (xtensa_modes_tieable_p): New function.
16442 * system.h (MODES_TIEABLE_P): Poison.
16444 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16445 Alan Hayward <alan.hayward@arm.com>
16446 David Sherwood <david.sherwood@arm.com>
16448 * target.def (hard_regno_mode_ok): New hook.
16449 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16450 (TARGET_HARD_REGNO_MODE_OK): ...this.
16451 * doc/tm.texi.in: Regenerate.
16452 * hooks.h (hook_bool_uint_mode_true): Declare.
16453 * hooks.c (hook_bool_uint_mode_true): New function.
16454 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16455 HARD_REGNO_MODE_OK.
16456 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16457 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16458 instead of HARD_REGNO_MODE_OK.
16459 * caller-save.c: Include target.h.
16460 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16461 HARD_REGNO_MODE_OK.
16462 * combine.c (can_combine_p): Likewise.
16463 (combinable_i3pat): Likewise.
16464 (can_change_dest_mode): Likewise.
16465 * expr.c (init_expr_target): Likewise.
16466 (convert_move): Likewise.
16467 (convert_modes): Likewise.
16468 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16469 (setup_prohibited_mode_move_regs): Likewise.
16470 * ira.h (target_ira): Likewise.
16471 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16472 * lra-constraints.c (process_alt_operands): Likewise.
16473 (split_reg): Likewise.
16474 * recog.c (peep2_find_free_register): Likewise.
16475 * ree.c (combine_reaching_defs): Likewise.
16476 * regcprop.c (maybe_mode_change): Likewise.
16477 * reginfo.c (init_reg_sets_1): Likewise.
16478 (choose_hard_reg_mode): Likewise.
16479 (simplifiable_subregs): Likewise.
16480 * regrename.c (check_new_reg_p): Likewise.
16481 * reload.c (find_valid_class): Likewise.
16482 (find_valid_class_1): Likewise.
16483 (reload_inner_reg_of_subreg): Likewise.
16484 (push_reload): Likewise.
16485 (combine_reloads): Likewise.
16486 (find_dummy_reload): Likewise.
16487 (find_reloads): Likewise.
16488 * reload1.c (find_reg): Likewise.
16489 (set_reload_reg): Likewise.
16490 (allocate_reload_reg): Likewise.
16491 (choose_reload_regs): Likewise.
16492 (reload_adjust_reg_for_temp): Likewise.
16493 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16494 (simplify_subreg_regno): Likewise.
16495 * sel-sched.c (init_regs_for_mode): Likewise.
16496 * varasm.c (make_decl_rtl): Likewise.
16497 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16498 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16499 HARD_REGNO_MODE_OK.
16500 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16501 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16502 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16503 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16504 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16505 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16506 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16507 (arc_mode_class): Delete.
16508 (HARD_REGNO_MODE_OK): Delete.
16509 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16510 (arc_hard_regno_mode_ok): Rename old array to...
16511 (arc_hard_regno_mode_ok_modes): ...this.
16512 (arc_conditional_register_usage): Update accordingly.
16513 (arc_mode_class): Make static.
16514 (arc_hard_regno_mode_ok): New function.
16515 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16516 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16517 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16518 (arm_hard_regno_mode_ok): Make static.
16519 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16520 HARD_REGNO_MODE_OK.
16521 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16522 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16523 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16525 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16526 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16527 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16528 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16529 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16530 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16531 * config/bfin/predicates.md (valid_reg_operand): Use
16532 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16533 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16534 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16535 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16536 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16537 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16538 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16539 (cr16_hard_regno_mode_ok): Make static and return a bool.
16540 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16541 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16542 (cris_hard_regno_mode_ok): New function.
16543 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16544 (epiphany_mode_class): Delete.
16545 (HARD_REGNO_MODE_OK): Delete.
16546 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16547 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16548 (hard_regno_mode_ok): Rename to...
16549 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16550 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16551 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16552 HARD_REGNO_MODE_OK.
16553 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16554 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16555 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16556 (frv_hard_regno_mode_ok): Make static and return a bool.
16557 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16558 HARD_REGNO_MODE_OK.
16559 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16560 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16561 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16562 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16564 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16565 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16566 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16567 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16569 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16570 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16571 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16572 (ia64_hard_regno_mode_ok): New function.
16573 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16574 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16575 (iq2000_hard_regno_mode_ok): New function.
16576 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16577 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16578 (lm32_hard_regno_mode_ok): New function.
16579 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16580 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16581 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16582 instead of HARD_REGNO_MODE_OK.
16583 (m32c_hard_regno_ok): Rename to...
16584 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16585 (m32c_cannot_change_mode_class): Update accordingly.
16586 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16587 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16588 (m32r_mode_class): Delete.
16589 (HARD_REGNO_MODE_OK): Delete.
16590 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16591 (m32r_hard_regno_mode_ok): Rename to...
16592 (m32r_hard_regno_modes): ...this.
16593 (m32r_mode_class): Make static.
16594 (m32r_hard_regno_mode_ok): New function.
16595 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16596 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16597 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16598 (m68k_hard_regno_mode_ok): Make static.
16599 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16600 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16601 (mcore_hard_regno_mode_ok): New function.
16602 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16603 (HARD_REGNO_MODE_OK): Delete.
16604 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16606 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16607 (microblaze_hard_regno_mode_ok): New function.
16608 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16609 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16610 (mips_hard_regno_mode_ok): Delete.
16611 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16612 (mips_hard_regno_mode_ok_p): ...this and make static.
16613 (mips_hard_regno_mode_ok_p): Rename to...
16614 (mips_hard_regno_mode_ok_uncached): ...this.
16615 (mips_hard_regno_mode_ok): New function.
16616 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16617 of HARD_REGNO_MODE_OK.
16618 (mips_option_override): Update after above name changes.
16619 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16620 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16621 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16622 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16623 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16624 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16625 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16626 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16627 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16628 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16629 (msp430_hard_regno_mode_ok): Make static and return a bool.
16630 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16631 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16632 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16634 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16635 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16636 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16637 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16638 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16639 (PA_HARD_REGNO_MODE_OK): ...this
16640 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16641 (PA_HARD_REGNO_MODE_OK): ...this.
16642 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16643 (pa_hard_regno_mode_ok): New function.
16644 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16645 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16646 (pdp11_hard_regno_mode_ok): New function.
16647 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16648 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16650 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16652 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16653 (rs6000_hard_regno_mode_ok): Rename to...
16654 (rs6000_hard_regno_mode_ok_uncached): ...this.
16655 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16656 (rs6000_hard_regno_mode_ok): New function.
16657 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16658 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16659 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16660 (riscv_hard_regno_mode_ok): ...this and make static.
16661 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16662 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16663 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16664 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16665 (rl78_hard_regno_mode_ok): Make static and return bool.
16666 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16667 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16669 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16670 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16671 (rs6000_hard_regno_mode_ok): Rename to...
16672 (rs6000_hard_regno_mode_ok_uncached): ...this.
16673 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16674 (rs6000_hard_regno_mode_ok): New function.
16675 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16676 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16677 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16678 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16679 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16680 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16681 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16682 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16683 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16684 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16685 (sh_hard_regno_mode_ok): Make static.
16686 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16687 instead of HARD_REGNO_MODE_OK.
16688 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16689 (sparc_mode_class): Delete.
16690 (HARD_REGNO_MODE_OK): Delete.
16691 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16692 (hard_regno_mode_classes): Make static.
16693 (sparc_mode_class): Likewise.
16694 (sparc_hard_regno_mode_ok): New function.
16695 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16696 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16697 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16699 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16700 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16701 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16702 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16703 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16704 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16705 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16706 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16707 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16708 (visium_hard_regno_mode_ok): New function.
16709 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16710 instead of HARD_REGNO_MODE_OK.
16711 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16712 (HARD_REGNO_MODE_OK): Delete.
16713 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16714 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16715 (xtensa_option_override): Update accordingly.
16716 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16717 (xtensa_hard_regno_mode_ok): New function.
16718 * system.h (HARD_REGNO_MODE_OK): Poison.
16720 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16721 Alan Hayward <alan.hayward@arm.com>
16722 David Sherwood <david.sherwood@arm.com>
16724 * target.def (hard_regno_call_part_clobbered): New hook.
16725 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16726 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16727 * doc/tm.texi: Regenerate.
16728 * hooks.h (hook_bool_uint_mode_false): Declare.
16729 * hooks.c (hook_bool_uint_mode_false): New function.
16730 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16731 * cselib.c (cselib_process_insn): Use
16732 targetm.hard_regno_call_part_clobbered instead of
16733 HARD_REGNO_CALL_PART_CLOBBERED.
16734 * ira-conflicts.c (ira_build_conflicts): Likewise.
16735 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16736 * lra-constraints.c (need_for_call_save_p): Likewise.
16737 * lra-lives.c: Include target.h.
16738 (check_pseudos_live_through_calls): Use
16739 targetm.hard_regno_call_part_clobbered instead of
16740 HARD_REGNO_CALL_PART_CLOBBERED.
16741 * regcprop.c: Include target.h.
16742 (copyprop_hardreg_forward_1): Use
16743 targetm.hard_regno_call_part_clobbered instead of
16744 HARD_REGNO_CALL_PART_CLOBBERED.
16745 * reginfo.c (choose_hard_reg_mode): Likewise.
16746 * regrename.c (check_new_reg_p): Likewise.
16747 * reload.c (find_equiv_reg): Likewise.
16748 * reload1.c (emit_reload_insns): Likewise.
16749 * sched-deps.c (deps_analyze_insn): Likewise.
16750 * sel-sched.c (init_regs_for_mode): Likewise.
16751 (mark_unavailable_hard_regs): Likewise.
16752 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16753 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16754 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16756 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16757 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16758 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16760 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16762 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16763 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16764 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16766 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16767 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16768 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16770 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16771 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16773 * config/powerpcspe/powerpcspe.c
16774 (rs6000_hard_regno_call_part_clobbered): New function.
16775 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16776 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16777 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16779 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16780 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16781 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16783 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16784 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16785 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16787 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16788 Alan Hayward <alan.hayward@arm.com>
16789 David Sherwood <david.sherwood@arm.com>
16791 * rtl.h (subreg_memory_offset): Declare.
16792 * emit-rtl.c (subreg_memory_offset): New function.
16793 * expmed.c (store_bit_field_1): Use it.
16794 * expr.c (undefined_operand_subword_p): Likewise.
16795 * simplify-rtx.c (simplify_subreg): Likewise.
16797 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16799 PR rtl-optimization/57448
16802 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16803 using atomic_load pattern.
16804 (expand_atomic_store): Likewise.
16806 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16809 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16810 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16811 handling. Use replace_call_with_value with NULL instead of
16812 gsi_replace, unlink_stmt_vdef and release_defs.
16814 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16817 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16819 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16822 * rtl.h (emit_library_call_value_1): Declare.
16823 (emit_library_call): Replace declaration with a series of overloads.
16824 Remove the parameter count argument.
16825 (emit_library_call_value): Likewise.
16826 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16827 with an "rtx_mode_t *".
16828 (emit_library_call_value): Delete.
16829 (emit_library_call): Likewise.
16830 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16831 (asan_emit_allocas_unpoison): Likewise.
16832 * builtins.c (expand_builtin_powi): Likewise.
16833 (expand_asan_emit_allocas_unpoison): Likewise.
16834 * cfgexpand.c (expand_main_function): Likewise.
16835 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16836 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16837 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16838 * config/arm/arm.c (arm_trampoline_init): Likewise.
16839 (arm_call_tls_get_addr): Likewise.
16840 (arm_expand_divmod_libfunc): Likewise.
16841 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16842 (smulsi3_highpart): Likewise.
16843 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16844 (c6x_expand_compare): Likewise.
16845 (c6x_expand_movmem): Likewise.
16846 * config/frv/frv.c (frv_trampoline_init): Likewise.
16847 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16848 (ix86_expand_divmod_libfunc): Likewise.
16849 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16850 (ia64_expand_compare): Likewise.
16851 (ia64_profile_hook): Likewise.
16852 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16853 (nonlocal_goto): Likewise.
16854 (restore_stack_nonlocal): Likewise.
16855 * config/m32r/m32r.c (block_move_call): Likewise.
16856 (m32r_trampoline_init): Likewise.
16857 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16858 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16859 (m68k_call_m68k_read_tp): Likewise.
16860 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16861 (microblaze_expand_divide): Likewise.
16862 * config/mips/mips.h (mips_args): Likewise.
16863 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16864 (MIPS_ICACHE_SYNC): Likewise.
16865 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16866 (nios2_trampoline_init): Likewise.
16867 * config/pa/pa.c (hppa_tls_call): Likewise.
16868 (pa_trampoline_init): Likewise.
16869 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16870 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16871 (expand_strn_compare): Likewise.
16872 (rs6000_generate_compare): Likewise.
16873 (rs6000_expand_float128_convert): Likewise.
16874 (output_profile_hook): Likewise.
16875 (rs6000_trampoline_init): Likewise.
16876 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16877 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16878 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16879 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16880 (rs6000_generate_compare): Likewise.
16881 (rs6000_expand_float128_convert): Likewise.
16882 (output_profile_hook): Likewise.
16883 (rs6000_trampoline_init): Likewise.
16884 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16885 * config/sh/sh.c (sh_trampoline_init): Likewise.
16886 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16887 (sparc_emit_float_lib_cmp): Likewise.
16888 (sparc32_initialize_trampoline): Likewise.
16889 (sparc64_initialize_trampoline): Likewise.
16890 (sparc_profile_hook): Likewise.
16891 * config/spu/spu.c (ea_load_store): Likewise.
16892 * config/spu/spu.md (floatunssidf2): Likewise.
16893 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16894 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16895 * config/visium/visium.c (expand_block_move_4): Likewise.
16896 (expand_block_move_2): Likewise.
16897 (expand_block_move_1): Likewise.
16898 (expand_block_set_4): Likewise.
16899 (expand_block_set_2): Likewise.
16900 (expand_block_set_1): Likewise.
16901 (visium_trampoline_init): Likewise.
16902 (visium_profile_hook): Likewise.
16903 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16904 (xtensa_setup_frame_addresses): Likewise.
16905 (xtensa_trampoline_init): Likewise.
16906 * except.c (sjlj_emit_function_enter): Likewise.
16907 (sjlj_emit_function_exit): Likewise.
16908 * explow.c (allocate_dynamic_stack_space): Likewise.
16909 (probe_stack_range): Likewise.
16910 * expr.c (convert_mode_scalar): Likewise.
16911 * optabs.c (expand_binop): Likewise.
16912 (expand_twoval_binop_libfunc): Likewise.
16913 (expand_unop): Likewise.
16914 (prepare_cmp_insn): Likewise.
16915 (prepare_float_lib_cmp): Likewise.
16916 (expand_float): Likewise.
16917 (expand_fix): Likewise.
16918 (expand_fixed_convert): Likewise.
16919 (maybe_emit_sync_lock_test_and_set): Likewise.
16920 (expand_atomic_compare_and_swap): Likewise.
16921 (expand_mem_thread_fence): Likewise.
16922 (expand_atomic_fetch_op): Likewise.
16924 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16926 * doc/generic.texi (OpenACC): Adjust URL.
16927 * doc/invoke.texi (C Dialect Options): Ditto.
16929 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16931 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16932 predicate for operand 1. Add (m,<S>) constraint.
16933 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16934 Prevent memory operand 1 with register operand 2.
16936 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16938 PR rtl-optimization/82024
16939 * combine.c (try_combine): If the combination result is a PARALLEL,
16940 and we only need to retain the SET in there that would be placed
16941 at I2, check that we can place that at I3 instead, before doing so.
16943 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16946 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16948 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16949 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16952 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16953 Jeff Law <law@redhat.com>
16955 * varasm.c (bss_initializer_p): Do not put constants into .bss
16956 (categorize_decl_for_section): Handle bss_initializer_p returning
16957 false when DECL_INITIAL is NULL.
16959 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16962 * config/s390/s390.c (s390_can_inline_p): New function.
16964 2017-09-01 Jeff Law <law@redhat.com>
16966 PR tree-optimization/82052
16967 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16968 Always initialize the returned slot after a hash table miss
16969 when INSERT is true.
16971 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16973 * config/s390/s390.md (mem_signal_fence): Remove.
16974 * doc/md.texi (mem_signal_fence): Remove.
16975 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16977 * target-insns.def (mem_signal_fence): Remove.
16979 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16982 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16985 * asan.c (create_odr_indicator): Strip name encoding from assembler
16986 name before appending it after __odr_asan_.
16988 2017-09-01 Martin Liska <mliska@suse.cz>
16990 PR tree-optimization/82059
16991 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16992 frequency only when an edge is redirected.
16994 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16996 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16997 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16998 (arc_conditional_register_usage): Remove ARC600 lp_count
17000 (arc_file_start): Emit Tag_ARC_CPU_variation.
17001 (arc_can_use_doloop_p): New conditions to use ZOLs.
17002 (hwloop_fail): New function.
17003 (hwloop_optimize): Likewise.
17004 (hwloop_pattern_reg): Likewise.
17005 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17006 (arc_reorg_loops): New function, calls reorg_loops.
17007 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17008 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17010 (arc_loop_hazard): Remove function, functionality moved into
17012 (arc_hazard): Remove arc_loop_hazard call.
17013 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17014 into hwloop_optimize.
17015 (arc_label_align): Remove ZOL handling.
17016 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17017 * config/arc/arc.md (doloop_begin): Remove pattern.
17018 (doloop_begin_i): Likewise.
17019 (doloop_end_i): Likewise.
17020 (doloop_fallback): Likewise.
17021 (doloop_fallback_m): Likewise.
17022 (doloop_end): Reimplement expand.
17023 (arc_lp): New pattern for LP instruction.
17024 (loop_end): New pattern.
17025 (loop_fail): Likewise.
17026 (decrement_and_branch_until_zero): Likewise.
17027 * config/arc/arc.opt (mlpc-width): New option.
17028 * doc/invoke.texi (mlpc-width): Document option.
17030 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17032 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17033 (arc_ccfsm_advance): Fix checking for delay slots.
17034 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17036 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17038 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17040 (movhi_insn): Update store instruction constraint which are saving
17041 6-bit short immediates.
17042 (movsi_insn): Consider also short scaled load operations.
17043 (zero_extendhisi2_i): Use Usd constraint instead of T.
17044 (extendhisi2_i): Add q constraint.
17045 (arc_clzsi2): Add type and length attributes.
17046 (arc_ctzsi2): Likewise.
17047 * config/arc/constraints.md (Usc): Update constraint, the
17048 assembler can parse two relocations for a single instruction.
17050 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17052 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17053 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17055 2017-08-31 Olivier Hainque <hainque@adacore.com>
17057 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17058 match as powerpc-wrs-vxworks*.
17060 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17062 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17063 register constraint for by-element operand.
17064 (aarch64_mls_elt_merge<mode>): Likewise.
17066 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17068 * config/arc/arc.c (arc_can_follow_jump): Check for short
17071 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17073 * config.gcc: Use g.opt for arc.
17074 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17075 functionality moved to ...
17076 (legitimate_scaled_address_p): New function, ...here.
17077 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17078 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17079 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17081 (arc_override_options): Handle G option.
17082 (arc_output_pic_addr_const): Correct function definition.
17083 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17084 (arc_decl_anon_ns_mem_p): Delete.
17085 (arc_in_small_data_p): Overhaul this function to take into
17086 consideration the value given via G option.
17087 (arc_rewrite_small_data_1): Renamed and corrected old
17088 arc_rewrite_small_data function.
17089 (arc_rewrite_small_data): New function.
17090 (small_data_pattern): Don't use pic_offset_table_rtx.
17091 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17092 * config/arc/simdext.md (movmisalignv2hi): Use
17093 prepare_move_operands function.
17095 (movmisalign*): Likewise.
17096 * doc/invoke.texi (ARC options): Document -G option.
17098 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17100 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17102 * config/arc/arc.c (arc_print_operand): Output scalled address for
17103 sdata whenever is possible.
17104 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17105 load/stores are available.
17106 (compact_sda_memory_operand): Check for the alignment required by
17107 code density instructions.
17108 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17110 * config/arc/constraints.md (Usd): Update constraint.
17111 (Us0): New constraint.
17112 (Usc): Update constraint.
17114 2017-08-31 Richard Biener <rguenther@suse.de>
17116 PR middle-end/82054
17117 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17118 function only once.
17120 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17122 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17123 Resize type_signature.
17125 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17126 Alan Hayward <alan.hayward@arm.com>
17127 David Sherwood <david.sherwood@arm.com>
17129 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17130 subregs whose inner modes can be stored in GPRs.
17131 (aarch64_classify_index): Likewise.
17133 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17134 Alan Hayward <alan.hayward@arm.com>
17135 David Sherwood <david.sherwood@arm.com>
17137 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17138 (V_INT_EQUIV): ...this.
17139 (v_cmp_result): Rename to...
17140 (v_int_equiv): ...this.
17141 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17142 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17143 (copysign<mode>3): Likewise.
17144 (aarch64_simd_bsl<mode>_internal): Likewise.
17145 (aarch64_simd_bsl<mode>): Likewise.
17146 (vec_cmp<mode><mode>): Likewise.
17147 (vcond<mode><mode>): Likewise.
17148 (vcond<v_cmp_mixed><mode>): Likewise.
17149 (vcondu<mode><v_cmp_mixed>): Likewise.
17150 (aarch64_cm<optab><mode>): Likewise.
17151 (aarch64_cmtst<mode>): Likewise.
17152 (aarch64_fac<optab><mode>): Likewise.
17153 (vec_perm_const<mode>): Likewise.
17154 (vcond_mask_<mode><v_cmp_result>): Rename to...
17155 (vcond_mask_<mode><v_int_equiv>): ...this.
17156 (vec_cmp<mode><v_cmp_result>): Rename to...
17157 (vec_cmp<mode><v_int_equiv>): ...this.
17159 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17160 Alan Hayward <alan.hayward@arm.com>
17161 David Sherwood <david.sherwood@arm.com>
17163 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17165 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17166 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17167 (UNSPEC_LD4_DREG): New unspecs.
17168 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17169 (aarch64_ld2<mode>_dreg_be): Replace with...
17170 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17172 (aarch64_ld3<mode>_dreg_le)
17173 (aarch64_ld3<mode>_dreg_be): Replace with...
17174 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17176 (aarch64_ld4<mode>_dreg_le)
17177 (aarch64_ld4<mode>_dreg_be): Replace with...
17178 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17181 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17183 PR tree-optimization/81987
17184 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17185 insert an initializer in a location not dominated by the stride
17188 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17190 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17191 on the entire header of the finally block in the fallthru case.
17193 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17195 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17197 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17199 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17200 rs6000_emit_move_from_cr and call renamed function.
17201 (rs6000_emit_prologue): Call renamed functions.
17202 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17203 movesi_from_cr, remove volatile CRs.
17205 2017-08-30 Jon Beniston <jon@beniston.com>
17206 Richard Biener <rguenther@suse.de>
17208 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17209 of VECTOR_MODE_P check.
17210 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17211 element vector types.
17213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17215 * df.h (df_read_modify_subreg_p): Remove in favor of...
17216 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17217 const_rtx instead of an rtx.
17218 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17219 * df-problems.c (df_word_lr_mark_ref): Likewise.
17220 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17221 (mark_pseudo_reg_dead): Likewise.
17222 (mark_ref_dead): Likewise.
17223 * reginfo.c (init_subregs_of_mode): Likewise.
17224 * sched-deps.c (sched_analyze_1): Likewise.
17225 * df-scan.c (df_def_record_1): Likewise.
17226 (df_uses_record): Likewise.
17227 (df_read_modify_subreg_p): Remove in favor of...
17228 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17229 const_rtx instead of an rtx.
17231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17232 Alan Hayward <alan.hayward@arm.com>
17233 David Sherwood <david.sherwood@arm.com>
17235 * rtl.h (partial_subreg_p): New function.
17236 * caller-save.c (save_call_clobbered_regs): Use it.
17237 * calls.c (expand_call): Likewise.
17238 * combine.c (combinable_i3pat): Likewise.
17239 (simplify_set): Likewise.
17240 (make_extraction): Likewise.
17241 (make_compound_operation_int): Likewise.
17242 (gen_lowpart_or_truncate): Likewise.
17243 (force_to_mode): Likewise.
17244 (make_field_assignment): Likewise.
17245 (reg_truncated_to_mode): Likewise.
17246 (record_truncated_value): Likewise.
17247 (move_deaths): Likewise.
17248 * cse.c (record_jump_cond): Likewise.
17249 (cse_insn): Likewise.
17250 * cselib.c (cselib_lookup_1): Likewise.
17251 * expmed.c (extract_bit_field_using_extv): Likewise.
17252 * function.c (assign_parm_setup_reg): Likewise.
17253 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17254 * ira-build.c (create_insn_allocnos): Likewise.
17255 * lra-coalesce.c (merge_pseudos): Likewise.
17256 * lra-constraints.c (match_reload): Likewise.
17257 (simplify_operand_subreg): Likewise.
17258 (curr_insn_transform): Likewise.
17259 * lra-lives.c (process_bb_lives): Likewise.
17260 * lra.c (new_insn_reg): Likewise.
17261 (lra_substitute_pseudo): Likewise.
17262 * regcprop.c (mode_change_ok): Likewise.
17263 (maybe_mode_change): Likewise.
17264 (copyprop_hardreg_forward_1): Likewise.
17265 * reload.c (push_reload): Likewise.
17266 (find_reloads): Likewise.
17267 (find_reloads_subreg_address): Likewise.
17268 * reload1.c (alter_reg): Likewise.
17269 (eliminate_regs_1): Likewise.
17270 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17272 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17274 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17275 back to if statements, including unpack.
17277 2017-08-30 Martin Liska <mliska@suse.cz>
17279 PR inline-asm/82001
17280 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17282 (func_checker::compare_asm_inputs_outputs): ... this function.
17283 (func_checker::compare_gimple_asm): Use the function to compare
17284 also ASM constrains.
17285 * ipa-icf-gimple.h: Rename the function.
17287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17288 Alan Hayward <alan.hayward@arm.com>
17289 David Sherwood <david.sherwood@arm.com>
17291 * coretypes.h (complex_mode): New type.
17292 * gdbhooks.py (build_pretty_printer): Handle it.
17293 * machmode.h (complex_mode): New class.
17294 (complex_mode::includes_p): New function.
17295 (is_complex_int_mode): Likewise.
17296 (is_complex_float_mode): Likewise.
17297 * genmodes.c (get_mode_class): Handle complex mode classes.
17298 * function.c (expand_function_end): Use is_complex_int_mode.
17300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17301 Alan Hayward <alan.hayward@arm.com>
17302 David Sherwood <david.sherwood@arm.com>
17304 * coretypes.h (scalar_mode_pod): New typedef.
17305 * gdbhooks.py (build_pretty_printer): Handle it.
17306 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17307 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17308 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17309 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17310 as_a <scalar_mode>.
17312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17313 Alan Hayward <alan.hayward@arm.com>
17314 David Sherwood <david.sherwood@arm.com>
17316 * machmode.h (mode_for_vector): Take a scalar_mode instead
17318 * stor-layout.c (mode_for_vector): Likewise.
17319 * explow.c (promote_mode): Use as_a <scalar_mode>.
17320 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17323 Alan Hayward <alan.hayward@arm.com>
17324 David Sherwood <david.sherwood@arm.com>
17326 * target.def (preferred_simd_mode): Take a scalar_mode
17327 instead of a machine_mode.
17328 * targhooks.h (default_preferred_simd_mode): Likewise.
17329 * targhooks.c (default_preferred_simd_mode): Likewise.
17330 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17331 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17332 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17333 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17334 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17335 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17336 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17337 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17339 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17340 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17341 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17342 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17343 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17344 * doc/tm.texi: Regenerate.
17345 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17349 Alan Hayward <alan.hayward@arm.com>
17350 David Sherwood <david.sherwood@arm.com>
17352 * target.def (scalar_mode_supported_p): Take a scalar_mode
17353 instead of a machine_mode.
17354 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17355 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17356 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17357 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17358 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17359 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17360 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17361 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17362 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17363 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17364 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17365 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17366 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17367 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17369 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17370 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17371 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17372 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17373 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17375 * doc/tm.texi: Regenerate.
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 (opt_scalar_mode): New typedef.
17382 * gdbhooks.py (build_pretty_printers): Handle it.
17383 * machmode.h (mode_iterator::get_2xwider): Add overload for
17385 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17387 * expr.c (convert_mode_scalar): Likewise.
17388 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17389 * optabs.c (expand_float): Likewise.
17390 (expand_fix): Likewise.
17391 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17394 Alan Hayward <alan.hayward@arm.com>
17395 David Sherwood <david.sherwood@arm.com>
17397 * optabs.c (expand_float): Explicitly check for scalars before
17398 using a branching expansion.
17399 (expand_fix): Likewise.
17401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17402 Alan Hayward <alan.hayward@arm.com>
17403 David Sherwood <david.sherwood@arm.com>
17405 * expr.c (convert_mode): Split scalar handling out into...
17406 (convert_mode_scalar): ...this new function. Treat the modes
17409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17410 Alan Hayward <alan.hayward@arm.com>
17411 David Sherwood <david.sherwood@arm.com>
17413 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17415 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17418 Alan Hayward <alan.hayward@arm.com>
17419 David Sherwood <david.sherwood@arm.com>
17421 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17422 rather than a machine_mode.
17423 (fixed_from_string): Likewise.
17424 (fixed_convert): Likewise.
17425 (fixed_convert_from_int): Likewise.
17426 (fixed_convert_from_real): Likewise.
17427 (real_convert_from_fixed): Likewise.
17428 * fixed-value.c (fixed_from_double_int): Likewise.
17429 (fixed_from_string): Likewise.
17430 (fixed_convert): Likewise.
17431 (fixed_convert_from_int): Likewise.
17432 (fixed_convert_from_real): Likewise.
17433 (real_convert_from_fixed): Likewise.
17434 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17437 Alan Hayward <alan.hayward@arm.com>
17438 David Sherwood <david.sherwood@arm.com>
17440 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17441 of separate mode class checks. Do not allow vector modes here.
17442 (immed_wide_int_const): Use as_a <scalar_mode>.
17443 * explow.c (trunc_int_for_mode): Likewise.
17444 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17445 (wi::shwi): Likewise.
17446 (wi::min_value): Likewise.
17447 (wi::max_value): Likewise.
17448 * dwarf2out.c (loc_descriptor): Likewise.
17449 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17450 for CONST_WIDE_INT.
17452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17453 Alan Hayward <alan.hayward@arm.com>
17454 David Sherwood <david.sherwood@arm.com>
17456 * tree.h (SCALAR_TYPE_MODE): New macro.
17457 * expr.c (expand_expr_addr_expr_1): Use it.
17458 (expand_expr_real_2): Likewise.
17459 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17460 (fold_convert_const_fixed_from_int): Likewise.
17461 (fold_convert_const_fixed_from_real): Likewise.
17462 (native_encode_fixed): Likewise
17463 (native_encode_complex): Likewise
17464 (native_encode_vector): Likewise.
17465 (native_interpret_fixed): Likewise.
17466 (native_interpret_real): Likewise.
17467 (native_interpret_complex): Likewise.
17468 (native_interpret_vector): Likewise.
17469 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17470 (simd_clone_adjust_argument_types): Likewise.
17471 (simd_clone_init_simd_arrays): Likewise.
17472 (simd_clone_adjust): Likewise.
17473 * stor-layout.c (layout_type): Likewise.
17474 * tree.c (build_minus_one_cst): Likewise.
17475 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17476 * tree-inline.c (estimate_move_cost): Likewise.
17477 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17478 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17479 (vectorizable_reduction): Likewise.
17480 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17481 (vect_recog_mixed_size_cond_pattern): Likewise.
17482 (check_bool_pattern): Likewise.
17483 (adjust_bool_pattern): Likewise.
17484 (search_type_for_mask_1): Likewise.
17485 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17486 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17487 (vectorizable_load): Likewise.
17488 (vectorizable_store): Likewise.
17489 * ubsan.c (ubsan_encode_value): Likewise.
17490 * varasm.c (output_constant): Likewise.
17492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17493 Alan Hayward <alan.hayward@arm.com>
17494 David Sherwood <david.sherwood@arm.com>
17496 * coretypes.h (scalar_mode): New class.
17497 * machmode.h (scalar_mode): Likewise.
17498 (scalar_mode::includes_p): New function.
17499 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17500 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17501 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17502 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17503 * expmed.c (store_bit_field_1): Likewise.
17504 (extract_bit_field_1): Likewise.
17505 * expr.c (write_complex_part): Likewise.
17506 (read_complex_part): Likewise.
17507 (emit_move_complex_push): Likewise.
17508 (expand_expr_real_2): Likewise.
17509 * function.c (assign_parm_setup_reg): Likewise.
17510 (assign_parms_unsplit_complex): Likewise.
17511 * optabs.c (expand_binop): Likewise.
17512 * rtlanal.c (subreg_get_info): Likewise.
17513 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17514 * varasm.c (output_constant_pool_2): Likewise.
17516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17517 Alan Hayward <alan.hayward@arm.com>
17518 David Sherwood <david.sherwood@arm.com>
17520 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17522 (expmed_mult_highpart_optab): Likewise.
17523 (expmed_mult_highpart): Likewise.
17525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17526 Alan Hayward <alan.hayward@arm.com>
17527 David Sherwood <david.sherwood@arm.com>
17529 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17530 instead of a machine_mode.
17531 (builtin_memset_read_str): Likewise.
17532 * builtins.c (c_readstr): Likewise.
17533 (builtin_memcpy_read_str): Likewise.
17534 (builtin_strncpy_read_str): Likewise.
17535 (builtin_memset_read_str): Likewise.
17536 (builtin_memset_gen_str): Likewise.
17537 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17538 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17539 instead of a machine_mode.
17540 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17542 (make_extraction): Likewise.
17543 (try_widen_shift_mode): Take and return scalar_int_modes instead
17545 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17546 a scalar_int_mode instead of a machine_mode.
17547 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17548 (avr_addr_space_pointer_mode): Likewise.
17549 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17550 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17551 (msp430_unwind_word_mode): Likewise.
17552 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17553 (spu_addr_space_pointer_mode): Likewise.
17554 (spu_addr_space_address_mode): Likewise.
17555 (spu_libgcc_cmp_return_mode): Likewise.
17556 (spu_libgcc_shift_count_mode): Likewise.
17557 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17558 (rl78_addr_space_pointer_mode): Likewise.
17559 (fl78_unwind_word_mode): Likewise.
17560 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17562 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17563 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17564 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17565 (mips_valid_pointer_mode): Likewise.
17566 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17567 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17568 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17570 (ft32_addr_space_address_mode): Likewise.
17571 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17572 scalar_int_mode instead of a machine_mode.
17573 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17575 (m32c_addr_space_address_mode): Likewise.
17576 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17577 (rs6000_eh_return_filter_mode): Likewise.
17578 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17579 (rs6000_eh_return_filter_mode): Likewise.
17580 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17581 (s390_libgcc_shift_count_mode): Likewise.
17582 (s390_unwind_word_mode): Likewise.
17583 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17585 * target.def (mode_rep_extended): Likewise.
17586 (valid_pointer_mode): Likewise.
17587 (addr_space.valid_pointer_mode): Likewise.
17588 (eh_return_filter_mode): Return a scalar_int_mode rather than
17590 (libgcc_cmp_return_mode): Likewise.
17591 (libgcc_shift_count_mode): Likewise.
17592 (unwind_word_mode): Likewise.
17593 (addr_space.pointer_mode): Likewise.
17594 (addr_space.address_mode): Likewise.
17595 * doc/tm.texi: Regenerate.
17596 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17598 (do_jump): Use scalar_int_mode for local variables.
17599 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17600 rather than a machine_mode.
17601 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17602 (scompare_loc_descriptor_wide): Likewise.
17603 (scompare_loc_descriptor_narrow): Likewise.
17604 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17606 * except.c (sjlj_emit_dispatch_table): Likewise.
17607 (expand_builtin_eh_copy_values): Likewise.
17608 * explow.c (convert_memory_address_addr_space_1): Likewise.
17609 Take a scalar_int_mode rather than a machine_mode.
17610 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17611 than a machine_mode.
17612 (memory_address_addr_space): Use scalar_int_mode for local variables.
17613 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17614 rather than a machine_mode.
17615 * expmed.c (mask_rtx): Likewise.
17616 (init_expmed_one_conv): Likewise.
17617 (expand_mult_highpart_adjust): Likewise.
17618 (extract_high_half): Likewise.
17619 (expmed_mult_highpart_optab): Likewise.
17620 (expmed_mult_highpart): Likewise.
17621 (expand_smod_pow2): Likewise.
17622 (expand_sdiv_pow2): Likewise.
17623 (emit_store_flag_int): Likewise.
17624 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17626 (extract_low_bits): Likewise.
17627 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17629 * expr.c (pieces_addr::adjust): Likewise.
17630 (can_store_by_pieces): Likewise.
17631 (store_by_pieces): Likewise.
17632 (clear_by_pieces_1): Likewise.
17633 (expand_expr_addr_expr_1): Likewise.
17634 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17635 (expand_expr_real_1): Likewise.
17636 (try_casesi): Likewise.
17637 * final.c (shorten_branches): Likewise.
17638 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17639 type of "mode" to machine_mode.
17640 * internal-fn.c (expand_arith_overflow_result_store): Take a
17641 scalar_int_mode rather than a machine_mode.
17642 (expand_mul_overflow): Use scalar_int_mode for local variables.
17643 * loop-doloop.c (doloop_modify): Likewise.
17644 (doloop_optimize): Likewise.
17645 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17646 than a machine_mode.
17647 (expand_doubleword_shift_condmove): Likewise.
17648 (expand_doubleword_shift): Likewise.
17649 (expand_doubleword_clz): Likewise.
17650 (expand_doubleword_popcount): Likewise.
17651 (expand_doubleword_parity): Likewise.
17652 (expand_absneg_bit): Use scalar_int_mode for local variables.
17653 (prepare_float_lib_cmp): Likewise.
17654 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17655 rather than a machine_mode.
17656 (convert_memory_address_addr_space): Likewise.
17657 (get_mode_bounds): Likewise.
17658 (get_address_mode): Return a scalar_int_mode rather than a
17660 * rtlanal.c (get_address_mode): Likewise.
17661 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17662 than a machine_mode.
17663 * targhooks.c (default_mode_rep_extended): Likewise.
17664 (default_valid_pointer_mode): Likewise.
17665 (default_addr_space_valid_pointer_mode): Likewise.
17666 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17667 than a machine_mode.
17668 (default_libgcc_cmp_return_mode): Likewise.
17669 (default_libgcc_shift_count_mode): Likewise.
17670 (default_unwind_word_mode): Likewise.
17671 (default_addr_space_pointer_mode): Likewise.
17672 (default_addr_space_address_mode): Likewise.
17673 * targhooks.h (default_eh_return_filter_mode): Likewise.
17674 (default_libgcc_cmp_return_mode): Likewise.
17675 (default_libgcc_shift_count_mode): Likewise.
17676 (default_unwind_word_mode): Likewise.
17677 (default_addr_space_pointer_mode): Likewise.
17678 (default_addr_space_address_mode): Likewise.
17679 (default_mode_rep_extended): Take a scalar_int_mode rather than
17681 (default_valid_pointer_mode): Likewise.
17682 (default_addr_space_valid_pointer_mode): Likewise.
17683 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17685 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17686 rather than a machine_mode.
17687 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17688 for local variables.
17689 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17690 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17691 than a machine_mode.
17693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17694 Alan Hayward <alan.hayward@arm.com>
17695 David Sherwood <david.sherwood@arm.com>
17697 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17698 the mode argument to scalar_int_mode.
17699 (do_jump_by_parts_zero_rtx): Likewise.
17700 (do_jump_by_parts_equality_rtx): Likewise.
17701 (do_jump_by_parts_greater): Take a mode argument.
17702 (do_jump_by_parts_equality): Likewise.
17703 (do_jump_1): Update calls accordingly.
17705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17706 Alan Hayward <alan.hayward@arm.com>
17707 David Sherwood <david.sherwood@arm.com>
17709 * is-a.h (safe_dyn_cast): New function.
17710 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17711 (jump_table_for_label): Likewise.
17712 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17713 instead of an rtx_insn *.
17714 (shorten_branches): Use dyn_cast instead of LABEL_P and
17715 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17716 rtx_jump_table_data::get_data_mode.
17717 (final_scan_insn): Likewise.
17719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17720 Alan Hayward <alan.hayward@arm.com>
17721 David Sherwood <david.sherwood@arm.com>
17723 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17724 trying to combine a full-register integer set with a subreg
17727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17728 Alan Hayward <alan.hayward@arm.com>
17729 David Sherwood <david.sherwood@arm.com>
17731 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17732 that is always either address_mode or pointer_mode.
17734 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17735 Alan Hayward <alan.hayward@arm.com>
17736 David Sherwood <david.sherwood@arm.com>
17738 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17739 when the two are known to be equal.
17741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17742 Alan Hayward <alan.hayward@arm.com>
17743 David Sherwood <david.sherwood@arm.com>
17745 * simplify-rtx.c (simplify_const_unary_operation): Use
17746 is_a <scalar_int_mode> instead of checking for a nonzero
17747 precision. Forcibly convert op_mode to a scalar_int_mode
17748 in that case. More clearly differentiate the operand and
17749 result modes and use the former when deciding what the value
17750 of a count-bits operation should be. Use is_int_mode instead
17751 of checking for a MODE_INT. Remove redundant check for whether
17752 this mode has a zero precision.
17754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17755 Alan Hayward <alan.hayward@arm.com>
17756 David Sherwood <david.sherwood@arm.com>
17758 * optabs.c (widen_leading): Change the type of the mode argument
17759 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17760 (widen_bswap): Likewise.
17761 (expand_parity): Likewise.
17762 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17763 (expand_ffs): Likewise.
17764 (epand_unop): Check for scalar integer modes before calling the
17767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17768 Alan Hayward <alan.hayward@arm.com>
17769 David Sherwood <david.sherwood@arm.com>
17771 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17773 (expand_expr_real_1): Update call accordingly.
17775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17776 Alan Hayward <alan.hayward@arm.com>
17777 David Sherwood <david.sherwood@arm.com>
17779 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17780 value_mode arguments. Use scalar_int_mode internally.
17781 (store_bit_field_1): Rename the new integer mode from imode
17782 to op0_mode and use it instead of GET_MODE (op0). Update calls
17783 to store_split_bit_field, store_bit_field_using_insv and
17784 store_fixed_bit_field.
17785 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17786 Use scalar_int_mode internally. Use a bit count rather than a mode
17787 when calculating the largest bit size for get_best_mode.
17788 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17789 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17790 Remove assertion that OP0 has a scalar integer mode.
17791 (store_split_bit_field): Add op0_mode and value_mode arguments.
17792 Update calls to extract_fixed_bit_field.
17793 (extract_bit_field_using_extv): Add an op0_mode argument.
17794 Use scalar_int_mode internally.
17795 (extract_bit_field_1): Rename the new integer mode from imode to
17796 op0_mode and use it instead of GET_MODE (op0). Update calls to
17797 extract_split_bit_field, extract_bit_field_using_extv and
17798 extract_fixed_bit_field.
17799 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17800 to extract_split_bit_field and extract_fixed_bit_field_1.
17801 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17802 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17803 on the target mode.
17804 (extract_split_bit_field): Add an op0_mode argument. Update call
17805 to extract_fixed_bit_field.
17807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17808 Alan Hayward <alan.hayward@arm.com>
17809 David Sherwood <david.sherwood@arm.com>
17811 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17812 * explow.c (hard_function_value): Likewise.
17813 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17814 convert_to_mode call outside the loop.
17815 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17816 for the mode iterator. Require the mode specified by max_pieces
17818 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17820 (copy_blkmode_to_reg): Likewise.
17821 (set_storage_via_setmem): Likewise.
17822 * optabs.c (prepare_cmp_insn): Likewise.
17823 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17824 * stor-layout.c (finish_bitfield_representative): Likewise.
17826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17827 Alan Hayward <alan.hayward@arm.com>
17828 David Sherwood <david.sherwood@arm.com>
17830 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17831 * expr.c (convert_move): Use them.
17832 (convert_modes): Likewise.
17833 (store_expr_with_bounds): Likewise.
17835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17836 Alan Hayward <alan.hayward@arm.com>
17837 David Sherwood <david.sherwood@arm.com>
17839 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17840 parameter for the mode of "x". Remove the "known_x", "known_mode"
17841 and "known_ret" arguments. Change the type of the mode argument
17842 to scalar_int_mode.
17843 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17844 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17845 (reg_num_sign_bit_copies_for_combine): Likewise.
17846 * rtlanal.c (nonzero_bits1): Likewise.
17847 (num_sign_bit_copies1): Likewise.
17848 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17849 (reg_num_sign_bit_copies_general): Likewise.
17850 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17851 (reg_nonzero_bits_general): Likewise.
17853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17854 Alan Hayward <alan.hayward@arm.com>
17855 David Sherwood <david.sherwood@arm.com>
17857 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17858 than in subroutines. Return 1 for non-integer modes.
17859 (cached_num_sign_bit_copies): Change the type of the mode parameter
17860 to scalar_int_mode.
17861 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17862 classes. Handle CONST_INT_P first and then check whether X also
17863 has a scalar integer mode. Check the same thing for inner registers
17864 of a SUBREG and for values that are being extended or truncated.
17866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17867 Alan Hayward <alan.hayward@arm.com>
17868 David Sherwood <david.sherwood@arm.com>
17870 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17871 in subroutines. Return the mode mask for non-integer modes.
17872 (cached_nonzero_bits): Change the type of the mode parameter
17873 to scalar_int_mode.
17874 (nonzero_bits1): Likewise. Remove early exit for other mode
17875 classes. Handle CONST_INT_P first and then check whether X
17876 also has a scalar integer mode.
17878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17879 Alan Hayward <alan.hayward@arm.com>
17880 David Sherwood <david.sherwood@arm.com>
17882 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17883 Return a scalar_int_mode and assert that the size is greater than
17885 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17886 (op_by_pieces_d::op_by_pieces_d): Likewise.
17887 (op_by_pieces_d::run): Likewise.
17888 (can_store_by_pieces): Likewise.
17890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17891 Alan Hayward <alan.hayward@arm.com>
17892 David Sherwood <david.sherwood@arm.com>
17894 * combine.c (extract_left_shift): Add a mode argument and update
17896 (make_compound_operation_int): Change the type of the mode parameter
17897 to scalar_int_mode and update the call to extract_left_shift.
17899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17900 Alan Hayward <alan.hayward@arm.com>
17901 David Sherwood <david.sherwood@arm.com>
17903 * combine.c (simplify_and_const_int): Change the type of the mode
17904 parameter to scalar_int_mode.
17905 (simplify_and_const_int_1): Likewise. Update recursive call.
17907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17908 Alan Hayward <alan.hayward@arm.com>
17909 David Sherwood <david.sherwood@arm.com>
17911 * combine.c (simplify_compare_const): Check that the mode is a
17912 scalar_int_mode (rather than VOIDmode) before testing its
17914 (simplify_comparison): Move COMPARISON_P handling out of the
17915 loop and restrict the latter part of the loop to scalar_int_modes.
17916 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17917 and when considering SUBREG_REGs. Use is_int_mode instead of
17918 checking GET_MODE_CLASS against MODE_INT.
17920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17921 Alan Hayward <alan.hayward@arm.com>
17922 David Sherwood <david.sherwood@arm.com>
17924 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17925 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17926 shift_unit_mode and for modes involved in scalar shifts.
17928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17929 Alan Hayward <alan.hayward@arm.com>
17930 David Sherwood <david.sherwood@arm.com>
17932 * combine.c (force_int_to_mode): New function, split out from...
17933 (force_to_mode): ...here. Keep xmode up-to-date and use it
17934 instead of GET_MODE (x).
17936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17937 Alan Hayward <alan.hayward@arm.com>
17938 David Sherwood <david.sherwood@arm.com>
17940 * optabs-query.h (extraction_insn::struct_mode): Change type to
17941 opt_scalar_int_mode and update comment.
17942 (extraction_insn::field_mode): Change type to scalar_int_mode.
17943 (extraction_insn::pos_mode): Likewise.
17944 * combine.c (make_extraction): Update accordingly.
17945 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17946 (get_optab_extraction_insn): Likewise.
17947 * recog.c (simplify_while_replacing): Likewise.
17948 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17949 parameter to opt_scalar_int_mode.
17951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17952 Alan Hayward <alan.hayward@arm.com>
17953 David Sherwood <david.sherwood@arm.com>
17955 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17956 to a scalar_int_mode instead of a machine_mode.
17957 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17958 (get_best_mode): Return a boolean and use a pointer argument to store
17959 the selected mode. Replace the limit mode parameter with a bit limit.
17960 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17961 for the values returned by bit_field_mode_iterator::next_mode.
17962 (store_bit_field): Update call to get_best_mode.
17963 (store_fixed_bit_field): Likewise.
17964 (extract_fixed_bit_field): Likewise.
17965 * expr.c (optimize_bitfield_assignment_op): Likewise.
17966 * fold-const.c (optimize_bit_field_compare): Likewise.
17967 (fold_truth_andor_1): Likewise.
17968 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17969 Update for new type of m_mode.
17970 (get_best_mode): As above.
17972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17973 Alan Hayward <alan.hayward@arm.com>
17974 David Sherwood <david.sherwood@arm.com>
17976 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17977 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17978 (store_bit_field): Check is_a <scalar_int_mode> before calling
17979 strict_volatile_bitfield_p.
17980 (extract_bit_field): Likewise.
17982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17983 Alan Hayward <alan.hayward@arm.com>
17984 David Sherwood <david.sherwood@arm.com>
17986 * target.def (cstore_mode): Return a scalar_int_mode.
17987 * doc/tm.texi: Regenerate.
17988 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17989 * targhooks.h (default_cstore_mode): Likewise.
17990 * targhooks.c (default_cstore_mode): Likewise, using a forced
17992 * expmed.c (emit_cstore): Expect the target of the cstore to be
17995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17996 Alan Hayward <alan.hayward@arm.com>
17997 David Sherwood <david.sherwood@arm.com>
17999 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18001 (niter_desc): Likewise mode.
18002 (iv_analyze): Add a mode parameter.
18004 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18005 and change its type to scalar_int_mode.
18006 * loop-iv.c: Update commentary at head of file.
18007 (iv_constant): Pass the mode paraeter before the rtx it describes
18008 and change its type to scalar_int_mode. Remove VOIDmode handling.
18009 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18010 (iv_extend): Likewise.
18011 (shorten_into_mode): Likewise.
18012 (iv_add): Use scalar_int_mode.
18013 (iv_mult): Likewise.
18014 (iv_shift): Likewise.
18015 (canonicalize_iv_subregs): Likewise.
18016 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18017 it describes and change its mode to scalar_int_mode. Also change
18018 the type of the returned inner_mode to scalar_int_mode.
18019 (get_biv_step): Likewise, turning outer_mode from a pointer
18020 into a direct parameter. Update call to get_biv_step_1.
18021 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18022 iv_constant and get_biv_step.
18023 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18024 and change its type to scalar_int_mode. Don't initialise iv->mode
18025 to VOIDmode and remove later checks for its still being VOIDmode.
18026 Update calls to iv_analyze_op and iv_analyze_expr. Check
18027 is_a <scalar_int_mode> when changing the mode under consideration.
18028 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18029 Update call to iv_analyze_expr.
18030 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18031 inner register is not also a scalar_int_mode. Update call to
18033 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18034 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18035 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18036 separate mode class checks. Update calls to iv_analyze. Remove
18037 fix-up of VOIDmodes after iv_analyze_biv.
18038 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18039 don't have a scalar_int_mode. Update call to biv_p.
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 * cfgexpand.c (convert_debug_memory_address): Use
18046 as_a <scalar_int_mode>.
18047 * combine.c (expand_compound_operation): Likewise.
18048 (make_extraction): Likewise.
18049 (change_zero_ext): Likewise.
18050 (simplify_comparison): Likewise.
18051 * cse.c (cse_insn): Likewise.
18052 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18053 (mem_loc_descriptor): Likewise.
18054 (loc_descriptor): Likewise.
18055 * expmed.c (init_expmed_one_mode): Likewise.
18056 (synth_mult): Likewise.
18057 (emit_store_flag_1): Likewise.
18058 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18059 of a comparison with size.
18060 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18061 (reduce_to_bit_field_precision): Likewise.
18062 * function.c (expand_function_end): Likewise.
18063 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18064 * loop-doloop.c (doloop_modify): Likewise.
18065 * optabs.c (expand_binop): Likewise.
18066 (expand_unop): Likewise.
18067 (expand_copysign_absneg): Likewise.
18068 (prepare_cmp_insn): Likewise.
18069 (maybe_legitimize_operand): Likewise.
18070 * recog.c (const_scalar_int_operand): Likewise.
18071 * rtlanal.c (get_address_mode): Likewise.
18072 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18073 (simplify_cond_clz_ctz): Likewise.
18074 * tree-nested.c (get_nl_goto_field): Likewise.
18075 * tree.c (build_vector_type_for_mode): Likewise.
18076 * var-tracking.c (use_narrower_mode): Likewise.
18078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18079 Alan Hayward <alan.hayward@arm.com>
18080 David Sherwood <david.sherwood@arm.com>
18082 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18083 * builtins.c (expand_builtin_signbit): Use it.
18084 * cfgexpand.c (expand_debug_expr): Likewise.
18085 * dojump.c (do_jump): Likewise.
18086 (do_compare_and_jump): Likewise.
18087 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18088 * expmed.c (make_tree): Likewise.
18089 * expr.c (expand_expr_real_2): Likewise.
18090 (expand_expr_real_1): Likewise.
18091 (try_casesi): Likewise.
18092 * fold-const-call.c (fold_const_call_ss): Likewise.
18093 * fold-const.c (unextend): Likewise.
18094 (extract_muldiv_1): Likewise.
18095 (fold_single_bit_test): Likewise.
18096 (native_encode_int): Likewise.
18097 (native_encode_string): Likewise.
18098 (native_interpret_int): Likewise.
18099 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18100 * internal-fn.c (expand_addsub_overflow): Likewise.
18101 (expand_neg_overflow): Likewise.
18102 (expand_mul_overflow): Likewise.
18103 (expand_arith_overflow): Likewise.
18104 * match.pd: Likewise.
18105 * stor-layout.c (layout_type): Likewise.
18106 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18107 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18108 * tree-ssanames.c (get_range_info): Likewise.
18109 * tree-switch-conversion.c (array_value_type) Likewise.
18110 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18111 (vect_recog_divmod_pattern): Likewise.
18112 (vect_recog_mixed_size_cond_pattern): Likewise.
18113 * tree-vrp.c (extract_range_basic): Likewise.
18114 (simplify_float_conversion_using_ranges): Likewise.
18115 * tree.c (int_fits_type_p): Likewise.
18116 * ubsan.c (instrument_bool_enum_load): Likewise.
18117 * varasm.c (mergeable_string_section): Likewise.
18118 (narrowing_initializer_constant_valid_p): Likewise.
18119 (output_constant): Likewise.
18121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18122 Alan Hayward <alan.hayward@arm.com>
18123 David Sherwood <david.sherwood@arm.com>
18125 * machmode.h (NARROWEST_INT_MODE): New macro.
18126 * expr.c (alignment_for_piecewise_move): Use it instead of
18127 GET_CLASS_NARROWEST_MODE (MODE_INT).
18128 (push_block): Likewise.
18129 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18131 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18134 Alan Hayward <alan.hayward@arm.com>
18135 David Sherwood <david.sherwood@arm.com>
18137 * postreload.c (move2add_valid_value_p): Change the type of the
18138 mode parameter to scalar_int_mode.
18139 (move2add_use_add2_insn): Add a mode parameter and use it instead
18141 (move2add_use_add3_insn): Likewise.
18142 (reload_cse_move2add): Update accordingly.
18144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18145 Alan Hayward <alan.hayward@arm.com>
18146 David Sherwood <david.sherwood@arm.com>
18148 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18150 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18151 * optabs.c (expand_unop): Likewise.
18153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18154 Alan Hayward <alan.hayward@arm.com>
18155 David Sherwood <david.sherwood@arm.com>
18157 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18158 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18159 (popcount_loc_descriptor): Likewise.
18160 (bswap_loc_descriptor): Likewise.
18161 (rotate_loc_descriptor): Likewise.
18162 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18163 calling the functions above.
18165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18166 Alan Hayward <alan.hayward@arm.com>
18167 David Sherwood <david.sherwood@arm.com>
18169 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18171 (try_combine): Likewise.
18172 (simplify_if_then_else): Likewise.
18173 * cse.c (cse_insn): Likewise.
18174 * dwarf2out.c (mem_loc_descriptor): Likewise.
18175 * emit-rtl.c (gen_lowpart_common): Likewise.
18176 * simplify-rtx.c (simplify_truncation): Likewise.
18177 (simplify_binary_operation_1): Likewise.
18178 (simplify_const_relational_operation): Likewise.
18179 (simplify_ternary_operation): Likewise.
18180 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18183 Alan Hayward <alan.hayward@arm.com>
18184 David Sherwood <david.sherwood@arm.com>
18186 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18187 * reload.c (push_reload): Likewise.
18188 (find_reloads): Likewise.
18190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18191 Alan Hayward <alan.hayward@arm.com>
18192 David Sherwood <david.sherwood@arm.com>
18194 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18195 (make_compound_operation_int): Likewise.
18196 (change_zero_ext): Likewise.
18197 * expr.c (convert_move): Likewise.
18198 (convert_modes): Likewise.
18199 * fwprop.c (forward_propagate_subreg): Likewise.
18200 * loop-iv.c (get_biv_step_1): Likewise.
18201 * optabs.c (widen_operand): Likewise.
18202 * postreload.c (move2add_valid_value_p): Likewise.
18203 * recog.c (simplify_while_replacing): Likewise.
18204 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18205 (simplify_binary_operation_1): Likewise. Remove redundant
18206 mode equality check.
18208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18209 Alan Hayward <alan.hayward@arm.com>
18210 David Sherwood <david.sherwood@arm.com>
18212 * combine.c (combine_simplify_rtx): Add checks for
18213 is_a <scalar_int_mode>.
18214 (simplify_if_then_else): Likewise.
18215 (make_field_assignment): Likewise.
18216 (simplify_comparison): Likewise.
18217 * ifcvt.c (noce_try_bitop): Likewise.
18218 * loop-invariant.c (canonicalize_address_mult): Likewise.
18219 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18222 Alan Hayward <alan.hayward@arm.com>
18223 David Sherwood <david.sherwood@arm.com>
18225 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18226 is_a <scalar_int_mode> instead of != BLKmode.
18228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18229 Alan Hayward <alan.hayward@arm.com>
18230 David Sherwood <david.sherwood@arm.com>
18232 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18233 instead of != VOIDmode.
18234 * combine.c (if_then_else_cond): Likewise.
18235 (change_zero_ext): Likewise.
18236 * dwarf2out.c (mem_loc_descriptor): Likewise.
18237 (loc_descriptor): Likewise.
18238 * rtlanal.c (canonicalize_condition): Likewise.
18239 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18242 Alan Hayward <alan.hayward@arm.com>
18243 David Sherwood <david.sherwood@arm.com>
18245 * simplify-rtx.c (simplify_binary_operation_1): Use
18246 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18249 Alan Hayward <alan.hayward@arm.com>
18250 David Sherwood <david.sherwood@arm.com>
18252 * wide-int.h (int_traits<unsigned char>) New class.
18253 (int_traits<unsigned short>) Likewise.
18254 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18255 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18257 * combine.c (set_nonzero_bits_and_sign_copies): Use
18258 is_a <scalar_int_mode>.
18259 (find_split_point): Likewise.
18260 (combine_simplify_rtx): Likewise.
18261 (simplify_logical): Likewise.
18262 (expand_compound_operation): Likewise.
18263 (expand_field_assignment): Likewise.
18264 (make_compound_operation): Likewise.
18265 (extended_count): Likewise.
18266 (change_zero_ext): Likewise.
18267 (simplify_comparison): Likewise.
18268 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18269 (ucompare_loc_descriptor): Likewise.
18270 (minmax_loc_descriptor): Likewise.
18271 (mem_loc_descriptor): Likewise.
18272 (loc_descriptor): Likewise.
18273 * expmed.c (init_expmed_one_mode): Likewise.
18274 * lra-constraints.c (lra_constraint_offset): Likewise.
18275 * optabs.c (prepare_libcall_arg): Likewise.
18276 * postreload.c (move2add_note_store): Likewise.
18277 * reload.c (operands_match_p): Likewise.
18278 * rtl.h (load_extend_op): Likewise.
18279 * rtlhooks.c (gen_lowpart_general): Likewise.
18280 * simplify-rtx.c (simplify_truncation): Likewise.
18281 (simplify_unary_operation_1): Likewise.
18282 (simplify_binary_operation_1): Likewise.
18283 (simplify_const_binary_operation): Likewise.
18284 (simplify_const_relational_operation): Likewise.
18285 (simplify_subreg): Likewise.
18286 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18287 * var-tracking.c (adjust_mems): Likewise.
18288 (prepare_call_arguments): Likewise.
18290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18291 Alan Hayward <alan.hayward@arm.com>
18292 David Sherwood <david.sherwood@arm.com>
18294 * machmode.h (is_int_mode): New fuction.
18295 * combine.c (find_split_point): Use it.
18296 (combine_simplify_rtx): Likewise.
18297 (simplify_if_then_else): Likewise.
18298 (simplify_set): Likewise.
18299 (simplify_shift_const_1): Likewise.
18300 (simplify_comparison): Likewise.
18301 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18302 * cse.c (notreg_cost): Likewise.
18303 (cse_insn): Likewise.
18304 * cselib.c (cselib_lookup_1): Likewise.
18305 * dojump.c (do_jump_1): Likewise.
18306 (do_compare_rtx_and_jump): Likewise.
18307 * dse.c (get_call_args): Likewise.
18308 * dwarf2out.c (rtl_for_decl_init): Likewise.
18309 (native_encode_initializer): Likewise.
18310 * expmed.c (emit_store_flag_1): Likewise.
18311 (emit_store_flag): Likewise.
18312 * expr.c (convert_modes): Likewise.
18313 (store_field): Likewise.
18314 (expand_expr_real_1): Likewise.
18315 * fold-const.c (fold_read_from_constant_string): Likewise.
18316 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18317 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18318 * optabs.c (expand_binop): Likewise.
18319 (expand_unop): Likewise.
18320 (expand_abs_nojump): Likewise.
18321 (expand_one_cmpl_abs_nojump): Likewise.
18322 * simplify-rtx.c (mode_signbit_p): Likewise.
18323 (val_signbit_p): Likewise.
18324 (val_signbit_known_set_p): Likewise.
18325 (val_signbit_known_clear_p): Likewise.
18326 (simplify_relational_operation_1): Likewise.
18327 * tree.c (vector_type_mode): Likewise.
18329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18330 Alan Hayward <alan.hayward@arm.com>
18331 David Sherwood <david.sherwood@arm.com>
18333 * machmode.h (smallest_mode_for_size): Fix formatting.
18334 (smallest_int_mode_for_size): New function.
18335 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18336 instead of smallest_mode_for_size.
18337 * combine.c (make_extraction): Likewise.
18338 * config/arc/arc.c (arc_expand_movmem): Likewise.
18339 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18340 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18341 * config/s390/s390.c (s390_expand_insv): Likewise.
18342 * config/sparc/sparc.c (assign_int_registers): Likewise.
18343 * config/spu/spu.c (spu_function_value): Likewise.
18344 (spu_function_arg): Likewise.
18345 * coverage.c (get_gcov_type): Likewise.
18346 (get_gcov_unsigned_t): Likewise.
18347 * dse.c (find_shift_sequence): Likewise.
18348 * expmed.c (store_bit_field_1): Likewise.
18349 * expr.c (convert_move): Likewise.
18350 (store_field): Likewise.
18351 * internal-fn.c (expand_arith_overflow): Likewise.
18352 * optabs-query.c (get_best_extraction_insn): Likewise.
18353 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18354 * stor-layout.c (layout_type): Likewise.
18355 (initialize_sizetypes): Likewise.
18356 * targhooks.c (default_get_mask_mode): Likewise.
18357 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18360 Alan Hayward <alan.hayward@arm.com>
18361 David Sherwood <david.sherwood@arm.com>
18363 * machmode.h (opt_mode::else_blk): New function.
18364 (int_mode_for_mode): Declare.
18365 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18366 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18368 * cfgexpand.c (expand_debug_expr): Likewise.
18369 * combine.c (gen_lowpart_or_truncate): Likewise.
18370 (gen_lowpart_for_combine): Likewise.
18371 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18372 * config/avr/avr.c (avr_to_int_mode): Likewise.
18373 (avr_out_plus_1): Likewise.
18374 (avr_out_plus): Likewise.
18375 (avr_out_round): Likewise.
18376 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18377 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18378 (s390_expand_vcond): Likewise.
18379 * config/spu/spu.c (spu_split_immediate): Likewise.
18380 (spu_expand_mov): Likewise.
18381 * dse.c (get_stored_val): Likewise.
18382 * expmed.c (store_bit_field_1): Likewise.
18383 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18385 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18386 (extract_low_bits): Likewise.
18387 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18388 handling rather than repeating the check.
18389 (emit_group_store): Likewise.
18390 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18391 * optabs.c (expand_absneg_bit): Likewise.
18392 (expand_copysign_absneg): Likewise.
18393 (expand_copysign_bit): Likewise.
18394 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18395 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18396 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18397 * var-tracking.c (prepare_call_arguments): Likewise.
18398 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18399 int_mode_for_mode instead of mode_for_size.
18400 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18403 Alan Hayward <alan.hayward@arm.com>
18404 David Sherwood <david.sherwood@arm.com>
18406 * machmode.h (int_mode_for_size): New function.
18407 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18408 instead of mode_for_size.
18409 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18411 * combine.c (expand_field_assignment): Use int_mode_for_size
18412 instead of mode_for_size.
18413 (make_extraction): Likewise.
18414 (simplify_shift_const_1): Likewise.
18415 (simplify_comparison): Likewise.
18416 * dojump.c (do_jump): Likewise.
18417 * dwarf2out.c (mem_loc_descriptor): Likewise.
18418 * emit-rtl.c (init_derived_machine_modes): Likewise.
18419 * expmed.c (flip_storage_order): Likewise.
18420 (convert_extracted_bit_field): Likewise.
18421 * expr.c (copy_blkmode_from_reg): Likewise.
18422 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18423 * internal-fn.c (expand_mul_overflow): Likewise.
18424 * lower-subreg.c (simple_move): Likewise.
18425 * optabs-libfuncs.c (init_optabs): Likewise.
18426 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18427 * tree.c (vector_type_mode): Likewise.
18428 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18429 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18430 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18431 * tree-vect-stmts.c (vectorizable_load): Likewise.
18432 (vectorizable_store): Likewise.
18434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18435 Alan Hayward <alan.hayward@arm.com>
18436 David Sherwood <david.sherwood@arm.com>
18438 * coretypes.h (pod_mode): New type.
18439 (scalar_int_mode_pod): New typedef.
18440 * machmode.h (pod_mode): New class.
18441 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18442 * genmodes.c (emit_mode_int_n): Update accordingly.
18443 * lower-subreg.h (target_lower_subreg): Change type to
18444 scalar_int_mode_pod.
18445 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18446 scalar_int_mode_pod.
18448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18449 Alan Hayward <alan.hayward@arm.com>
18450 David Sherwood <david.sherwood@arm.com>
18452 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18453 machine_mode to scalar_int_mode.
18454 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18455 (rs6000_option_override_internal): Remove cast to int.
18456 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18457 machine_mode to scalar_int_mode.
18458 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18459 (rs6000_option_override_internal): Remove cast to int.
18460 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18461 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18463 * config/s390/s390.c (s390_expand_builtin): Likewise.
18464 * coretypes.h (scalar_int_mode): New type.
18465 (opt_scalar_int_mode): New typedef.
18466 * machmode.h (scalar_int_mode): New class.
18467 (scalar_int_mode::includes_p): New function.
18468 (byte_mode): Change type to scalar_int_mode.
18469 (word_mode): Likewise.
18470 (ptr_mode): Likewise.
18471 * emit-rtl.c (byte_mode): Likewise.
18472 (word_mode): Likewise.
18473 (ptr_mode): Likewise.
18474 (init_derived_machine_modes): Update accordingly.
18475 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18476 and MODE_PARTIAL_INT.
18477 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18478 opt_scalar_int_mode.
18480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18481 Alan Hayward <alan.hayward@arm.com>
18482 David Sherwood <david.sherwood@arm.com>
18484 * target.def (libgcc_floating_mode_supported_p): Take a
18486 * doc/tm.texi: Regenerate.
18487 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18489 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18490 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18494 Alan Hayward <alan.hayward@arm.com>
18495 David Sherwood <david.sherwood@arm.com>
18497 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18498 * doc/tm.texi: Regenerate.
18499 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18500 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18501 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18502 * targhooks.h (default_floatn_mode): Likewise.
18503 * targhooks.c (default_floatn_mode): Likewise.
18504 * tree.c (build_common_tree_nodes): Update accordingly.
18506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18507 Alan Hayward <alan.hayward@arm.com>
18508 David Sherwood <david.sherwood@arm.com>
18510 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18511 (mode_iterator::iterate_p): Likewise.
18512 (mode_iterator::get_wider): Likewise.
18513 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18516 Alan Hayward <alan.hayward@arm.com>
18517 David Sherwood <david.sherwood@arm.com>
18519 * coretypes.h (opt_scalar_float_mode): New typedef.
18520 * machmode.h (float_mode_for_size): New function.
18521 * emit-rtl.c (double_mode): Delete.
18522 (init_emit_once): Use float_mode_for_size.
18523 * stor-layout.c (layout_type): Likewise.
18524 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18527 Alan Hayward <alan.hayward@arm.com>
18528 David Sherwood <david.sherwood@arm.com>
18530 * output.h (assemble_real): Take a scalar_float_mode.
18531 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18532 * config/arm/arm.md (consttable_4): Likewise.
18533 (consttable_8): Likewise.
18534 (consttable_16): Likewise.
18535 * config/mips/mips.md (consttable_float): Likewise.
18536 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18537 * varasm.c (assemble_real): Take a scalar_float_mode.
18538 (output_constant_pool_2): Update accordingly.
18539 (output_constant): Likewise.
18541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18542 Alan Hayward <alan.hayward@arm.com>
18543 David Sherwood <david.sherwood@arm.com>
18545 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18546 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18547 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18548 (native_encode_real): Likewise.
18549 (native_interpret_real): Likewise.
18550 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18551 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18554 Alan Hayward <alan.hayward@arm.com>
18555 David Sherwood <david.sherwood@arm.com>
18557 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18558 <scalar_float_mode>. Simplify.
18559 (gen_extend_conv_libfunc): Likewise.
18561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18562 Alan Hayward <alan.hayward@arm.com>
18563 David Sherwood <david.sherwood@arm.com>
18565 * coretypes.h (scalar_float_mode): New type.
18566 * machmode.h (mode_traits::from_int): Use machine_mode if
18567 USE_ENUM_MODES is defined.
18568 (is_a): New function.
18570 (dyn_cast): Likewise.
18571 (scalar_float_mode): New class.
18572 (scalar_float_mode::includes_p): New function.
18573 (is_float_mode): Likewise.
18574 * gdbhooks.py (MachineModePrinter): New class.
18575 (build_pretty_printer): Use it for scalar_float_mode.
18576 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18577 (format_helper::format_helper): Turn into a template.
18578 * genmodes.c (get_mode_class): New function.
18579 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18580 or machine_mode if none.
18581 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18582 as_a <scalar_float_mode>.
18583 * dwarf2out.c (mem_loc_descriptor): Likewise.
18584 (insert_float): Likewise.
18585 (add_const_value_attribute): Likewise.
18586 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18587 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18588 (expand_unop): Update accordingly.
18589 (expand_abs_nojump): Likewise.
18590 (expand_copysign_absneg): Take a scalar_float_mode.
18591 (expand_copysign_bit): Likewise.
18592 (expand_copysign): Update accordingly.
18594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18595 Alan Hayward <alan.hayward@arm.com>
18596 David Sherwood <david.sherwood@arm.com>
18598 * coretypes.h (opt_mode): New class.
18599 * machmode.h (opt_mode): Likewise.
18600 (opt_mode::else_void): New function.
18601 (opt_mode::require): Likewise.
18602 (opt_mode::exists): Likewise.
18603 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18604 (GET_MODE_2XWIDER_MODE): Likewise.
18605 (mode_iterator::get_wider): Update accordingly.
18606 (mode_iterator::get_2xwider): Likewise.
18607 (mode_iterator::get_known_wider): Likewise, turning into a template.
18608 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18609 forcing a wider mode to exist.
18610 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18611 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18612 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18613 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18614 * lower-subreg.c (init_lower_subreg): Likewise.
18615 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18616 on the final iteration.
18617 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18618 a wider mode exists before asking for a move pattern.
18619 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18620 forcing a wider mode to exist.
18621 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18622 returning false if no such mode exists.
18623 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18624 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18625 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18626 Avoid checking for a MODE_INT if we already know the mode is not a
18628 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18629 forcing a wider mode to exist.
18630 (expmed_mult_highpart_optab): Likewise.
18631 (expmed_mult_highpart): Likewise.
18632 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18634 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18635 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18636 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18637 * internal-fn.c (expand_mul_overflow): Update use of
18638 GET_MODE_2XWIDER_MODE.
18639 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18640 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18641 GET_MODE_WIDER_MODE.
18642 (convert_plusminus_to_widen): Likewise.
18643 * tree-switch-conversion.c (array_value_type): Likewise.
18644 * var-tracking.c (emit_note_insn_var_location): Likewise.
18645 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18646 Return false inside rather than outside the loop if no wider mode
18648 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18649 and GET_MODE_2XWIDER_MODE
18650 (can_compare_p): Use else_void.
18651 * gdbhooks.py (OptMachineModePrinter): New class.
18652 (build_pretty_printer): Use it for opt_mode.
18654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18655 Alan Hayward <alan.hayward@arm.com>
18656 David Sherwood <david.sherwood@arm.com>
18658 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18659 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18662 Alan Hayward <alan.hayward@arm.com>
18663 David Sherwood <david.sherwood@arm.com>
18665 * machmode.h (mode_traits): New structure.
18666 (get_narrowest_mode): New function.
18667 (mode_iterator::start): Likewise.
18668 (mode_iterator::iterate_p): Likewise.
18669 (mode_iterator::get_wider): Likewise.
18670 (mode_iterator::get_known_wider): Likewise.
18671 (mode_iterator::get_2xwider): Likewise.
18672 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18673 (FOR_EACH_MODE): Likewise.
18674 (FOR_EACH_MODE_FROM): Likewise.
18675 (FOR_EACH_MODE_UNTIL): Likewise.
18676 (FOR_EACH_WIDER_MODE): Likewise.
18677 (FOR_EACH_2XWIDER_MODE): Likewise.
18678 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18679 * combine.c (simplify_comparison): Likewise
18680 * config/i386/i386.c (type_natural_mode): Likewise.
18681 * cse.c (cse_insn): Likewise.
18682 * dse.c (find_shift_sequence): Likewise.
18683 * emit-rtl.c (init_derived_machine_modes): Likewise.
18684 (init_emit_once): Likewise.
18685 * explow.c (hard_function_value): Likewise.
18686 * expmed.c (extract_fixed_bit_field_1): Likewise.
18687 (extract_bit_field_1): Likewise.
18688 (expand_divmod): Likewise.
18689 (emit_store_flag_1): Likewise.
18690 * expr.c (init_expr_target): Likewise.
18691 (convert_move): Likewise.
18692 (alignment_for_piecewise_move): Likewise.
18693 (widest_int_mode_for_size): Likewise.
18694 (emit_block_move_via_movmem): Likewise.
18695 (copy_blkmode_to_reg): Likewise.
18696 (set_storage_via_setmem): Likewise.
18697 (compress_float_constant): Likewise.
18698 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18699 * optabs-query.c (get_best_extraction_insn): Likewise.
18700 * optabs.c (expand_binop): Likewise.
18701 (expand_twoval_unop): Likewise.
18702 (expand_twoval_binop): Likewise.
18703 (widen_leading): Likewise.
18704 (widen_bswap): Likewise.
18705 (expand_parity): Likewise.
18706 (expand_unop): Likewise.
18707 (prepare_cmp_insn): Likewise.
18708 (prepare_float_lib_cmp): Likewise.
18709 (expand_float): Likewise.
18710 (expand_fix): Likewise.
18711 (expand_sfix_optab): Likewise.
18712 * postreload.c (move2add_use_add2_insn): Likewise.
18713 * reg-stack.c (reg_to_stack): Likewise.
18714 * reginfo.c (choose_hard_reg_mode): Likewise.
18715 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18716 * stor-layout.c (mode_for_size): Likewise.
18717 (smallest_mode_for_size): Likewise.
18718 (mode_for_vector): Likewise.
18719 (finish_bitfield_representative): Likewise.
18720 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18721 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18722 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18723 * var-tracking.c (prepare_call_arguments): Likewise.
18725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18726 Alan Hayward <alan.hayward@arm.com>
18727 David Sherwood <david.sherwood@arm.com>
18729 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18730 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18731 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18732 * machmode.h (mode_size): Move earlier in file.
18733 (mode_precision): Likewise.
18734 (mode_inner): Likewise.
18735 (mode_nunits): Likewise.
18736 (mode_unit_size): Likewise.
18737 (unit_unit_precision): Likewise.
18738 (mode_wider): Likewise.
18739 (mode_2xwider): Likewise.
18740 (machine_mode): New class.
18741 (mode_to_bytes): New function.
18742 (mode_to_bits): Likewise.
18743 (mode_to_precision): Likewise.
18744 (mode_to_inner): Likewise.
18745 (mode_to_unit_size): Likewise.
18746 (mode_to_unit_precision): Likewise.
18747 (mode_to_nunits): Likewise.
18748 (GET_MODE_SIZE): Use mode_to_bytes.
18749 (GET_MODE_BITSIZE): Use mode_to_bits.
18750 (GET_MODE_PRECISION): Use mode_to_precision.
18751 (GET_MODE_INNER): Use mode_to_inner.
18752 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18753 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18754 (GET_MODE_NUNITS): Use mode_to_nunits.
18755 * system.h (ALWAYS_INLINE): New macro.
18756 * config/powerpcspe/powerpcspe-c.c
18757 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18758 int for arg1_mode and arg2_mode.
18760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18761 Alan Hayward <alan.hayward@arm.com>
18762 David Sherwood <david.sherwood@arm.com>
18764 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18765 Prefix mode names with E_ in case statements.
18766 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18767 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18768 (aarch64_split_simd_move): Likewise.
18769 (aarch64_gen_storewb_pair): Likewise.
18770 (aarch64_gen_loadwb_pair): Likewise.
18771 (aarch64_gen_store_pair): Likewise.
18772 (aarch64_gen_load_pair): Likewise.
18773 (aarch64_get_condition_code_1): Likewise.
18774 (aarch64_constant_pool_reload_icode): Likewise.
18775 (get_rsqrte_type): Likewise.
18776 (get_rsqrts_type): Likewise.
18777 (get_recpe_type): Likewise.
18778 (get_recps_type): Likewise.
18779 (aarch64_gimplify_va_arg_expr): Likewise.
18780 (aarch64_simd_container_mode): Likewise.
18781 (aarch64_emit_load_exclusive): Likewise.
18782 (aarch64_emit_store_exclusive): Likewise.
18783 (aarch64_expand_compare_and_swap): Likewise.
18784 (aarch64_gen_atomic_cas): Likewise.
18785 (aarch64_emit_bic): Likewise.
18786 (aarch64_emit_atomic_swap): Likewise.
18787 (aarch64_emit_atomic_load_op): Likewise.
18788 (aarch64_evpc_trn): Likewise.
18789 (aarch64_evpc_uzp): Likewise.
18790 (aarch64_evpc_zip): Likewise.
18791 (aarch64_evpc_ext): Likewise.
18792 (aarch64_evpc_rev): Likewise.
18793 (aarch64_evpc_dup): Likewise.
18794 (aarch64_gen_ccmp_first): Likewise.
18795 (aarch64_gen_ccmp_next): Likewise.
18796 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18797 (alpha_emit_xfloating_libcall): Likewise.
18798 (emit_insxl): Likewise.
18799 (alpha_arg_type): Likewise.
18800 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18801 (arc_preferred_simd_mode): Likewise.
18802 (arc_secondary_reload): Likewise.
18803 (get_arc_condition_code): Likewise.
18804 (arc_print_operand): Likewise.
18805 (arc_legitimate_constant_p): Likewise.
18806 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18807 * config/arc/arc.md (casesi_load): Likewise.
18808 (casesi_compact_jump): Likewise.
18809 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18810 (cc_use_register): Likewise.
18811 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18812 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18813 (arm_init_iwmmxt_builtins): Likewise.
18814 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18815 (neon_expand_vector_init): Likewise.
18816 (arm_attr_length_move_neon): Likewise.
18817 (maybe_get_arm_condition_code): Likewise.
18818 (arm_emit_vector_const): Likewise.
18819 (arm_preferred_simd_mode): Likewise.
18820 (arm_output_iwmmxt_tinsr): Likewise.
18821 (thumb1_output_casesi): Likewise.
18822 (thumb2_output_casesi): Likewise.
18823 (arm_emit_load_exclusive): Likewise.
18824 (arm_emit_store_exclusive): Likewise.
18825 (arm_expand_compare_and_swap): Likewise.
18826 (arm_evpc_neon_vuzp): Likewise.
18827 (arm_evpc_neon_vzip): Likewise.
18828 (arm_evpc_neon_vrev): Likewise.
18829 (arm_evpc_neon_vtrn): Likewise.
18830 (arm_evpc_neon_vext): Likewise.
18831 (arm_validize_comparison): Likewise.
18832 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18833 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18834 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18835 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18836 (c6x_preferred_simd_mode): Likewise.
18837 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18838 (epiphany_rtx_costs): Likewise.
18839 * config/epiphany/predicates.md (proper_comparison_operator):
18841 * config/frv/frv.c (condexec_memory_operand): Likewise.
18842 (frv_emit_move): Likewise.
18843 (output_move_single): Likewise.
18844 (output_condmove_single): Likewise.
18845 (frv_hard_regno_mode_ok): Likewise.
18846 (frv_matching_accg_mode): Likewise.
18847 * config/h8300/h8300.c (split_adds_subs): Likewise.
18848 (h8300_rtx_costs): Likewise.
18849 (h8300_print_operand): Likewise.
18850 (compute_mov_length): Likewise.
18851 (output_logical_op): Likewise.
18852 (compute_logical_op_length): Likewise.
18853 (compute_logical_op_cc): Likewise.
18854 (h8300_shift_needs_scratch_p): Likewise.
18855 (output_a_shift): Likewise.
18856 (compute_a_shift_length): Likewise.
18857 (compute_a_shift_cc): Likewise.
18858 (expand_a_rotate): Likewise.
18859 (output_a_rotate): Likewise.
18860 * config/i386/i386.c (classify_argument): Likewise.
18861 (function_arg_advance_32): Likewise.
18862 (function_arg_32): Likewise.
18863 (function_arg_64): Likewise.
18864 (function_value_64): Likewise.
18865 (ix86_gimplify_va_arg): Likewise.
18866 (ix86_legitimate_constant_p): Likewise.
18867 (put_condition_code): Likewise.
18868 (split_double_mode): Likewise.
18869 (ix86_avx256_split_vector_move_misalign): Likewise.
18870 (ix86_expand_vector_logical_operator): Likewise.
18871 (ix86_split_idivmod): Likewise.
18872 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18873 (ix86_build_const_vector): Likewise.
18874 (ix86_build_signbit_mask): Likewise.
18875 (ix86_match_ccmode): Likewise.
18876 (ix86_cc_modes_compatible): Likewise.
18877 (ix86_expand_branch): Likewise.
18878 (ix86_expand_sse_cmp): Likewise.
18879 (ix86_expand_sse_movcc): Likewise.
18880 (ix86_expand_int_sse_cmp): Likewise.
18881 (ix86_expand_vec_perm_vpermi2): Likewise.
18882 (ix86_expand_vec_perm): Likewise.
18883 (ix86_expand_sse_unpack): Likewise.
18884 (ix86_expand_int_addcc): Likewise.
18885 (ix86_split_to_parts): Likewise.
18886 (ix86_vectorize_builtin_gather): Likewise.
18887 (ix86_vectorize_builtin_scatter): Likewise.
18888 (avx_vpermilp_parallel): Likewise.
18889 (inline_memory_move_cost): Likewise.
18890 (ix86_tieable_integer_mode_p): Likewise.
18891 (x86_maybe_negate_const_int): Likewise.
18892 (ix86_expand_vector_init_duplicate): Likewise.
18893 (ix86_expand_vector_init_one_nonzero): Likewise.
18894 (ix86_expand_vector_init_one_var): Likewise.
18895 (ix86_expand_vector_init_concat): Likewise.
18896 (ix86_expand_vector_init_interleave): Likewise.
18897 (ix86_expand_vector_init_general): Likewise.
18898 (ix86_expand_vector_set): Likewise.
18899 (ix86_expand_vector_extract): Likewise.
18900 (emit_reduc_half): Likewise.
18901 (ix86_emit_i387_round): Likewise.
18902 (ix86_mangle_type): Likewise.
18903 (ix86_expand_round_sse4): Likewise.
18904 (expand_vec_perm_blend): Likewise.
18905 (canonicalize_vector_int_perm): Likewise.
18906 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18907 (expand_vec_perm_1): Likewise.
18908 (expand_vec_perm_interleave3): Likewise.
18909 (expand_vec_perm_even_odd_pack): Likewise.
18910 (expand_vec_perm_even_odd_1): Likewise.
18911 (expand_vec_perm_broadcast_1): Likewise.
18912 (ix86_vectorize_vec_perm_const_ok): Likewise.
18913 (ix86_expand_vecop_qihi): Likewise.
18914 (ix86_expand_mul_widen_hilo): Likewise.
18915 (ix86_expand_sse2_abs): Likewise.
18916 (ix86_expand_pextr): Likewise.
18917 (ix86_expand_pinsr): Likewise.
18918 (ix86_preferred_simd_mode): Likewise.
18919 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18920 * config/i386/sse.md (*andnot<mode>3): Likewise.
18921 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18922 (*<code><mode>3): Likewise.
18923 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18924 (ia64_expand_atomic_op): Likewise.
18925 (ia64_arg_type): Likewise.
18926 (ia64_mode_to_int): Likewise.
18927 (ia64_scalar_mode_supported_p): Likewise.
18928 (ia64_vector_mode_supported_p): Likewise.
18929 (expand_vec_perm_broadcast): Likewise.
18930 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18931 (iq2000_function_arg_advance): Likewise.
18932 (iq2000_function_arg): Likewise.
18933 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18934 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18935 (m68k_libcall_value): Likewise.
18936 (m68k_function_value): Likewise.
18937 (sched_attr_op_type): Likewise.
18938 * config/mcore/mcore.c (mcore_output_move): Likewise.
18939 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18941 (microblaze_function_arg): Likewise.
18942 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18943 (mips_print_operand): Likewise.
18944 (mips_mode_ok_for_mov_fmt_p): Likewise.
18945 (mips_vector_mode_supported_p): Likewise.
18946 (mips_preferred_simd_mode): Likewise.
18947 (mips_expand_vpc_loongson_even_odd): Likewise.
18948 (mips_expand_vec_unpack): Likewise.
18949 (mips_expand_vi_broadcast): Likewise.
18950 (mips_expand_vector_init): Likewise.
18951 (mips_expand_vec_reduc): Likewise.
18952 (mips_expand_msa_cmp): Likewise.
18953 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18954 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18955 (cc_flags_for_mode): Likewise.
18956 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18957 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18958 (nds32_output_casesi_pc_relative): Likewise.
18959 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18960 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18961 (nvptx_gen_unpack): Likewise.
18962 (nvptx_gen_pack): Likewise.
18963 (nvptx_gen_shuffle): Likewise.
18964 (nvptx_gen_wcast): Likewise.
18965 (nvptx_preferred_simd_mode): Likewise.
18966 * config/pa/pa.c (pa_secondary_reload): Likewise.
18967 * config/pa/predicates.md (base14_operand): Likewise.
18968 * config/powerpcspe/powerpcspe-c.c
18969 (altivec_resolve_overloaded_builtin): Likewise.
18970 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18972 (rs6000_preferred_simd_mode): Likewise.
18973 (output_vec_const_move): Likewise.
18974 (rs6000_expand_vector_extract): Likewise.
18975 (rs6000_split_vec_extract_var): Likewise.
18976 (reg_offset_addressing_ok_p): Likewise.
18977 (rs6000_legitimate_offset_address_p): Likewise.
18978 (rs6000_legitimize_address): Likewise.
18979 (rs6000_emit_set_const): Likewise.
18980 (rs6000_const_vec): Likewise.
18981 (rs6000_emit_move): Likewise.
18982 (spe_build_register_parallel): Likewise.
18983 (rs6000_darwin64_record_arg_recurse): Likewise.
18984 (swap_selector_for_mode): Likewise.
18985 (spe_init_builtins): Likewise.
18986 (paired_init_builtins): Likewise.
18987 (altivec_init_builtins): Likewise.
18988 (do_load_for_compare): Likewise.
18989 (rs6000_generate_compare): Likewise.
18990 (rs6000_expand_float128_convert): Likewise.
18991 (emit_load_locked): Likewise.
18992 (emit_store_conditional): Likewise.
18993 (rs6000_output_function_epilogue): Likewise.
18994 (rs6000_handle_altivec_attribute): Likewise.
18995 (rs6000_function_value): Likewise.
18996 (emit_fusion_gpr_load): Likewise.
18997 (emit_fusion_p9_load): Likewise.
18998 (emit_fusion_p9_store): Likewise.
18999 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19000 (fusion_gpr_mem_load): Likewise.
19001 (fusion_addis_mem_combo_load): Likewise.
19002 (fusion_addis_mem_combo_store): Likewise.
19003 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19004 (fusion_gpr_mem_load): Likewise.
19005 (fusion_addis_mem_combo_load): Likewise.
19006 (fusion_addis_mem_combo_store): Likewise.
19007 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19009 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19010 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19011 (rs6000_preferred_simd_mode): Likewise.
19012 (output_vec_const_move): Likewise.
19013 (rs6000_expand_vector_extract): Likewise.
19014 (rs6000_split_vec_extract_var): Likewise.
19015 (reg_offset_addressing_ok_p): Likewise.
19016 (rs6000_legitimate_offset_address_p): Likewise.
19017 (rs6000_legitimize_address): Likewise.
19018 (rs6000_emit_set_const): Likewise.
19019 (rs6000_const_vec): Likewise.
19020 (rs6000_emit_move): Likewise.
19021 (rs6000_darwin64_record_arg_recurse): Likewise.
19022 (swap_selector_for_mode): Likewise.
19023 (paired_init_builtins): Likewise.
19024 (altivec_init_builtins): Likewise.
19025 (rs6000_expand_float128_convert): Likewise.
19026 (emit_load_locked): Likewise.
19027 (emit_store_conditional): Likewise.
19028 (rs6000_output_function_epilogue): Likewise.
19029 (rs6000_handle_altivec_attribute): Likewise.
19030 (rs6000_function_value): Likewise.
19031 (emit_fusion_gpr_load): Likewise.
19032 (emit_fusion_p9_load): Likewise.
19033 (emit_fusion_p9_store): Likewise.
19034 * config/rx/rx.c (rx_gen_move_template): Likewise.
19035 (flags_from_mode): Likewise.
19036 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19037 (s390_slb_comparison): Likewise.
19038 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19039 (s390_vector_mode_supported_p): Likewise.
19040 (s390_cc_modes_compatible): Likewise.
19041 (s390_match_ccmode_set): Likewise.
19042 (s390_canonicalize_comparison): Likewise.
19043 (s390_emit_compare_and_swap): Likewise.
19044 (s390_branch_condition_mask): Likewise.
19045 (s390_rtx_costs): Likewise.
19046 (s390_secondary_reload): Likewise.
19047 (__SECONDARY_RELOAD_CASE): Likewise.
19048 (s390_expand_cs): Likewise.
19049 (s390_preferred_simd_mode): Likewise.
19050 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19051 * config/sh/sh.c (sh_print_operand): Likewise.
19052 (dump_table): Likewise.
19053 (sh_secondary_reload): Likewise.
19054 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19055 * config/sh/sh.md (casesi_worker_1): Likewise.
19056 (casesi_worker_2): Likewise.
19057 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19058 (fcc_comparison_operator): Likewise.
19059 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19060 (emit_soft_tfmode_cvt): Likewise.
19061 (sparc_preferred_simd_mode): Likewise.
19062 (output_cbranch): Likewise.
19063 (sparc_print_operand): Likewise.
19064 (sparc_expand_vec_perm_bmask): Likewise.
19065 (vector_init_bshuffle): Likewise.
19066 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19067 (spu_vector_mode_supported_p): Likewise.
19068 (spu_expand_insv): Likewise.
19069 (spu_emit_branch_or_set): Likewise.
19070 (spu_handle_vector_attribute): Likewise.
19071 (spu_builtin_splats): Likewise.
19072 (spu_builtin_extract): Likewise.
19073 (spu_builtin_promote): Likewise.
19074 (spu_expand_sign_extend): Likewise.
19075 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19076 (tilegx_simd_int): Likewise.
19077 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19078 (tilepro_simd_int): Likewise.
19079 * config/v850/v850.c (const_double_split): Likewise.
19080 (v850_print_operand): Likewise.
19081 (ep_memory_offset): Likewise.
19082 * config/vax/vax.c (vax_rtx_costs): Likewise.
19083 (vax_output_int_move): Likewise.
19084 (vax_output_int_add): Likewise.
19085 (vax_output_int_subtract): Likewise.
19086 * config/visium/predicates.md (visium_branch_operator): Likewise.
19087 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19088 (visium_print_operand_address): Likewise.
19089 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19090 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19091 (xtensa_expand_conditional_branch): Likewise.
19092 (xtensa_copy_incoming_a7): Likewise.
19093 (xtensa_output_literal): Likewise.
19094 * dfp.c (decimal_real_maxval): Likewise.
19095 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19098 Alan Hayward <alan.hayward@arm.com>
19099 David Sherwood <david.sherwood@arm.com>
19101 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19102 (mode_nunits_inline): Likewise.
19103 (mode_inner_inline): Likewise.
19104 (mode_unit_size_inline): Likewise.
19105 (mode_unit_precision_inline): Likewise.
19106 (emit_insn_modes_h): Likewise. Also emit a #define of the
19108 (emit_mode_wider): Add an E_ prefix to mode names.
19109 (emit_mode_complex): Likewise.
19110 (emit_mode_inner): Likewise.
19111 (emit_mode_adjustments): Likewise.
19112 (emit_mode_int_n): Likewise.
19113 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19114 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19115 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19116 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19117 (CRC32_BUILTIN, ENTRY): Likewise.
19118 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19119 (aarch64_pop_regs): Likewise.
19120 (aarch64_process_components): Likewise.
19121 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19122 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19123 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19124 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19125 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19126 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19128 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19129 (pcmpistr): Likewise.
19130 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19131 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19132 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19134 * config/rl78/rl78.c (mduc_regs): Likewise.
19135 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19136 (htm_expand_builtin): Likewise.
19137 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19138 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19139 * config/xtensa/xtensa.c (print_operand): Likewise.
19140 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19141 (NUM_MODE_VECTOR_INT): Likewise.
19142 * genoutput.c (null_operand): Likewise.
19143 (output_operand_data): Likewise.
19144 * genrecog.c (print_parameter_value): Likewise.
19145 * lra.c (debug_operand_data): Likewise.
19147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19149 * dwarf2out.c (scompare_loc_descriptor_wide)
19150 (scompare_loc_descriptor_narrow): New functions, split out from...
19151 (scompare_loc_descriptor): ...here.
19152 * expmed.c (emit_store_flag_int): New function, split out from...
19153 (emit_store_flag): ...here.
19155 2017-08-30 Richard Biener <rguenther@suse.de>
19157 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19158 (dwarf2out_early_finish): Move setting of AT_pubnames from
19159 early debug output to early finish.
19161 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19163 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19164 and -mdata-region to the assembler.
19166 2017-08-30 Richard Biener <rguenther@suse.de>
19168 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19170 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19171 (dwarf2out_early_global_decl): Only generate a DIE for the
19172 abstract origin if it doesn't already exist or is a declaration DIE.
19173 (resolve_addr): Do not add the linkage name twice when
19174 generating a stub DIE for the DW_TAG_GNU_call_site target.
19176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19178 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19179 Use machine_mode rather than int for arg1_mode.
19181 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19184 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19185 that the second argument of the built-in functions to unpack
19186 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19187 switch statement instead a lot of if statements.
19188 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19189 Allow 64-bit values to be in Altivec registers as well as
19190 traditional floating point registers.
19191 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19193 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19195 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19196 MAX_REGS_PER_ADDRESS == 1.
19198 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19200 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19201 * config/i386/i386.c (ix86_option_override_internal): Simplify
19202 setting of opts->x_flag_entry.
19204 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19205 Jakub Jelinek <jakub@redhat.com>
19206 Richard Biener <rguenther@suse.de>
19208 PR tree-optimization/81503
19209 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19210 folded constant fits in the target type; reorder tests for clarity.
19212 2017-08-29 Martin Liska <mliska@suse.cz>
19214 * passes.def: Include pass_lower_switch.
19215 * stmt.c (dump_case_nodes): Remove and move to
19216 tree-switch-conversion.
19217 (case_values_threshold): Likewise.
19218 (expand_switch_as_decision_tree_p): Likewise.
19219 (emit_case_decision_tree): Likewise.
19220 (expand_case): Likewise.
19221 (balance_case_nodes): Likewise.
19222 (node_has_low_bound): Likewise.
19223 (node_has_high_bound): Likewise.
19224 (node_is_bounded): Likewise.
19225 (emit_case_nodes): Likewise.
19226 (struct simple_case_node): New struct.
19227 (add_case_node): Remove.
19228 (emit_case_dispatch_table): Use vector instead of case_list.
19229 (reset_out_edges_aux): Remove.
19230 (compute_cases_per_edge): Likewise.
19231 (expand_case): Build list of simple_case_node.
19232 (expand_sjlj_dispatch_table): Use it.
19233 * tree-switch-conversion.c (struct case_node): Moved from
19234 stmt.c and adjusted.
19235 (emit_case_nodes): Likewise.
19236 (node_has_low_bound): Likewise.
19237 (node_has_high_bound): Likewise.
19238 (node_is_bounded): Likewise.
19239 (case_values_threshold): Likewise.
19240 (reset_out_edges_aux): Likewise.
19241 (compute_cases_per_edge): Likewise.
19242 (add_case_node): Likewise.
19243 (dump_case_nodes): Likewise.
19244 (balance_case_nodes): Likewise.
19245 (expand_switch_as_decision_tree_p): Likewise.
19246 (emit_jump): Likewise.
19247 (emit_case_decision_tree): Likewise.
19248 (try_switch_expansion): Likewise.
19249 (do_jump_if_equal): Likewise.
19250 (emit_cmp_and_jump_insns): Likewise.
19251 (fix_phi_operands_for_edge): New function.
19252 (record_phi_operand_mapping): Likewise.
19253 (class pass_lower_switch): New pass.
19254 (pass_lower_switch::execute): New function.
19255 (make_pass_lower_switch): Likewise.
19256 (conditional_probability):
19257 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19258 * tree-pass.h: Add make_pass_lower_switch.
19260 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19263 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19266 2017-08-29 Richard Biener <rguenther@suse.de>
19268 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19269 we do not add a DW_AT_inline attribute twice.
19270 (gen_subprogram_die): Remove code setting DW_AT_inline on
19271 DECL_ABSTRACT_P nodes.
19273 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19275 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19276 calls to internal functions.
19277 (gimplify_modify_expr): Likewise.
19278 * tree-call-cdce.c (use_internal_fn): Likewise.
19279 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19280 (convert_to_divmod): Set the nothrow flag.
19281 * tree-if-conv.c (predicate_mem_writes): Likewise.
19282 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19283 (vectorizable_call): Likewise.
19284 (vectorizable_store): Likewise.
19285 (vectorizable_load): Likewise.
19286 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19287 (vect_recog_mask_conversion_pattern): Likewise.
19289 2017-08-29 Martin Liska <mliska@suse.cz>
19292 * gcc.c (driver_handle_option): Add new argument.
19293 * opts-common.c (handle_option): Pass
19294 target_option_override_hook.
19295 * opts-global.c (lang_handle_option): Add new option.
19296 (set_default_handlers): Add new argument.
19297 (decode_options): Likewise.
19298 * opts.c (target_handle_option): Likewise.
19299 (common_handle_option): Call target_option_override_hook.
19300 * opts.h (struct cl_option_handler_func): Add hook for
19301 target option override.
19302 (struct cl_option_handlers): Likewise.
19303 (set_default_handlers): Add new argument.
19304 (decode_options): Likewise.
19305 (common_handle_option): Likewise.
19306 (target_handle_option): Likewise.
19307 * toplev.c (toplev::main): Pass targetm.target_option.override
19310 2017-08-29 Richard Biener <rguenther@suse.de>
19311 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19313 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19314 life to the active subtree.
19316 2017-08-28 Jeff Law <law@redhat.com>
19318 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19319 derive_equivalences.
19320 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19321 Code moved into....
19322 (edge_info::derive_equivalences): New private member function
19324 * tree-ssa-dom.c (class edge_info): Changed from a struct
19325 to a class. Add ctor/dtor, methods and data members.
19326 (edge_info::edge_info): Renamed from allocate_edge_info.
19327 Initialize additional members.
19328 (edge_info::~edge_info): New.
19329 (free_dom_edge_info): Delete the edge info.
19330 (record_edge_info): Use new class & associated member functions.
19331 Tighten forms for testing for edge equivalences.
19332 (record_temporary_equivalences): Iterate over the simple
19333 equivalences rather than assuming there's only one per edge.
19334 (cprop_into_successor_phis): Iterate over the simple
19335 equivalences rather than assuming there's only one per edge.
19336 (optimize_stmt): Use operand_equal_p rather than pointer
19337 equality for mini-DSE code.
19339 2017-08-28 Nathan Sidwell <nathan@acm.org>
19341 * gcc.c (execute): Fold SIGPIPE handling into switch
19342 statement. Adjust internal error message.
19344 2017-08-28 Richard Biener <rguenther@suse.de>
19347 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19348 Do nothing for removed DIEs.
19350 2017-08-28 Richard Biener <rguenther@suse.de>
19352 PR tree-optimization/81977
19353 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19356 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19359 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19360 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19361 using targetm.gen_mem_thread_fence.
19363 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19366 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19367 predicate to register_operand. Reorder operands.
19368 (*btr<mode>): Ditto.
19369 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19370 (*btr<mode>_mask): Ditto.
19372 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19374 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19375 * config/rs6000/xmmintrin.h: New file.
19376 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19378 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19381 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19382 parameter and_insn and return it.
19383 (recombine_lvx_pattern): Insert a copy to ensure availability of
19384 the base register of the copied masking operation at the point of
19385 the instruction replacement.
19386 (recombine_stvx_pattern): Likewise.
19388 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19390 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19391 undocumented switches.
19392 (-mpower9-dform-vector): Likewise.
19393 (-mpower9-dform): Likewise.
19394 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19395 comments to delete references to -mpower9-dform* switches.
19396 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19397 Delete reference to -mpower9-dform* switches, test for
19398 -mpower9-vector instead.
19399 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19400 (OTHER_P9_VECTOR_MASKS): Likewise.
19401 (POWERPC_MASKS): Likewise.
19402 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19403 tests against -mpower9-dform* to -mpower9-vector. Delete code
19404 that checked for -mpower9-dform* consistancy with other options.
19405 Add test for -mpower9-misc to enable other power9 switches.
19406 (rs6000_init_hard_regno_mode_ok): Likewise.
19407 (rs6000_option_override_internal): Likewise.
19408 (rs6000_emit_prologue): Likewise.
19409 (rs6000_emit_epilogue): Likewise.
19410 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19411 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19412 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19414 (emit_fusion_p9_store): Likewise.
19415 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19416 resetting these macros if the assembler does not support ISA 3.0
19418 (TARGET_P9_DFORM_VECTOR): Likewise.
19419 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19420 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19422 2017-08-25 Alan Modra <amodra@gmail.com>
19424 PR rtl-optimization/81747
19425 * cse.c (cse_extended_basic_block): Don't attempt to record
19426 equivalences for degenerate conditional branches that branch
19427 to their fall-through.
19429 2017-08-24 Martin Sebor <msebor@redhat.com>
19431 PR middle-end/81908
19432 * gimple-fold.c (size_must_be_zero_p): New function.
19433 (gimple_fold_builtin_memory_op): Call it.
19435 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19437 * config/rs6000/mm_malloc.h: New file.
19439 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19441 PR tree-optimization/81913
19442 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19443 analysis when either IVs in condition can wrap.
19445 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19447 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19448 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19450 2017-08-24 Richard Biener <rguenther@suse.de>
19453 * targhooks.c (default_target_can_inline_p): Properly
19454 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19455 is present and always compare.
19456 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19457 infer -mfpmath=sse from TARGET_SSE_P.
19458 (ix86_can_inline_p): Properly use target_option_default_node when
19459 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19461 2017-08-24 Richard Biener <rguenther@suse.de>
19464 * dwarf2out.c (output_die): Handle flag_generate_offload like
19466 (output_comp_unit): Likewise.
19467 (gen_array_type_die): Likewise.
19468 (dwarf2out_early_finish): Likewise.
19469 (note_variable_value_in_expr): Likewise.
19470 (dwarf2out_finish): Likewise. Adjust assert.
19471 * cgraphunit.c (symbol_table::compile): Move setting of
19472 flag_generate_offload earlier ...
19473 (symbol_table::finalize_compilation_unit): ... here, before
19474 early debug finalization.
19476 2017-08-24 Richard Biener <rguenther@suse.de>
19478 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19479 and ipa-fnsummary.h.
19480 (ix86_can_inline_p): When ix86_fpmath flags do not match
19481 check whether the callee uses FP math at all.
19483 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19485 PR middle-end/81931
19486 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19487 instead of TYPE_PRECISION.
19489 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19490 Alan Hayward <alan.hayward@arm.com>
19491 David Sherwood <david.sherwood@arm.com>
19493 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19495 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19497 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19499 (ix86_frame::stack_realign_allocate): New field.
19500 (struct machine_frame_state): Modify comments.
19501 (machine_frame_state::sp_realigned_fp_end): New field.
19502 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19503 layout calculation.
19504 (sp_valid_at): Add assertion to assure no attempt to access invalid
19505 offset of a realigned stack.
19506 (fp_valid_at): Likewise.
19507 (choose_baseaddr): Modify comments.
19508 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19509 ix86_expand_prologue.
19510 (ix86_expand_prologue): Modify stack realignment and allocation.
19511 (ix86_expand_epilogue): Modify comments.
19512 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19513 avx2_runtime, avx512f, and avx512f_runtime.
19515 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19517 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19518 (mstackrealign): Do not init to -1.
19519 * config/i386/i386.c (ix86_option_override_internal):
19520 Check opts_set, not opts when setting default value of
19521 opts->x_ix86_force_align_arg_pointer.
19523 2017-08-23 Richard Biener <rguenther@suse.de>
19525 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19526 lang_hooks.decl_printable_name.
19527 * print-rtl-function.c (print_rtx_function): Likewise.
19528 * tree-pretty-print.c (dump_function_header): Likewise.
19530 2017-08-23 Richard Biener <rguenther@suse.de>
19533 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19534 -g0 at compile-time.
19536 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19538 PR middle-end/19706
19539 * doc/sourcebuild.texi (Other hardware attributes):
19542 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19544 PR middle-end/19706
19545 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19546 Fix single-use check.
19548 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19550 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19552 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19554 * doc/install.texi: Modify to add more details on running selected
19557 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19559 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19560 is combined with -mabi=ms.
19561 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19564 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19566 PR tree-optimization/81488
19567 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19568 and cached_basis fields.
19569 (MAX_SPREAD): New constant.
19570 (alloc_cand_and_find_basis): Initialize new fields.
19571 (clear_visited): New function.
19572 (create_phi_basis_1): Rename from create_phi_basis, set visited
19573 and cached_basis fields.
19574 (create_phi_basis): New wrapper function.
19575 (phi_add_costs_1): Rename from phi_add_costs, add spread
19576 parameter, set visited field, short-circuit when limits reached.
19577 (phi_add_costs): New wrapper function.
19578 (record_phi_increments_1): Rename from record_phi_increments, set
19580 (record_phi_increments): New wrapper function.
19581 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19582 (phi_incr_cost): New wrapper function.
19583 (all_phi_incrs_profitable_1): Rename from
19584 all_phi_incrs_profitable, set visited field.
19585 (all_phi_incrs_profitable): New wrapper function.
19587 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19588 Alan Hayward <alan.hayward@arm.com>
19589 David Sherwood <david.sherwood@arm.com>
19591 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19592 that takes the outer and inner modes.
19593 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19594 comparison as the canonical test for a paradoxical subreg.
19595 * combine.c (simplify_set): Use paradoxical_subreg_p.
19596 (make_extraction): Likewise.
19597 (force_to_mode): Likewise.
19598 (rtx_equal_for_field_assignment_p): Likewise.
19599 (gen_lowpart_for_combine): Likewise.
19600 (simplify_comparison): Likewise.
19601 * cse.c (equiv_constant): Likewise.
19602 * expmed.c (store_bit_field_1): Likewise.
19603 * final.c (alter_subreg): Likewise.
19604 * fwprop.c (propagate_rtx): Likewise.
19605 (forward_propagate_subreg): Likewise.
19606 * ira-conflicts.c (ira_build_conflicts): Likewise.
19607 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19608 * lra-constraints.c (curr_insn_transform): Likewise.
19609 (split_reg): Likewise.
19610 * lra-eliminations.c (move_plus_up): Likewise.
19611 (lra_eliminate_regs_1): Likewise.
19612 * recog.c (general_operand): Likewise.
19613 * ree.c (combine_reaching_defs): Likewise.
19614 * reload.c (push_reload): Likewise.
19615 (find_reloads): Likewise.
19616 * reload1.c (elimination_effects): Likewise.
19617 (compute_reload_subreg_offset): Likewise.
19618 (choose_reload_regs): Likewise.
19619 * rtlanal.c (subreg_lsb_1): Likewise.
19620 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19621 (simplify_subreg): Likewise.
19622 * var-tracking.c (track_loc_p): Likewise.
19623 * emit-rtl.c (byte_lowpart_offset): Likewise.
19624 (paradoxical_subreg_p): Delete out-of-line definition.
19626 2017-08-22 Jeff Law <law@redhat.com>
19628 PR tree-optimization/81741
19629 PR tree-optimization/71947
19630 * tree-ssa-dom.c: Include tree-inline.h.
19631 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19632 equivalences if one is more expensive to compute than the other.
19633 * tree-ssa-scopedtables.h (class const_or_copies): Make
19634 record_const_or_copy_raw method private.
19635 (class avail_exprs_stack): New method simplify_binary_operation.
19636 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19637 avail_exprs_stack::simplify_binary_operation as needed.
19638 (avail_exprs_stack::simplify_binary_operation): New function.
19640 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19642 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19643 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19644 (DOT_SYMBOLS): Likewise.
19645 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19646 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19647 (RS6000_ABI_NAME): Likewise.
19648 (TARGET_CMODEL): Likewise.
19649 (TOC_SECTION_ASM_OP): Likewise.
19650 (SET_CMODEL): New macro.
19651 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19653 2017-08-22 Richard Biener <rguenther@suse.de>
19655 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19656 to allow for free-lang-data replacements similar to verify_type_variant.
19658 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19659 Alan Hayward <alan.hayward@arm.com>
19660 David Sherwood <david.sherwood@arm.com>
19662 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19663 VOIDmode for the LABEL_REF.
19665 2017-08-22 Richard Biener <rguenther@suse.de>
19667 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19669 2017-08-22 Richard Biener <rguenther@suse.de>
19671 * common.opt (feliminate-dwarf2-dups): Ignore.
19672 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19673 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19674 same_die_p_wrap, compute_section_prefix,
19675 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19676 (comdat_symbol_id, comdat_symbol_number): Likewise.
19677 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19679 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19680 (output_die): Mark unreachable path unreachable.
19681 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19682 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19683 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19684 (dwarf2out_early_finish): Likewise.
19686 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19688 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19690 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19693 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19694 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19695 (avr_attribute_table) <io, io_low, address>: Initialize
19696 .decl_required with true.
19698 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19700 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19701 undocumented debugging options.
19702 (-mvsx-scalar-double): Likewise.
19703 (-mallow-df-permute): Likewise.
19704 (-mvectorize-builtins): Likewise.
19705 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19706 (rs6000_builtin_vectorized_function): Likewise.
19707 (rs6000_builtin_md_vectorized_function): Likewise.
19708 (rs6000_opt_vars): Likewise.
19710 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19713 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19714 (*btrq_imm): Rename from *btrq.
19715 (*btcq_imm): Rename from *btcq.
19716 (btsc): New code attribute.
19717 (*<btsc><mode>): New insn pattern.
19718 (*btr<mode>): Ditto.
19719 (*<btsc><mode>_mask): New insn_and_split pattern.
19720 (*btr<mode>_mask): Ditto.
19722 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19723 Alan Hayward <alan.hayward@arm.com>
19724 David Sherwood <david.sherwood@arm.com>
19726 * function.c (pad_below): Simplify padding calculation.
19728 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19729 Alan Hayward <alan.hayward@arm.com>
19730 David Sherwood <david.sherwood@arm.com>
19732 * target.def (function_prologue): Remove frame size argument.
19733 (function_epilogue): Likewise.
19734 * doc/tm.texi: Regenerate.
19735 * final.c (final_start_function): Update call to function_prologue.
19736 (final_end_function): Update call to function_epilogue.
19737 (default_function_pro_epilogue): Remove frame size argument.
19738 * output.h (default_function_pro_epilogue): Likewise.
19739 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19740 (arm_output_function_prologue): Likewise.
19741 * config/frv/frv.c (frv_function_prologue): Likewise.
19742 (frv_function_epilogue): Likewise.
19743 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19744 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19745 (ia64_output_function_epilogue): Likewise.
19746 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19747 (m32r_output_function_epilogue): Likewise.
19748 * config/microblaze/microblaze.c (microblaze_function_prologue)
19749 (microblaze_function_epilogue): Likewise.
19750 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19751 (mips_output_function_epilogue): Likewise.
19752 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19753 (mmix_target_asm_function_epilogue): Likewise.
19754 * config/msp430/msp430.c (msp430_start_function): Likewise.
19755 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19756 (nds32_asm_function_epilogue): Likewise.
19757 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19758 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19759 (pa_output_function_epilogue): Likewise.
19760 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19761 (rs6000_output_function_epilogue): Likewise.
19762 * config/rl78/rl78.c (rl78_start_function): Likewise.
19763 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19764 (rs6000_output_function_epilogue): Likewise.
19765 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19766 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19767 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19768 (sparc_asm_function_epilogue): Likewise.
19770 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19772 * tree.h (type_has_mode_precision_p): New function.
19773 * convert.c (convert_to_integer_1): Use it.
19774 * expr.c (expand_expr_real_2): Likewise.
19775 (expand_expr_real_1): Likewise.
19776 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19777 * match.pd: Likewise.
19778 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19779 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19780 * tree-tailcall.c (process_assignment): Likewise.
19781 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19782 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19783 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19784 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19785 (vectorizable_assignment): Likewise.
19786 (vectorizable_shift): Likewise.
19787 (vectorizable_operation): Likewise.
19788 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19790 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19792 * match.pd: Add pow (C, x) simplification.
19794 2017-08-21 Richard Biener <rguenther@suse.de>
19796 PR tree-optimization/81900
19797 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19798 for blocks with abnormal predecessors.
19799 (compute_antic): Do not set visited flag prematurely.
19801 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19804 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19806 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19808 * stor-layout.h (vector_type_mode): Move to...
19809 * tree.h (vector_type_mode): ...here.
19810 * stor-layout.c (vector_type_mode): Move to...
19811 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19813 2017-08-21 Richard Biener <rguenther@suse.de>
19815 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19816 register_external_die hooks.
19817 (debug_false_tree_charstarstar_uhwistar): Declare.
19818 (debug_nothing_tree_charstar_uhwi): Likewise.
19819 * debug.c (do_nothing_debug_hooks): Adjust.
19820 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19821 (debug_nothing_tree_charstar_uhwi): Likewise.
19822 * dbxout.c (dbx_debug_hooks): Adjust.
19823 (xcoff_debug_hooks): Likewise.
19824 * sdbout.c (sdb_debug_hooks): Likewise.
19825 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19826 * dwarf2out.c (macinfo_label_base): New global.
19827 (dwarf2out_register_external_die): New function for the
19828 register_external_die hook.
19829 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19830 (dwarf2_debug_hooks): Use them.
19831 (dwarf2_lineno_debug_hooks): Adjust.
19832 (struct die_struct): Add with_offset flag.
19833 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19834 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19835 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19836 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19837 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19838 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19839 defining section names for the early LTO debug variants.
19840 (reset_indirect_string): New helper.
19841 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19842 (print_dw_val): Add support for offsetted symbol references.
19843 (get_ultimate_context): Split out from is_cxx.
19844 (is_cxx): Use get_ultimate_context.
19845 (is_fortran): Add decl overload.
19846 (compute_comp_unit_symbol): Split out worker from
19847 compute_section_prefix.
19848 (compute_section_prefix): Call compute_comp_unit_symbol and
19849 set comdat_type_p here.
19850 (output_die): Skip DIE symbol output for the LTO added one.
19851 Handle DIE symbol references with offset.
19852 (output_comp_unit): Guard section name mangling properly.
19853 For LTO debug sections emit a symbol at the section beginning
19854 which we use to refer to its DIEs.
19855 (add_abstract_origin_attribute): For DIEs registered via
19856 dwarf2out_register_external_die directly refer to the early
19857 DIE rather than indirectly through the shadow one we created.
19858 Remove obsolete call to dwarf2out_abstract_function for
19859 non-function/block origins.
19860 (gen_array_type_die): When generating early LTO debug do
19861 not emit DW_AT_string_length.
19862 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19863 late when in LTO. As suggested place a gcc_unreachable for
19864 the DECL_ABSTRACT_P case.
19865 (gen_subprogram_die): Avoid another specification DIE
19866 for early built declarations/definitions for the late LTO case.
19867 (gen_variable_die): Add type references for late duplicated VLA dies
19869 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19870 we have the abstract instance already.
19871 (process_scope_var): Adjust decl DIE contexts in LTO which
19872 first puts them in limbo.
19873 (gen_decl_die): Do not generate type DIEs late apart from
19874 types for VLAs or for decls we do not yet have a DIE. Do not
19875 call dwarf2out_abstract_function late.
19876 (dwarf2out_early_global_decl): Make sure to create DIEs
19877 for abstract instances of a decl first.
19878 (dwarf2out_late_global_decl): Adjust comment.
19879 (output_macinfo_op): With multiple macro sections use
19880 macinfo_label_base to distinguish labels.
19881 (output_macinfo): Likewise. Update macinfo_label_base.
19882 Pass in the line info label.
19883 (note_variable_value_in_expr): When generating LTO resolve
19884 all variable values here by generating DIEs as needed.
19885 (init_sections_and_labels): Add early LTO debug flag parameter
19886 and generate different sections and names if set. Add generation
19887 counter for the labels so we can have multiple of them.
19888 (reset_dies): Helper to allow DIEs to be output multiple times.
19889 (dwarf2out_finish): When outputting DIEs to the fat part of an
19890 LTO object first reset DIEs.
19891 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19892 (modified_type_die): Check for decl_ultimate_origin being self
19894 (gen_type_die_with_usage): Likewise.
19895 (gen_typedef_die): Allow decl_ultimate_origin being self.
19896 (set_decl_abstract_flags): Remove.
19897 (set_block_abstract_flags): Likewise.
19898 (dwarf2out_abstract_function): Treat the early generated DIEs
19899 as the abstract copy and only add DW_AT_inline and
19900 DW_AT_artificial here and call set_decl_origin_self.
19901 If the DIE has an abstract origin don't do anything.
19902 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19903 if we have none yet (Go fails to build one, PR78628).
19904 (variably_modified_type_p): Prevent endless recursion for Ada
19905 cyclic pointer types.
19906 * lto-streamer-in.c: Include debug.h.
19907 (dref_queue): New global.
19908 (lto_read_tree_1): Stream in DIE references.
19909 (lto_input_tree): Register DIE references.
19910 (input_function): Stream DECL_DEBUG_ARGS.
19911 * lto-streamer-out.c: Include debug.h.
19912 (lto_write_tree_1): Output DIE references.
19913 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19914 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19915 (output_function): Stream DECL_DEBUG_ARGS.
19916 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19917 Stream DECL_ABSTRACT_ORIGIN.
19918 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19919 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19920 DECL_CONTEXT for file-scope decls.
19921 * lto-streamer.h (struct dref_entry): Declare.
19922 (dref_queue): Likewise.
19923 * cfgexpand.c (pass_expand::execute): Do not call the
19924 outlining_inline_function hook here.
19925 * lto-wrapper.c (debug_obj): New global.
19926 (tool_cleanup): Unlink it if required.
19927 (debug_objcopy): New function.
19928 (run_gcc): Handle early debug sections in the IL files by
19929 extracting them to separate files, partially linkin them and
19930 feeding the result back as result to the linker.
19931 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19932 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19933 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19934 sections into a separate segment.
19935 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19937 (darwin_asm_dwarf_section): Likewise.
19938 (darwin_asm_output_dwarf_offset): Likewise.
19939 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19941 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19942 Alan Hayward <alan.hayward@arm.com>
19943 David Sherwood <david.sherwood@arm.com>
19945 * read-md.h (md_reader::record_potential_iterator_use): Replace
19946 pointer argument with an rtx and an index.
19947 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19948 (apply_mode_iterator): Likewise.
19949 (apply_code_iterator): Likewise.
19950 (apply_int_iterator): Likewise.
19951 (apply_subst_iterator): Likewise.
19952 (record_iterator_use): Likewise.
19953 (record_attribute_use): Likewise.
19954 (md_reader::record_potential_iterator_use): Likewise. Update calls
19955 to record_iterator_use and apply_iterator.
19956 (iterator_use): Replace ptr with x and index.
19957 (attribute_use): Likewise.
19958 (apply_attribute_uses): Update calls to apply_iterator.
19959 (apply_iterators): Likewise. Update initialization of iterator_use.
19960 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19961 and record_potential_iterator_use.
19962 (rtx_reader::read_rtx_operand): Likewise.
19964 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19965 Alan Hayward <alan.hayward@arm.com>
19966 David Sherwood <david.sherwood@arm.com>
19968 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19971 2017-08-21 Richard Biener <rguenther@suse.de>
19973 PR middle-end/81884
19974 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19975 at struct end conservatively when comparing common bases.
19977 2017-08-21 Richard Biener <rguenther@suse.de>
19979 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19980 (mem_ref_in_stmt): Remove.
19981 (determine_max_movement): Use ref index to get at the reference.
19982 (invariantness_dom_walker::before_dom_children): Deal with
19983 lim data already initialized.
19984 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19986 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19988 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19989 (TARGET_ISA_ROUND): Ditto.
19990 (TARGET_ROUND): Ditto.
19991 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19992 * config/i386/i386.md: Ditto.
19993 * config/i386/sse.md: Ditto.
19994 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19995 with OPTION_MASK_ISA_SSE4_1.
19997 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20000 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20001 __builtin_ia32_lzcnt_u16.
20003 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20006 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20007 (rs6000_set_current_function): Rewrite function to use it.
20009 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20012 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20013 and TYPE_WARN_IF_NOT_ALIGN.
20014 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20015 (handle_warn_if_not_align): New.
20016 (place_union_field): Call handle_warn_if_not_align.
20017 (place_field): Call handle_warn_if_not_align.
20018 Copy TYPE_WARN_IF_NOT_ALIGN.
20019 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20020 (layout_type): Likewise.
20021 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20023 (tree_decl_common): Add warn_if_not_align.
20024 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20025 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20026 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20027 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20028 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20029 * doc/extend.texi: Document warn_if_not_aligned attribute.
20030 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20032 2017-08-17 Martin Liska <mliska@suse.cz>
20035 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20036 (get_data_dependence): Use it as pointer type.
20037 (distribute_loop): Likewise.
20039 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20041 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20042 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20043 (p8_vmrgow_<mode>_direct): New define_insn.
20044 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20045 handle endianness for vmrgew and vmrgow permute patterns.
20047 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20049 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20050 * config/rs6000/rs6000-cpus.def: Remove comment.
20051 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20052 (POWERPC_MASKS): Likewise.
20053 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20054 use of TARGET_VSX_TIMODE.
20055 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20056 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20057 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20058 (rs6000_option_override_internal): Remove dead code.
20059 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20060 (rs6000_legitimize_reload_address): Likewise.
20061 (rs6000_legitimate_address_p): Likewise.
20062 (rs6000_opt_masks): Delete "vsx-timode".
20063 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20064 from function comment.
20065 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20066 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20067 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20069 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20070 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20071 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20072 (VSX_TI): Likewise.
20074 (define_peephole2): Likewise.
20076 2017-08-17 Martin Sebor <msebor@redhat.com>
20079 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20080 past the end of an array.
20081 (test_pp_format): Add test cases.
20083 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20085 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20086 missing ECF_NOTHROW flags.
20088 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20091 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20092 operands residing in integer registers.
20093 (*vsx_le_perm_load_<mode>): Likewise.
20094 (*vsx_le_perm_store_<mode>): Likewise.
20095 (define_peephole2): Add peepholes to optimize the above.
20097 2017-08-17 Marek Polacek <polacek@redhat.com>
20099 PR middle-end/81814
20100 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20101 to mimic what shorten_compare did. Change the return type to bool.
20102 (fold_cond_expr_with_comparison): Update call to
20103 operand_equal_for_comparison_p.
20104 (fold_ternary_loc): Likewise.
20106 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20108 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20110 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20112 2017-08-17 Richard Biener <rguenther@suse.de>
20114 * tree-ssa-structalias.c (solve_graph): When propagating
20115 to successors update the graphs succ edges and avoid duplicate work.
20117 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20120 * config/i386/i386.c (ix86_option_override_internal): Save target
20121 specific options after ix86_stack_protector_guard_reg was changed.
20123 2017-08-17 Richard Biener <rguenther@suse.de>
20125 PR tree-optimization/81827
20126 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20127 (new_var_info): Initialize it conservatively.
20128 (get_call_vi): Mark register vars.
20129 (new_scalar_tmp_constraint_exp): Likewise.
20130 (handle_rhs_call): Likewise.
20131 (handle_const_call): Likewise.
20132 (create_function_info_for): Likewise.
20133 (solve_constraints): Sort varinfos to separate register from
20134 non-register vars to pack points-to solution bitmaps during
20137 2017-08-17 Marek Polacek <polacek@redhat.com>
20139 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20141 2017-08-17 Richard Biener <rguenther@suse.de>
20143 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20144 to true when overflow is undefined and we saturated the result.
20146 2017-08-17 Alan Modra <amodra@gmail.com>
20149 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20150 Don't use store multiple if only one reg needs saving.
20151 (interesting_frame_related_regno): New function.
20152 (rs6000_frame_related): Don't emit frame info for regs that
20154 (rs6000_emit_epilogue): Likewise.
20156 2017-08-16 Nathan Sidwell <nathan@acm.org>
20158 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20159 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20160 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20161 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20162 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20163 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20164 (verify_type): Adjust for TYPE_BINFO move.
20165 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20166 process TYPE_BINFO directly.
20167 (hash_tree): Likewise.
20168 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20170 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20173 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20175 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20177 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20180 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20181 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20182 Add ix86_binary_operator_ok to insn constraint.
20183 (*iordi_1_bts): Ditto.
20184 (*xordi_1_btc): Ditto.
20185 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20186 Update corresponding peephole2 pattern.
20190 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20192 PR tree-optimization/81832
20193 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20194 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20196 2017-08-16 Marek Polacek <polacek@redhat.com>
20199 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20200 regarding size_zero_node.
20202 2017-08-16 Martin Liska <mliska@suse.cz>
20205 * config.gcc: Respect previously set extra_objs in case
20208 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20210 PR tree-optimization/81835
20211 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20212 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20213 not depend on the phi.
20215 2017-08-16 Alan Modra <amodra@gmail.com>
20217 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20220 2017-08-16 Alan Modra <amodra@gmail.com>
20222 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20223 (save_reg_p): ..into this. Update all callers.
20224 (first_reg_to_save): Simplify.
20226 2017-08-16 Alan Modra <amodra@gmail.com>
20228 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20231 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20235 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20236 constant count if that count is less than 32.
20238 2017-08-15 Nathan Sidwell <nathan@acm.org>
20240 * gcc.c (execute): Emit friendlier message if inferior is killed
20241 by an external cause.
20243 2017-08-15 Richard Biener <rguenther@suse.de>
20245 PR tree-optimization/81790
20246 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20247 CONSTRUCTORs from simplifying and VN.
20249 2017-08-14 Martin Sebor <msebor@redhat.com>
20251 * builtin-attrs.def: Add comments.
20253 2017-08-14 Martin Sebor <msebor@redhat.com>
20256 * doc/extend.texi (attribute nonstring): Document new attribute.
20258 2017-08-14 Martin Sebor <msebor@redhat.com>
20261 * tree-diagnostic.c (default_tree_printer): Handle %G.
20262 * gimple-pretty-print.h (percent_G_format): Declare new function.
20263 * gimple-pretty-print.c (percent_G_format): Define.
20264 * tree-pretty-print.c (percent_K_format): Add argument.
20266 2017-08-14 Martin Sebor <msebor@redhat.com>
20268 PR translation/79998
20269 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20270 Remove a stray space.
20272 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20275 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20276 (*iordi_1_bts): Ditto.
20277 (*xordi_1_btc): Ditto.
20279 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20282 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20283 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20285 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20286 quoted strings, and make more translator-friendly.
20287 (darwin_rs6000_override_options): Likewise.
20288 (rs6000_option_override_internal): Likewise.
20289 (rs6000_return_in_memory): Fix overlong line.
20290 (init_cmulative_args): Use quoted strings, and make more
20291 translator-friendly.
20292 (rs6000_pass_by_reference): Fix overlong line.
20293 (def_builtin): Use quoted strings.
20294 (altivec_expand_predicate_builtin): Use quoted strings, and make
20295 more translator-friendly.
20296 (htm_expand_builtin): Use quoted strings.
20297 (cpu_expand_builtin): Use quoted strings, and make more
20298 translator-friendly.
20299 (altivec_expand_builtin): Likewise.
20300 (paired_expand_predicate_builtin): Likewise.
20301 (rs6000_invalid_builtin): Likewise.
20302 (builtin_function_type): Use quoted strings.
20303 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20304 more translator-friendly.
20305 (rs6000_trampoline_init): Likewise.
20306 (rs6000_handle_altivec_attribute): Likewise.
20307 (rs6000_inner_target_options): Use quoted strings.
20308 (rs6000_disable_incompatible_switches): Likewise.
20309 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20310 strings, and make more translator-friendly.
20311 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20313 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20315 PR tree-optimization/81799
20316 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20317 cond_expr to simple gimple operand.
20319 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20321 PR middle-end/46932
20322 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20324 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20328 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20329 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20330 TARGET_GASISR_PROLOGUES.
20331 * config/avr/avr.c (avr_option_override): Same.
20332 (avr_pass_pre_proep::execute): Same.
20334 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20337 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20338 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20340 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20342 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20343 define_insn_and_split. Split to a memory load from 0 in
20344 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20345 using PTR mode iterator.
20346 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20347 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20348 (*add_tp_<mode>): Redefine as define_insn_and_split.
20349 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20350 address space. Merge with *add_tp_x32 using PTR mode iterator.
20351 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20352 Split to an add with a memory load from 0 in
20353 DEFAULT_TLS_SEG_REG address space.
20355 2017-08-12 Andrew Pinski <apinski@cavium.com>
20357 * config/aarch64/aarch64-option-extensions.def (rdma):
20358 Fix feature string to what Linux prints out in /proc/cpuinfo.
20360 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20363 * dwarf2out.c (modified_type_die): For C typedef types that have
20364 an ultimate origin, process the ultimate origin instead of the
20366 (gen_typedef_die): Assert that input DECLs have no ultimate
20368 (gen_type_die_with_usage): For typedef variants that have an
20369 ultimate origin, just call gen_decl_die on the original DECL.
20370 (process_scope_var): Avoid creating DIEs for local typedefs and
20371 concrete static variables.
20373 2017-08-12 Alan Modra <amodra@gmail.com>
20377 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20378 match gnu-user.h startfile.
20379 (ENDFILE_LINUX_SPEC): Similarly.
20381 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20384 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20386 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20388 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20389 * config/aarch64/aarch64.md (mov<mode>): Change.
20390 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20391 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20392 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20394 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20396 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20397 for storage order barriers.
20399 2017-08-11 Martin Liska <mliska@suse.cz>
20402 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20403 variables of void type.
20405 2017-08-11 Martin Liska <mliska@suse.cz>
20407 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20408 TARGET_SUPPORTS_ALIASES.
20409 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20410 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20411 (optimize_weakref): Likewise.
20412 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20413 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20414 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20416 2017-08-11 Martin Liska <mliska@suse.cz>
20419 * config/i386/i386.c (make_resolver_func): Do complete
20420 refactoring of the function.
20422 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20425 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20426 * config/i386/i386.c (ix86_stack_protect_guard): Use
20427 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20428 * doc/invoke.texi (x86 Options): Document
20429 -mstack-protector-guard-symbol= option.
20431 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20433 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20434 * config/i386/i386.c (ix86_split_stack_guard): New function.
20435 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20436 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20437 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20438 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20439 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20440 (split_stack_space_check): Call ix86_split_stack_guard.
20442 2017-08-10 Martin Sebor <msebor@redhat.com>
20444 * print-tree.c (print_node): Print location using the established
20446 Replace spaces with colons.
20447 (debug_raw, debug): Ditto.
20449 2017-08-10 Martin Sebor <msebor@redhat.com>
20452 * pretty-print.c (pp_format): Correct the handling of %s precision.
20454 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20457 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20459 (ix86_finalize_stack_frame_flags): This. Also clear
20460 frame_pointer_needed if -fno-omit-frame-pointer is used without
20462 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20463 with ix86_finalize_stack_frame_flags.
20464 (ix86_expand_epilogue): Likewise.
20465 (ix86_expand_split_stack_prologue): Likewise.
20466 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20468 2017-08-10 Martin Liska <mliska@suse.cz>
20471 * c-attribs.c (handle_target_attribute):
20472 Report warning for an empty string argument of target attribute.
20474 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20477 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20479 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20480 or DECL_NONLOCAL labels.
20481 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20482 or DECL_NONLOCAL labels here.
20484 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20486 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20487 to indicate when early gimple folding has been disabled.
20488 (rs6000_gimple_fold_builtin): Add debug content.
20489 (rs6000_invalid_builtin): Fix whitespace.
20490 (rs6000_expand_builtin): Fix whitespace.
20491 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20493 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20496 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20497 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20499 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20501 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20502 (qdf24xx): Likewise.
20503 * config/aarch64/aarch64.md: Include falkor.md.
20504 * config/aarch64/falkor.md: New.
20506 2017-08-09 Marek Polacek <polacek@redhat.com>
20509 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20510 * diagnostic.c (emit_diagnostic): Add a comment.
20511 (emit_diagnostic_valist): New function.
20513 2017-08-09 Marek Polacek <polacek@redhat.com>
20516 * input.c (make_location): New overload.
20517 * input.h (make_location): Declare.
20519 2017-08-08 Alan Modra <amodra@gmail.com>
20520 H.J. Lu <hongjiu.lu@intel.com>
20523 * gcc.c (NO_PIE_SPEC): Delete.
20524 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20526 (LINK_PIE_SPEC): ..to here.
20527 (LINK_COMMAND_SPEC): Support -no-pie.
20528 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20529 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20531 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20532 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20533 (ENDFILE_CRTEND_SPEC): Similarly.
20535 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20538 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20539 (mstack-protector-guard-offset=): Ditto.
20540 * config/i386/i386.c (ix86_option_override): Handle
20541 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20543 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20544 ix86_stack_protect_guard_offset variables.
20545 (TARGET_STACK_PROTECT_GUARD): Always define.
20546 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20547 and -mstack-protector-guard-offset= options.
20549 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20551 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20552 boundary case for the last candidate.
20554 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20556 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20557 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20559 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20561 PR middle-end/19706
20562 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20563 * config/aarch64/aarch64-builtins.c
20564 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20565 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20566 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20568 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20569 Andrew Pinski <pinskia@gmail.com>
20571 PR middle-end/19706
20572 * internal-fn.def (XORSIGN): New.
20573 * optabs.def (xorsign_optab): New.
20574 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20575 (convert_expand_mult_copysign): New.
20576 (pass_optimize_widening_mul::execute): Call
20577 convert_expand_mult_copysign.
20579 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20581 PR tree-optimization/81354
20582 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20583 Insert on edges rather than explicitly creating landing pads.
20584 (analyze_candidates_and_replace): Commit edge inserts.
20586 2017-08-08 Richard Biener <rguenther@suse.de>
20588 PR middle-end/81719
20589 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20590 (expand_simple_operations): Also look through ADDR_EXPRs with
20591 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20593 2017-08-08 Richard Biener <rguenther@suse.de>
20595 PR tree-optimization/81723
20596 * tree-vect-slp.c (struct bst_traits): New hash traits.
20597 (bst_fail): New global.
20598 (vect_build_slp_tree_2): New worker, split out from ...
20599 (vect_build_slp_tree): ... this now wrapping it with using
20600 bst_fail set to cache SLP tree build fails. Properly handle
20602 (vect_analyze_slp_instance): Allocate and free bst_fail.
20604 2017-08-08 Martin Liska <mliska@suse.cz>
20607 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20608 LABEL_DECLs that can be from a different function.
20610 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20612 PR tree-optimization/81744
20613 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20614 loop's number of iterations.
20616 2017-08-08 Martin Liska <mliska@suse.cz>
20618 * asan.c: Include header files.
20619 * attribs.c (build_decl_attribute_variant): New function moved
20621 (build_type_attribute_qual_variant): Likewise.
20622 (cmp_attrib_identifiers): Likewise.
20623 (simple_cst_list_equal): Likewise.
20624 (omp_declare_simd_clauses_equal): Likewise.
20625 (attribute_value_equal): Likewise.
20626 (comp_type_attributes): Likewise.
20627 (build_type_attribute_variant): Likewise.
20628 (lookup_ident_attribute): Likewise.
20629 (remove_attribute): Likewise.
20630 (merge_attributes): Likewise.
20631 (merge_type_attributes): Likewise.
20632 (merge_decl_attributes): Likewise.
20633 (merge_dllimport_decl_attributes): Likewise.
20634 (handle_dll_attribute): Likewise.
20635 (attribute_list_equal): Likewise.
20636 (attribute_list_contained): Likewise.
20637 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20638 (lookup_attribute_by_prefix): Likewise.
20639 * bb-reorder.c: Include header files.
20640 * builtins.c: Likewise.
20641 * calls.c: Likewise.
20642 * cfgexpand.c: Likewise.
20643 * cgraph.c: Likewise.
20644 * cgraphunit.c: Likewise.
20645 * convert.c: Likewise.
20646 * dwarf2out.c: Likewise.
20647 * final.c: Likewise.
20648 * fold-const.c: Likewise.
20649 * function.c: Likewise.
20650 * gimple-expr.c: Likewise.
20651 * gimple-fold.c: Likewise.
20652 * gimple-pretty-print.c: Likewise.
20653 * gimple.c: Likewise.
20654 * gimplify.c: Likewise.
20655 * hsa-common.c: Likewise.
20656 * hsa-gen.c: Likewise.
20657 * internal-fn.c: Likewise.
20658 * ipa-chkp.c: Likewise.
20659 * ipa-cp.c: Likewise.
20660 * ipa-devirt.c: Likewise.
20661 * ipa-fnsummary.c: Likewise.
20662 * ipa-inline.c: Likewise.
20663 * ipa-visibility.c: Likewise.
20665 * lto-cgraph.c: Likewise.
20666 * omp-expand.c: Likewise.
20667 * omp-general.c: Likewise.
20668 * omp-low.c: Likewise.
20669 * omp-offload.c: Likewise.
20670 * omp-simd-clone.c: Likewise.
20671 * opts-global.c: Likewise.
20672 * passes.c: Likewise.
20673 * predict.c: Likewise.
20674 * sancov.c: Likewise.
20675 * sanopt.c: Likewise.
20676 * symtab.c: Likewise.
20677 * toplev.c: Likewise.
20678 * trans-mem.c: Likewise.
20679 * tree-chkp.c: Likewise.
20680 * tree-eh.c: Likewise.
20681 * tree-into-ssa.c: Likewise.
20682 * tree-object-size.c: Likewise.
20683 * tree-parloops.c: Likewise.
20684 * tree-profile.c: Likewise.
20685 * tree-ssa-ccp.c: Likewise.
20686 * tree-ssa-live.c: Likewise.
20687 * tree-ssa-loop.c: Likewise.
20688 * tree-ssa-sccvn.c: Likewise.
20689 * tree-ssa-structalias.c: Likewise.
20690 * tree-ssa.c: Likewise.
20691 * tree-streamer-in.c: Likewise.
20692 * tree-vectorizer.c: Likewise.
20693 * tree-vrp.c: Likewise.
20694 * tsan.c: Likewise.
20695 * ubsan.c: Likewise.
20696 * varasm.c: Likewise.
20697 * varpool.c: Likewise.
20698 * tree.c: Remove functions moved to attribs.[ch].
20699 * tree.h: Likewise.
20700 * config/aarch64/aarch64.c: Add attrs.h header file.
20701 * config/alpha/alpha.c: Likewise.
20702 * config/arc/arc.c: Likewise.
20703 * config/arm/arm.c: Likewise.
20704 * config/avr/avr.c: Likewise.
20705 * config/bfin/bfin.c: Likewise.
20706 * config/c6x/c6x.c: Likewise.
20707 * config/cr16/cr16.c: Likewise.
20708 * config/cris/cris.c: Likewise.
20709 * config/darwin.c: Likewise.
20710 * config/epiphany/epiphany.c: Likewise.
20711 * config/fr30/fr30.c: Likewise.
20712 * config/frv/frv.c: Likewise.
20713 * config/ft32/ft32.c: Likewise.
20714 * config/h8300/h8300.c: Likewise.
20715 * config/i386/winnt.c: Likewise.
20716 * config/ia64/ia64.c: Likewise.
20717 * config/iq2000/iq2000.c: Likewise.
20718 * config/lm32/lm32.c: Likewise.
20719 * config/m32c/m32c.c: Likewise.
20720 * config/m32r/m32r.c: Likewise.
20721 * config/m68k/m68k.c: Likewise.
20722 * config/mcore/mcore.c: Likewise.
20723 * config/microblaze/microblaze.c: Likewise.
20724 * config/mips/mips.c: Likewise.
20725 * config/mmix/mmix.c: Likewise.
20726 * config/mn10300/mn10300.c: Likewise.
20727 * config/moxie/moxie.c: Likewise.
20728 * config/msp430/msp430.c: Likewise.
20729 * config/nds32/nds32-isr.c: Likewise.
20730 * config/nds32/nds32.c: Likewise.
20731 * config/nios2/nios2.c: Likewise.
20732 * config/nvptx/nvptx.c: Likewise.
20733 * config/pa/pa.c: Likewise.
20734 * config/pdp11/pdp11.c: Likewise.
20735 * config/powerpcspe/powerpcspe.c: Likewise.
20736 * config/riscv/riscv.c: Likewise.
20737 * config/rl78/rl78.c: Likewise.
20738 * config/rx/rx.c: Likewise.
20739 * config/s390/s390.c: Likewise.
20740 * config/sh/sh.c: Likewise.
20741 * config/sol2.c: Likewise.
20742 * config/sparc/sparc.c: Likewise.
20743 * config/spu/spu.c: Likewise.
20744 * config/stormy16/stormy16.c: Likewise.
20745 * config/tilegx/tilegx.c: Likewise.
20746 * config/tilepro/tilepro.c: Likewise.
20747 * config/v850/v850.c: Likewise.
20748 * config/vax/vax.c: Likewise.
20749 * config/visium/visium.c: Likewise.
20750 * config/xtensa/xtensa.c: Likewise.
20752 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20755 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20756 constraints since the -mupper-regs-* switches have been
20758 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20759 into a vector from a double word element that was extracted from
20760 another vector, and eliminate extra XXPERMDI instructions.
20761 (vsx_concat_<mode>_2): Likewise.
20762 (vsx_concat_<mode>_3): Likewise.
20763 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20764 concat to allow optimizing inserts from previous extracts.
20766 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20768 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20769 memory reference to a SSP offset in TLS address space.
20770 (ix86_print_operand) <case '@'>: Remove.
20771 (ix86_print_operand_punct_valid_p): Remove '@' code.
20772 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20773 UNSPEC_SP_TLS_TEST.
20774 (stack_tls_protect_set_<mode>): Remove.
20775 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20776 (stack_tls_protect_test_<mode>): Remove.
20777 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20779 2017-08-07 Olivier Hainque <hainque@adacore.com>
20782 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20784 2017-08-07 Douglas Rupp <rupp@adacore.com>
20786 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20787 variable was referenced as multidir in command.
20789 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20792 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20795 2017-08-07 Martin Liska <mliska@suse.cz>
20797 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20798 * config/rl78/rl78.c: Add include of attribs.h.
20799 * config/sh/sh.c: Likewise.
20800 * config/v850/v850.c: Likewise.
20802 2017-08-07 Tom de Vries <tom@codesourcery.com>
20804 PR middle-end/78266
20805 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20807 2017-08-07 Martin Liska <mliska@suse.cz>
20809 * config/mips/mips.c: Include attribs.h.
20811 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20814 * doc/invoke.texi: Document change in behvaior for -Ofast for
20817 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20819 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20821 (aarch64_pop_regs): Likewise.
20822 (aarch64_gen_load_pair): Likewise.
20823 (aarch64_save_callee_saves): Likewise.
20824 (aarch64_restore_callee_saves): Likewise.
20826 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20828 * config/i386/i386.c: Revert the last change.
20830 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20833 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20835 (ix86_finalize_stack_frame_flags): This. Also clear
20836 frame_pointer_needed if -fno-omit-frame-pointer is used without
20838 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20839 with ix86_finalize_stack_frame_flags.
20840 (ix86_expand_epilogue): Likewise.
20841 (ix86_expand_split_stack_prologue): Likewise.
20843 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20846 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20847 to P_AES for Westmere.
20849 2017-08-07 Jonathan Yong <10walls@gmail.com>
20851 * config/i386/mingw.opt (fset-stack-executable): Removed.
20852 * config/i386/cygming.opt (fset-stack-executable): Moved
20854 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20856 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20858 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20860 2017-08-07 Marek Polacek <polacek@redhat.com>
20862 PR middle-end/81737
20863 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20865 2017-08-07 Martin Liska <mliska@suse.cz>
20867 * attribs.h (canonicalize_attr_name): New function.
20868 (cmp_attribs): Move from c-format.c and adjusted.
20869 (is_attribute_p): Moved from tree.h.
20870 * tree-inline.c: Add new includes.
20871 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20872 (private_is_attribute_p): Remove.
20873 (private_lookup_attribute): Likewise.
20874 (private_lookup_attribute_by_prefix): Simplify.
20875 (remove_attribute): Use is_attribute_p.
20876 * tree.h: Remove removed declarations.
20878 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20880 PR middle-end/81698
20881 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20882 instead of computing it in the function. Formatting fix.
20883 (expand_case): Don't rely on default_edge being the first edge,
20884 clear it if removing it, pass default_edge to
20885 emit_case_dispatch_table.
20886 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20889 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20891 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20892 insn in the function, emit NOP after the insn.
20894 2017-08-06 Tom de Vries <tom@codesourcery.com>
20896 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20899 2017-08-06 Tom de Vries <tom@codesourcery.com>
20901 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20904 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20906 PR tree-optimization/57371
20907 * match.pd: New pattern.
20909 2017-08-04 Marek Polacek <polacek@redhat.com>
20911 PR middle-end/81695
20912 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20913 perform the computation in offset_int.
20915 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20917 PR tree-optimization/81136
20918 * tree-vectorizer.h: Include tree-hash-traits.h.
20919 (vec_base_alignments): New typedef.
20920 (vec_info): Add a base_alignments field.
20921 (vect_record_base_alignments): Declare.
20922 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20924 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20925 (create_data_ref): Add an is_conditional_in_stmt argument.
20926 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20927 the is_conditional_in_stmt field.
20928 (data_ref_loc): Add an is_conditional_in_stmt field.
20929 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20930 (find_data_references_in_stmt): Update call to create_data_ref.
20931 (graphite_find_data_references_in_stmt): Likewise.
20932 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20933 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20934 (vect_record_base_alignment): New function.
20935 (vect_record_base_alignments): Likewise.
20936 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20937 for nested statements even if we fail to compute a misalignment.
20938 Use pooled base alignments for unconditional references.
20939 (vect_find_same_alignment_drs): Compare base addresses instead
20941 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20942 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20944 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20946 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20947 Add an explicit name for the enum. Use auto_vec for slp_instances
20948 and grouped_stores.
20949 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20951 (_bb_vec_info): Add a constructor and destructor.
20952 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20953 (destroy_loop_vec_info): Delete.
20954 (vect_destroy_datarefs): Likewise.
20955 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20956 (vec_info::vec_info): New function.
20957 (vec_info::~vec_info): Likewise.
20958 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20959 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20960 destroy_loop_vec_info.
20961 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20962 (_loop_vec_info::_loop_vec_info): ...this.
20963 (destroy_loop_vec_info): Replace with...
20964 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20965 the stmt_vec_infos. Leave handling of vec_info information to its
20966 destructor. Remove explicit vector releases.
20967 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20968 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20969 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20970 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20971 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20972 (destroy_bb_vec_info): Replace with...
20973 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20974 information to its destructor.
20975 (vect_slp_analyze_bb_1): Use new and delete instead of
20976 new_bb_vec_info and destroy_bb_vec_info.
20977 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20980 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20982 * tree-data-ref.h (subscript): Add access_fn field.
20983 (data_dependence_relation): Add could_be_independent_p.
20984 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20985 (same_access_functions): Move to tree-data-ref.c.
20986 * tree-data-ref.c (ref_contains_union_access_p): New function.
20987 (access_fn_component_p): Likewise.
20988 (access_fn_components_comparable_p): Likewise.
20989 (dr_analyze_indices): Add a reference to access_fn_component_p.
20990 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20992 (constant_access_functions): Likewise.
20993 (add_other_self_distances): Likewise.
20994 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20995 (initialize_data_dependence_relation): Use XCNEW and remove
20996 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20997 of access functions that have the same type. Allow the
20998 subsequence to end with different bases in some circumstances.
20999 Record the chosen access functions in SUB_ACCESS_FN.
21000 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21001 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21002 (subscript_dependence_tester_1): Likewise dra and drb.
21003 (build_classic_dist_vector): Update calls accordingly.
21004 (subscript_dependence_tester): Likewise.
21005 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21006 DDR_COULD_BE_INDEPENDENT_P.
21007 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21008 comp_alias_ddrs instead of may_alias_ddrs.
21009 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21011 (vect_analyze_data_ref_dependence): Use it if
21012 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21013 distance vectors if that fails.
21014 (dependence_distance_ge_vf): New function.
21015 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21016 LOOP_VINFO_MAY_ALIAS_DDRS.
21018 2017-08-04 Richard Biener <rguenther@suse.de>
21020 PR middle-end/81705
21021 * fold-const.c (fold_binary_loc): Properly restrict
21022 minus_var0 && minus_var1 case when associating undefined overflow
21025 2017-08-04 Tom de Vries <tom@codesourcery.com>
21027 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21029 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21031 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21032 Don't start diagnostic messages with a capital letter.
21033 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21035 (rs6000_invalid_builtin): Likewise.
21036 (rs6000_trampoline_init): Likewise.
21038 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21041 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21042 after setting changeable df flags.
21044 2017-08-03 Richard Biener <rguenther@suse.de>
21046 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21047 up if the use is in USE - X.
21049 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21051 * toplev.c (dumpfile.h): New include.
21052 (internal_error_reentered): New static function. Use it...
21053 (internal_error_function): ...here to handle reentered internal_error.
21055 2017-08-03 Richard Biener <rguenther@suse.de>
21057 PR middle-end/81148
21058 * fold-const.c (split_tree): Add minus_var and minus_con
21059 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21060 here but always use minus_*.
21061 (associate_trees): Assert we never associate with MINUS_EXPR
21062 and NULL first operand. Do not recurse for PLUS_EXPR operands
21063 when associating as MINUS_EXPR either.
21064 (fold_binary_loc): Track minus_var and minus_con.
21066 2017-08-03 Tom de Vries <tom@codesourcery.com>
21069 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21070 ACCEL_COMPILER, apply finish_options on
21071 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21073 2017-08-03 Tom de Vries <tom@codesourcery.com>
21076 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21077 function_entry_patch_area_size > 0.
21079 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21082 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21083 instead of 10??LU, perform unit multiplication in wide_int,
21084 don't change alloc_object_size_limit if the limit is larger
21087 PR tree-optimization/81655
21088 PR tree-optimization/81588
21089 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21090 the case when ranges[i].low and high are 1 for unsigned type with
21093 PR middle-end/81052
21094 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21095 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21097 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21099 * tree-vrp.h: Add include guard.
21101 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21104 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21105 (ud2): New insn pattern.
21106 * config/i386/i386.c (ix86_expand_epilogue):
21107 For naked functions, generate ud2 instead of trap insn.
21109 2017-08-02 Marek Polacek <polacek@redhat.com>
21112 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21114 2017-08-02 Tom de Vries <tom@codesourcery.com>
21115 Cesar Philippidis <cesar@codesourcery.com>
21117 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21118 Add missing edge probabilities.
21120 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21122 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21123 Correct endianness.
21125 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21127 PR middle-end/79499
21128 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21129 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21130 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21131 prologue_seq sequences - if any.
21133 2017-08-02 Richard Biener <rguenther@suse.de>
21135 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21136 via vectors if supported, integer extracts via punning if supported
21137 or otherwise vector extracts.
21139 2017-08-02 Richard Biener <rguenther@suse.de>
21141 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21143 (bitmap_insert_into_set): ... this.
21145 2017-08-02 Richard Biener <rguenther@suse.de>
21147 PR tree-optimization/81633
21149 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21151 PR tree-optimization/71752
21152 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21154 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21156 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21157 (machine_function::call_ms2sysv_pad_out): Remove field.
21158 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21159 (ix86_compute_frame_layout): Likewise.
21161 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21164 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21165 attribute with interrupt attribute.
21167 2017-08-01 Andrew Pinski <apinski@cavium.com>
21169 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21170 BIT_INSERT_EXPR's operand 1
21171 to see if the types precision matches.
21173 2017-08-01 Martin Liska <mliska@suse.cz>
21175 PR middle-end/70140
21176 * builtins.c (expand_builtin_memcpy_args): Remove.
21177 (expand_builtin_memcpy): Call newly added function
21178 expand_builtin_memory_copy_args.
21179 (expand_builtin_memcpy_with_bounds): Likewise.
21180 (expand_builtin_mempcpy): Remove last argument.
21181 (expand_builtin_mempcpy_with_bounds): Likewise.
21182 (expand_builtin_memory_copy_args): New function created from
21183 expand_builtin_mempcpy_args with small modifications.
21184 (expand_builtin_mempcpy_args): Remove.
21185 (expand_builtin_stpcpy): Remove unused argument.
21186 (expand_builtin): Likewise.
21187 (expand_builtin_with_bounds): Likewise.
21189 2017-08-01 Martin Liska <mliska@suse.cz>
21192 Make mempcpy more optimal (PR middle-end/70140).
21194 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21197 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21198 __builtin_vec_cmpne verify both arguments are compatible vectors
21199 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21200 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21201 move computation of aligned to after checking the argument types.
21205 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21206 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21209 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21210 Doug Rupp <rupp@adacore.com>
21211 Olivier Hainque <hainque@adacore.com>
21213 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21214 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21216 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21217 for TARGET_OS_CPP_BUILTIN.
21218 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21219 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21221 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21222 passing required abi options to the assembler for EABI configurations.
21223 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21224 of .text.hot and .text.unlikely sections for kernel modules when
21225 using ARM style exceptions.
21226 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21227 options. Add EXTRA_CC1_SPEC.
21228 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21230 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21232 (ARM_TARGET2_DWARF_FORMAT): Define.
21233 * config/arm/t-vxworks: Adjust multilib control to removal of the
21234 Diab command line options.
21236 2017-08-01 Martin Liska <mliska@suse.cz>
21238 PR gcov-profile/81561
21239 * gcov.c (unblock): Make unblocking safe as we need to preserve
21240 index correspondence of blocks and block_lists.
21242 2017-08-01 Richard Biener <rguenther@suse.de>
21244 PR tree-optimization/81181
21245 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21246 (compute_antic): ... end of iteration here.
21248 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21250 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21251 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21252 (ftree-slp-vectorize): Likewise.
21253 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21254 can no longer be set independent of flag_tree_loop_vectorize.
21255 * omp-general.c (emp_max_vf): Likewise.
21256 * opts.c (enable_fdo_optimizations): Remove references to
21257 flag_tree_vectorize, these are now implicit.
21258 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21259 and leave it for the options machinery.
21261 2017-08-01 Martin Liska <mliska@suse.cz>
21263 PR middle-end/70140
21264 * builtins.c (expand_builtin_memcpy_args): Remove.
21265 (expand_builtin_memcpy): Call newly added function
21266 expand_builtin_memory_copy_args.
21267 (expand_builtin_memcpy_with_bounds): Likewise.
21268 (expand_builtin_mempcpy): Remove last argument.
21269 (expand_builtin_mempcpy_with_bounds): Likewise.
21270 (expand_builtin_memory_copy_args): New function created from
21271 expand_builtin_mempcpy_args with small modifications.
21272 (expand_builtin_mempcpy_args): Remove.
21273 (expand_builtin_stpcpy): Remove unused argument.
21274 (expand_builtin): Likewise.
21275 (expand_builtin_with_bounds): Likewise.
21277 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21280 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21281 print "ds:" only for immediates in generic address space.
21283 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21286 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21287 (ix86_function_ok_for_sibcall): Return false for naked functions.
21289 2017-08-01 Richard Biener <rguenther@suse.de>
21291 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21292 (compute_antic): Seed worklist with exit block predecessors.
21293 * cfganal.c (dfs_find_deadend): For a cycle return the source
21294 of the edge closing it.
21296 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21298 * config/aarch64/aarch64.c
21299 (aarch64_can_const_movi_rtx_p): Move 0 check.
21301 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21303 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21304 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21306 * match.pd: Ditto in address comparison pattern.
21308 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21310 PR tree-optimization/81627
21311 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21312 closed ssa form for store-store chain.
21314 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21316 PR tree-optimization/81620
21317 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21318 for store-store chain.
21320 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21322 PR tree-optimization/81588
21323 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21324 ranges[i].in_p, invert comparison code ccode. For >/>=,
21325 swap rhs1 and rhs2 and comparison code unconditionally,
21326 for </<= don't do that. Don't swap rhs1/rhs2 again if
21327 ranges[i].in_p, instead invert comparison code ccode if
21328 opcode or oe->rank is BIT_IOR_EXPR.
21331 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21332 a direct optab to conversion optab.
21333 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21334 with GET_MODE_INNER as last argument instead of optab_handler.
21335 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21336 vector extraction if possible and optab is available.
21337 * expr.c (store_constructor): Use convert_optab_handler instead
21338 of optab_handler. Use vector initialization from smaller
21339 vectors if possible and optab is available.
21340 * tree-vect-stmts.c (vectorizable_load): Likewise.
21341 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21342 now have two modes.
21343 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21344 of vec_init from half-sized vectors with the same element mode.
21345 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21346 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21347 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21348 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21349 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21350 after mode in gen_vec_extract* calls.
21351 (vec_extract<mode>): Renamed to ...
21352 (vec_extract<mode><ssescalarmodelower>): ... this.
21353 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21354 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21355 element mode after mode in gen_vec_init* calls.
21356 (VEC_INIT_HALF_MODE): New mode iterator.
21357 (vec_init<mode>): Renamed to ...
21358 (vec_init<mode><ssescalarmodelower>): ... this.
21359 (vec_init<mode><ssehalfvecmodelower>): New expander.
21360 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21361 (vec_extractv2sfsf): ... this.
21362 (vec_initv2sf): Renamed to ...
21363 (vec_initv2sfsf): ... this.
21364 (vec_extractv2si): Renamed to ...
21365 (vec_extractv2sisi): ... this.
21366 (vec_initv2si): Renamed to ...
21367 (vec_initv2sisi): ... this.
21368 (vec_extractv4hi): Renamed to ...
21369 (vec_extractv4hihi): ... this.
21370 (vec_initv4hi): Renamed to ...
21371 (vec_initv4hihi): ... this.
21372 (vec_extractv8qi): Renamed to ...
21373 (vec_extractv8qiqi): ... this.
21374 (vec_initv8qi): Renamed to ...
21375 (vec_initv8qiqi): ... this.
21376 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21377 (vec_init<mode>): Renamed to ...
21378 (vec_init<mode><VEC_base_l>): ... this.
21379 (vec_extract<mode>): Renamed to ...
21380 (vec_extract<mode><VEC_base_l>): ... this.
21381 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21382 (vec_initv2sfsf): ... this.
21383 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21384 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21385 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21386 element mode after mode in gen_vec_init* calls.
21387 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21388 (vec_init<mode><Vel>): ... this.
21389 (vec_extract<mode>): Renamed to ...
21390 (vec_extract<mode><Vel>): ... this.
21391 * config/aarch64/iterators.md (Vel): New mode attribute.
21392 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21393 Add element mode after mode in gen_vec_extract* calls.
21394 * config/s390/vector.md (non_vec_l): New mode attribute.
21395 (vec_extract<mode>): Renamed to ...
21396 (vec_extract<mode><non_vec_l>): ... this.
21397 (vec_init<mode>): Renamed to ...
21398 (vec_init<mode><non_vec_l>): ... this.
21399 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21400 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21402 * config/arm/iterators.md (V_elem_l): New mode attribute.
21403 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21404 (vec_extract<mode><V_elem_l>): ... this.
21405 (vec_extractv2di): Renamed to ...
21406 (vec_extractv2didi): ... this.
21407 (vec_init<mode>): Renamed to ...
21408 (vec_init<mode><V_elem_l>): ... this.
21409 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21410 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21411 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21412 Add element mode after gen_vec_extract* calls.
21413 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21414 (vec_init<mode><unitmode>): ... this.
21415 (vec_extract<mode>): Renamed to ...
21416 (vec_extract<mode><unitmode>): ... this.
21417 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21418 (vec_init<mode><unitmode>): ... this.
21419 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21420 (vec_initv2sfsf): ... this.
21421 (vec_extractv2sf): Renamed to ...
21422 (vec_extractv2sfsf): ... this.
21423 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21424 Add element mode after gen_vec_extract* calls.
21425 * config/mips/mips.md (unitmode): New mode iterator.
21426 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21427 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21428 * config/spu/spu.md (inner_l): New mode attribute.
21429 (vec_init<mode>): Renamed to ...
21430 (vec_init<mode><inner_l>): ... this.
21431 (vec_extract<mode>): Renamed to ...
21432 (vec_extract<mode><inner_l>): ... this.
21433 * config/sparc/sparc.md (veltmode): New mode iterator.
21434 (vec_init<VMALL:mode>): Renamed to ...
21435 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21436 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21437 (vec_initv2sisi): ... this.
21438 (vec_initv2sf): Renamed to ...
21439 (vec_initv2sfsf): ... this.
21440 (vec_extractv2sf): Renamed to ...
21441 (vec_extractv2sfsf): ... this.
21442 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21443 (vec_init<mode>): Renamed to ...
21444 (vec_init<mode><VEC_base_l>): ... this.
21445 (vec_extract<mode>): Renamed to ...
21446 (vec_extract<mode><VEC_base_l>): ... this.
21447 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21448 (vec_initv2sfsf): ... this.
21449 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21450 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21451 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21452 gen_vec_init* calls.
21454 2017-08-01 Richard Biener <rguenther@suse.de>
21456 PR tree-optimization/81297
21457 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21458 TREE_OVERFLOW from INTEGER_CSTs.
21460 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21462 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21464 2017-07-31 Carl Love <cel@us.ibm.com>
21466 * config/rs6000/rs6000-c: Add support for built-in functions
21467 vector signed char vec_xl_be (signed long long, signed char *);
21468 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21469 vector signed int vec_xl_be (signed long long, signed int *);
21470 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21471 vector signed long long vec_xl_be (signed long long, signed long long *);
21472 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21473 vector signed short vec_xl_be (signed long long, signed short *);
21474 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21475 vector double vec_xl_be (signed long long, double *);
21476 vector float vec_xl_be (signed long long, float *);
21477 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21478 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21479 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21481 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21482 (altivec_expand_builtin): Add switch statement to call
21483 altivec_expand_xl_be for each builtin.
21484 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21485 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21486 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21487 __builtin_vsx_le_be_v16qi.
21488 * doc/extend.texi: Update the built-in documentation file for the
21489 new built-in functions.
21491 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21494 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21496 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21498 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21500 * config.gcc: Add z14.
21501 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21502 CPU model numbers for z13s and z14.
21503 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21505 * config/s390/s390-opts.h (enum processor_type): Rename
21506 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21507 * config/s390/s390.c (processor_table): Add field for CPU name to
21508 be passed to Binutils.
21509 (s390_asm_output_machine_for_arch): Use the new field in
21510 processor_table for Binutils.
21511 (s390_expand_builtin): Replace arch12 with z14.
21512 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21513 (s390_get_sched_attrmask): Likewise.
21514 (s390_get_unit_mask): Likewise.
21515 * config/s390/s390.opt: Add z14 to processor_type enum.
21517 2017-07-31 Martin Jambor <mjambor@suse.cz>
21520 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21521 regardless of optimization level.
21523 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21524 Martin Liska <mliska@suse.cz>
21526 * predict.def: Remove old comment and adjust probability.
21527 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21528 PREDICT statements.
21530 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21533 * config/i386/i386.c (ix86_function_naked): New function.
21534 (ix86_can_use_return_insn_p): Return false for naked functions.
21535 (ix86_expand_prologue): Skip prologue for naked functions.
21536 (ix86_expand_epilogue): Skip epilogue for naked functions
21537 and emit trap instruction.
21538 (ix86_warn_func_return): New function.
21539 (ix86_attribute_table): Add "naked" attribute specification.
21540 (TARGET_WARN_FUNC_RETURN): Define.
21541 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21543 2017-07-31 Martin Liska <mliska@suse.cz>
21545 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21546 (dump_gimple_bb_header): Always dump BB info.
21547 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21549 2017-07-31 Martin Liska <mliska@suse.cz>
21552 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21553 also with current_function_decl non-null equality.
21555 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21558 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21559 change type to the element type, instead add eltype variable and
21560 use it where we are interested in the element type.
21562 PR tree-optimization/81603
21563 * ipa-polymorphic-call.c
21564 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21565 offset arithmetic in offset_int, bail out if the resulting bit offset
21566 doesn't fit into shwi.
21568 2017-07-31 Martin Liska <mliska@suse.cz>
21570 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21571 (gimplify_save_expr): Fix comment.
21573 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21576 * config/i386/i386.c (ix86_function_arg): Update arguments for
21578 (ix86_compute_frame_layout): Set the initial stack offset to
21579 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21580 INCOMING_FRAME_SP_OFFSET.
21581 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21582 stack before exception handler returns.
21583 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21584 the 'ERROR_CODE' for exception handler.
21586 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21588 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21589 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21590 (ASM_OUTPUT_REG_POP): Ditto.
21591 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21592 instead of asm_fprintf to output pure string.
21594 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21596 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21597 to imported_module_or_decl hook.
21598 (debug_nothing_tree_tree_tree_bool): Remove.
21599 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21600 * debug.c (do_nothing_debug_hooks): Use
21601 debug_nothing_tree_tree_tree_bool_bool instead of
21602 debug_nothing_tree_tree_tree_bool.
21603 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21604 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21605 * sdbout.c (sdb_debug_hooks): Likewise.
21606 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21607 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21609 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21610 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21611 attribute, don't add anything.
21613 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21615 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21616 (fold_build2_stat_loc): Likewise.
21617 (fold_build3_stat_loc): Likewise.
21618 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21619 (fold_build1_loc): Remove macro.
21620 (fold_build2_loc): Likewise.
21621 (fold_build3_loc): Likewise.
21623 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21625 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21626 (gimple_build_debug_bind_source_stat): Likewise.
21627 * gimple.h (gimple_build_debug_bind): Remove macro.
21628 (gimple_build_debug_bind_source): Likewise.
21630 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21632 * bitmap.c (bitmap_alloc): Adjust.
21633 (bitmap_gc_alloc): Likewise.
21634 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21636 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21638 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21639 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21640 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21641 (bitmap_gc_alloc_stat): Likewise.
21642 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21644 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21646 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21647 * rtl.h (shallow_copy_rtx): Remove macro.
21649 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21651 * emit-rtl.c (gen_raw_REG): Adjust.
21652 * gengenrtl.c (gendef): Likewise.
21653 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21654 * rtl.h (rtx_alloc): Remove macro.
21656 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21658 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21659 (build_tree_list_stat): Likewise.
21660 * tree.h (build_tree_list): Remove macro.
21661 (build_tree_list_vec): Likewise.
21663 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21665 * tree.c (make_vector_stat): Remove _stat from name.
21666 (build_vector_stat): Likewise.
21667 * tree.h (make_vector_stat): Remove macro.
21668 (build_vector_stat): Likewise.
21670 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21672 * tree.h (build_var_debug_value): Remove prototype.
21674 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21676 * tree.c (tree_cons_stat): Remove _stat from name.
21677 * tree.h (tree_cons): Remove macro.
21679 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21681 * tree.c (build_vl_exp_stat): Remove _stat from name.
21682 * tree.h (build_vl_exp): Remove macro.
21684 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21686 * tree.c (build_decl_stat): Remove _stat from name.
21687 * tree.h (build_decl): Remove macro.
21689 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21691 * gimple.c (gimple_build_with_ops_stat): Adjust.
21692 (gimple_alloc_stat): Remove _stat from name.
21693 * gimple.h (gimple_alloc): Remove macro.
21695 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21697 * tree.c (make_tree_vec_stat): Remove _stat from name.
21698 (grow_tree_vec_stat): Likewise.
21699 * tree.h (make_tree_vec_stat): Adjust prototype.
21700 (grow_tree_vec_stat): Likewise.
21701 (make_tree_vec): Remove macro.
21702 (grow_tree_vec): Likewise.
21704 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21706 * fold-const.c (fold_build1_stat_loc): Adjust.
21707 (fold_build2_stat_loc): Likewise.
21708 (fold_build3_stat_loc): Likewise.
21709 * tree.c (build0_stat): Remove _stat from name.
21710 (build1_stat): Likewise.
21711 (build2_stat): Likewise.
21712 (build3_stat): Likewise.
21713 (build4_stat): Likewise.
21714 (build5_stat): Likewise.
21715 * tree.h (build1_loc): Remove macro, and rename _stat function
21717 (build2_loc): Likewise.
21718 (build3_loc): Likewise.
21719 (build4_loc): Likewise.
21720 (build5_loc): Likewise.
21722 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21724 * tree.c (make_int_cst_stat): Remove _stat from name.
21725 * tree.h (make_int_cst_stat): Adjust prototype.
21726 (make_int_cst): Remove macro.
21728 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21730 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21731 * tree.h (make_tree_binfo_stat): Adjust prototype.
21732 (make_tree_binfo): Remove.
21734 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21736 * tree.c (copy_node_stat): Rename to copy_node.
21737 (build_distinct_type_copy): Adjust.
21738 * tree.h (copy_node_stat): Adjust prototype.
21739 (copy_node): Remove macro.
21741 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21743 * tree.c (make_node_stat): rename to make_node.
21744 (build_tree_list_stat): Adjust.
21745 (build0_stat): Likewise.
21746 (build2_stat): Likewise.
21747 (build3_stat): Likewise.
21748 (build4_stat): Likewise.
21749 (build5_stat): Likewise.
21750 (build_decl_stat): Likewise.
21751 * tree.h (make_node_stat): Adjust prototype.
21752 (make_node): remove macro.
21754 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21756 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21757 (PPC_FEATURE2_SCV): Likewise.
21758 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21760 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21762 * config/aarch64/aarch64.c
21763 (aarch64_internal_mov_immediate): Add new special pattern.
21764 * config/aarch64/aarch64.md (*movdi_aarch64):
21765 Add reg/32bit const mov case.
21767 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21768 Richard Sandiford <richard.sandiford@linaro.org>
21770 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21771 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21772 Add integer and movi cases.
21773 (movi-split-hf-df-sf split, fp16): New.
21774 (enabled): Added TARGET_FP_F16INST.
21775 * config/aarch64/iterators.md (GPF_HF): New.
21776 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21778 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21780 * config/aarch64/aarch64.c
21781 (aarch64_simd_container_mode): Add prototype.
21782 (aarch64_expand_mov_immediate): Add HI support.
21783 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21784 (aarch64_can_const_movi_rtx_p): New.
21785 (aarch64_preferred_reload_class):
21786 Remove restrictions of using FP registers for certain SIMD operations.
21787 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21788 (aarch64_valid_floating_const): Add integer move validation.
21789 (aarch64_simd_imm_scalar_p): Remove.
21790 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21791 (aarch64_legitimate_constant_p): Expand list of supported cases.
21792 * config/aarch64/aarch64-protos.h
21793 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21794 (aarch64_reinterpret_float_as_int): New.
21795 (aarch64_simd_imm_scalar_p): Remove.
21796 * config/aarch64/constraints.md (Uvi): New.
21797 (Dd): Split into Ds and new Dd.
21798 * config/aarch64/aarch64.md (*movsi_aarch64):
21800 (*movdi_aarch64): Add SIMD mov case.
21802 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21804 * tree-predcom.c: (struct chain): Handle store-store chain in which
21805 stores for elimination only store loop invariant values.
21806 (execute_pred_commoning_chain): Ditto.
21807 (prepare_initializers_chain_store_elim): Ditto.
21808 (prepare_finalizers): Ditto.
21809 (is_inv_store_elimination_chain): New function.
21810 (initialize_root_vars_store_elim_1): New function.
21812 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21814 * tree-predcom.c: Revise general description of the pass.
21815 (enum chain_type): New enum type for store elimination.
21816 (struct chain): New field supporting store elimination.
21817 (struct component): Ditto.
21818 (dump_chain): Dump store-stores chain.
21819 (release_chain): Release resources.
21820 (split_data_refs_to_components): Compute and create component
21821 contains only stores for elimination.
21822 (get_chain_last_ref_at): New function.
21823 (make_invariant_chain): Initialization.
21824 (make_rooted_chain): Specify chain type in parameter and record it.
21825 (add_looparound_copies): Skip for store-stores chain.
21826 (determine_roots_comp): Compute type of chain and pass it to
21828 (initialize_root_vars_store_elim_2): New function.
21829 (finalize_eliminated_stores): New function.
21830 (remove_stmt): Handle store for elimination.
21831 (execute_pred_commoning_chain): Execute predictive commoning on
21832 store-store chains.
21833 (determine_unroll_factor): Skip unroll for store-stores chain.
21834 (prepare_initializers_chain_store_elim): New function.
21835 (prepare_initializers_chain): Hanlde store-store chain.
21836 (prepare_finalizers_chain, prepare_finalizers): New function.
21837 (tree_predictive_commoning_loop): Return integer value indicating
21838 if loop is unrolled or lcssa form is corrupted.
21839 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21841 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21843 * tree-predcom.c (initialize_root): Delete.
21844 (execute_pred_commoning_chain): Initialize root vars and replace
21845 reference of non-combined chain directly, rather than call above
21848 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21850 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21851 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21853 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21855 * tree-predcom.c (struct chain): New field init_seq.
21856 (release_chain): Release init_seq.
21857 (prepare_initializers_chain): Record intialization stmts in above
21859 (insert_init_seqs): New function.
21860 (tree_predictive_commoning_loop): Call insert_init_seqs.
21862 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21864 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21866 2017-07-28 Richard Biener <rguenther@suse.de>
21868 * match.pd: Remove superfluous :c.
21869 * genmatch.c (simplify::id): Add member.
21870 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21872 (current_id): New global.
21873 (dt_node::parent): Move from ...
21874 (dt_operand::parent): ... here. Add for_id member.
21875 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21876 (decision_tree::find_node): Relax order requirement when
21877 merging DT_TRUE nodes to ones inbetween the current simplify
21878 and the one we try to merge with. Add diagnostic whenever
21879 we need to enforce pattern order by not merging.
21880 (decision_tree::insert): Set current_id.
21881 (decision_tree::print_node): Dump parent node and for_id.
21882 (parser::last_id): Add member.
21883 (parser::push_simplify): Assign unique id.
21884 (parser::parser): Initialize last_id.
21886 2017-07-28 Martin Liska <mliska@suse.cz>
21889 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21890 gimple_build_debug_bind.
21892 2017-07-28 Richard Biener <rguenther@suse.de>
21894 PR tree-optimization/81502
21895 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21897 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21899 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21900 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21901 for BIT_FIELD_REF args.
21902 * fold-const.c (make_bit_field_ref): Likewise.
21903 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21905 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21908 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21909 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21910 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21911 Or it into SANITIZER_UNDEFINED.
21912 * ubsan.c: Include gimple-fold.h and varasm.h.
21913 (ubsan_expand_ptr_ifn): New function.
21914 (instrument_pointer_overflow): New function.
21915 (maybe_instrument_pointer_overflow): New function.
21916 (instrument_object_size): Formatting fix.
21917 (pass_ubsan::execute): Call instrument_pointer_overflow
21918 and maybe_instrument_pointer_overflow.
21919 * internal-fn.c (expand_UBSAN_PTR): New function.
21920 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21921 * sanitizer.def (__ubsan_handle_pointer_overflow,
21922 __ubsan_handle_pointer_overflow_abort): New builtins.
21923 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21924 * internal-fn.def (UBSAN_PTR): New internal function.
21925 * opts.c (sanitizer_opts): Add pointer-overflow.
21926 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21927 * fold-const.c (build_range_check): Compute pointer range check in
21928 integral type if pointer arithmetics would be needed. Formatting
21931 2017-07-28 Martin Liska <mliska@suse.cz>
21934 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21935 parameters that are of a variable-length.
21937 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21939 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21941 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21942 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21943 (CRT_CALL_STATIC_FUNCTION): Likewise.
21944 (ASM_DEFAULT_SPEC): New define.
21945 (ASM_SPEC32): Likewise.
21946 (ASM_SPEC64): Likewise.
21947 (ASM_SPEC_COMMON): Likewise.
21948 (ASM_SPEC): Likewise.
21949 (INVALID_64BIT): Likewise.
21950 (LINK_OS_DEFAULT_SPEC): Likewise.
21951 (LINK_OS_SPEC32): Likewise.
21952 (LINK_OS_SPEC64): Likewise.
21953 (POWERPC_LINUX): Likewise.
21954 (PTRDIFF_TYPE): Likewise.
21955 (RESTORE_FP_PREFIX): Likewise.
21956 (RESTORE_FP_SUFFIX): Likewise.
21957 (SAVE_FP_PREFIX): Likewise.
21958 (SAVE_FP_SUFFIX): Likewise.
21959 (SIZE_TYPE): Likewise.
21960 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21961 (TARGET_64BIT): Likewise.
21962 (TARGET_64BIT): Likewise.
21963 (TARGET_AIX): Likewise.
21964 (WCHAR_TYPE_SIZE): Likewise.
21965 (WCHAR_TYPE): Undefine.
21966 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21967 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21968 (CPP_OS_RTEMS_SPEC): Delete.
21969 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21970 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21972 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21974 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21976 PR tree-optimization/81578
21977 * tree-parloops.c (build_new_reduction): Bail out if
21978 reduction_code isn't one of the standard OpenMP reductions.
21979 Move the details printing after that decision.
21981 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21983 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21984 related to reload_in_progress.
21985 (splat_input_operand): Likewise.
21986 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21988 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21990 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21991 (TARGET_INSTANTIATE_DECLS): Likewise.
21992 (legitimate_indexed_address_p): Delete reload_in_progress code.
21993 (rs6000_debug_legitimate_address_p): Likewise.
21994 (rs6000_eliminate_indexed_memrefs): Likewise.
21995 (rs6000_emit_le_vsx_store): Likewise.
21996 (rs6000_emit_move_si_sf_subreg): Likewise.
21997 (rs6000_emit_move): Likewise.
21998 (register_to_reg_type): Likewise.
21999 (rs6000_pre_atomic_barrier): Likewise.
22000 (rs6000_machopic_legitimize_pic_address): Likewise.
22001 (rs6000_allocate_stack_temp): Likewise.
22002 (rs6000_address_for_fpconvert): Likewise.
22003 (rs6000_address_for_altivec): Likewise.
22004 (rs6000_secondary_memory_needed_rtx): Delete function.
22005 (rs6000_check_sdmode): Likewise.
22006 (rs6000_alloc_sdmode_stack_slot): Likewise.
22007 (rs6000_instantiate_decls): Likewise.
22008 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22009 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22010 Delete reload_in_progress.
22011 (*vec_reload_and_plus_<mptrsize>): Likewise.
22012 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22013 (vsx_div_v2di): Likewise.
22014 (vsx_udiv_v2di): Likewise.
22016 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22018 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22019 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22020 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22021 (rs6000_debug_reg_global): Delete print of LRA status.
22022 (rs6000_option_override_internal): Delete dead LRA related code.
22023 (rs6000_lra_p): Delete function.
22024 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22026 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22028 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22029 * config/riscv/rtems.h: New file.
22031 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22032 Sudakshina Das <sudi.das@arm.com>
22034 * config/aarch64/aarch64.md
22035 (define_split for and<mode>3nr_compare): Move
22036 non aarch64_logical_operand to a register.
22037 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22038 register immediate operand to a register.
22039 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22041 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22043 PR middle-end/81564
22044 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22046 2017-07-27 Richard Biener <rguenther@suse.de>
22048 PR tree-optimization/81573
22049 PR tree-optimization/81494
22050 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22051 multi defuse cycle case.
22053 2017-07-27 Richard Biener <rguenther@suse.de>
22055 PR tree-optimization/81571
22056 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22059 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22061 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22062 earlier and only if MASK_FPU is set. Adjust formatting.
22064 2017-07-27 Martin Liska <mliska@suse.cz>
22066 * opt-functions.awk: Add validation of value of Init.
22067 * optc-gen.awk: Pass new argument.
22069 2017-07-27 Martin Liska <mliska@suse.cz>
22071 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22072 Fix wrong condition.
22074 2017-07-27 Martin Liska <mliska@suse.cz>
22076 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22077 BBs and edges seen by autoFDO.
22079 2017-07-27 Richard Biener <rguenther@suse.de>
22081 PR tree-optimization/81502
22082 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22083 with incompatible but same sized type.
22084 (execute_update_addresses_taken): Likewise.
22086 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22088 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22089 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22091 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22094 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22095 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22096 Change s_operand to memory_operand.
22098 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22100 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22101 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22102 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22103 Emit instructions rather than returning an expression. Handle TFmode
22104 and KFmode by casting to TImode.
22105 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22106 (rs6000_emit_le_vsx_store): Likewise.
22107 * config/rs6000/vsx.md (VSX_TI): New iterator.
22108 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22109 (*vsx_le_undo_permute_<mode>): Likewise.
22110 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22111 emit the split sequence.
22112 (*vsx_le_perm_store_<mode>): Likewise.
22114 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22116 PR tree-optimization/81555
22117 PR tree-optimization/81556
22118 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22119 if true, force CHANGED for the recursive invocation.
22120 (reassociate_bb): Remember original length of ops array, pass
22121 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22123 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22124 attributes for noipa attribute. For naked attribute use
22125 lookup_attribute first before lookup_attribute_spec.
22126 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22128 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22129 for functions with noipa attribute.
22130 (cgraph_externally_visible_p): Return true for functions with noipa
22132 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22133 for functions with noipa attribute.
22134 * doc/extend.texi: Document noipa function attribute.
22135 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22136 also for functions with noipa attribute.
22137 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22139 2017-07-26 Andrew Pinski <apinski@cavium.com>
22141 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22142 vec_unalign_load_cost and vec_unalign_store_cost.
22144 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22146 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22147 -mvsx-small-integer option.
22148 (ISA_3_0_MASKS_IEEE): Likewise.
22149 (OTHER_VSX_VECTOR_MASKS): Likewise.
22150 (POWERPC_MASKS): Likewise.
22151 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22152 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22153 code, only testing for DImode being allowed in non-VSX floating
22155 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22156 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22158 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22159 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22160 TARGET_P8_VECTOR test.
22161 (rs6000_secondary_reload_simple_move): Likewise.
22162 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22163 since TARGET_P9_VECTOR was already tested.
22164 (rs6000_opt_masks): Remove -mvsx-small-integer.
22165 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22166 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22168 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22169 test for TARGET_VEXTRACTUB was used, and that uses
22171 (p9 extract splitter): Likewise.
22172 (vsx_extract_<mode>_di_p9): Likewise.
22173 (vsx_extract_<mode>_store_p9): Likewise.
22174 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22175 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22176 the elimination of TARGET_VSX_SMALL_INTEGER.
22177 (vsx_extract_<mode>_p8): Likewise.
22178 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22179 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22180 (vsx_set_<mode>_p9): Likewise.
22181 (vsx_set_v4sf_p9): Likewise.
22182 (vsx_set_v4sf_p9_zero): Likewise.
22183 (vsx_insert_extract_v4sf_p9): Likewise.
22184 (vsx_insert_extract_v4sf_p9_2): Likewise.
22185 * config/rs6000/rs6000.md (sign extend splitter): Change
22186 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22187 (floatsi<mode>2_lfiwax_mem): Likewise.
22188 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22189 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22190 since a test for TARGET_P9_VECTOR was used.
22191 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22192 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22193 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22194 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22195 TARGET_P8_VECTOR test.
22196 (fix_trunc<mode>si2_stfiwx): Likewise.
22197 (fix_trunc<mode>si2_internal): Likewise.
22198 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22199 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22201 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22202 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22203 TARGET_P8_VECTOR test.
22204 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22205 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22206 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22208 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22209 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22210 since a test for TARGET_P9_VECTOR was used.
22211 (splitter for loading small constants): Likewise.
22213 2017-07-26 Andrew Pinski <apinski@cavium.com>
22215 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22218 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22221 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22222 (fp_valid_at): Likewise.
22224 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22226 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22227 (qdf24xx_addrcost_table): Likewise.
22228 (cortexa57_tunings): Update to use generic_branch_cost.
22229 (cortexa72_tunings): Likewise.
22230 (cortexa73_tunings): Likewise.
22231 (qdf24xx_tunings): Likewise.
22233 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22235 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22236 (thunderx2t99_branch_cost): Likewise.
22237 (cortexa35_tunings): Update to use generic_branch_cost.
22238 (cortexa53_tunings): Likewise.
22239 (cortexa57_tunings): Likewise.
22240 (cortexa72_tunings): Likewise.
22241 (cortexa73_tunings): Likewise.
22242 (thunderx2t99_tunings): Likewise.
22244 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22246 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22247 (sparc_option_override): Honour MASK_FSMULD.
22248 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22249 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22250 * config/sparc/sparc.opt (mfsmuld): New option.
22251 * doc/invoke.texi (mfsmuld): Document option.
22253 2017-07-26 Marek Polacek <polacek@redhat.com>
22255 PR middle-end/70992
22256 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22258 2017-07-26 Richard Biener <rguenther@suse.de>
22260 * gimple-match-head.c (do_valueize): Return OP if valueize
22262 (get_def): New helper to get at the def stmt of a SSA name
22263 if valueize allows.
22264 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22265 do_valueize to get at the def stmt.
22266 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22268 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22270 PR middle-end/46932
22271 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22273 2017-07-26 Martin Liska <mliska@suse.cz>
22276 * function.c (expand_function_start): Make expansion of
22277 nonlocal_goto_save_area after parm_birth_insn.
22279 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22281 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22282 from all CPU target flags enable members.
22284 2017-07-26 Richard Biener <rguenther@suse.de>
22286 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22287 (decision_tree::gen): Make 'type' const.
22288 (write_predicate): Likewise.
22290 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22292 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22293 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22294 (rs6000_option_override_internal): Likewise.
22295 (rs6000_expand_vector_set): Likewise.
22296 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22297 (TARGET_UPPER_REGS_SF): Likewise.
22298 (TARGET_UPPER_REGS_DI): Likewise.
22299 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22300 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22301 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22302 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22303 (Splitters for DI constants in Altivec registers): Likewise.
22304 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22305 (vsx_set_v4sf_p9): Likewise.
22306 (vsx_set_v4sf_p9_zero): Likewise.
22307 (vsx_insert_extract_v4sf_p9): Likewise.
22308 (vsx_insert_extract_v4sf_p9_2): Likewise.
22310 2017-07-25 Carl Love <cel@us.ibm.com>
22312 * doc/extend.texi: Update the built-in documentation file for the
22313 existing built-in functions
22314 vector signed char vec_cnttz (vector signed char);
22315 vector unsigned char vec_cnttz (vector unsigned char);
22316 vector signed short vec_cnttz (vector signed short);
22317 vector unsigned short vec_cnttz (vector unsigned short);
22318 vector signed int vec_cnttz (vector signed int);
22319 vector unsigned int vec_cnttz (vector unsigned int);
22320 vector signed long long vec_cnttz (vector signed long long);
22321 vector unsigned long long vec_cnttz (vector unsigned long long);
22323 2017-07-25 Andrew Pinski <apinski@cavium.com>
22325 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22326 accesses where the use is for the first operand of a BIT_INSERT.
22328 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22331 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22332 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22334 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22336 * config/i386/gstabs.h: Delete.
22337 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22339 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22341 * config/i386/i386.c (ix86_decompose_address): Do not check for
22342 register RTX when looking at index_reg or base_reg.
22343 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22345 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22347 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22348 to update EH info here.
22350 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22352 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22354 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22356 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22358 2017-07-25 Torsten Duwe <duwe@suse.de>
22360 * common.opt: Introduce -fpatchable-function-entry
22361 command line option, and its variables function_entry_patch_area_size
22362 and function_entry_patch_area_start.
22363 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22364 including a two-value parser.
22365 * target.def (print_patchable_function_entry): New target hook.
22366 * targhooks.h (default_print_patchable_function_entry): New function.
22367 * targhooks.c (default_print_patchable_function_entry): Likewise.
22368 * toplev.c (process_options): Switch off IPA-RA if
22369 patchable function entries are being generated.
22370 * varasm.c (assemble_start_function): Look at the
22371 patchable-function-entry command line switch and current
22372 function attributes and maybe generate NOP instructions by
22373 calling the print_patchable_function_entry hook.
22374 * doc/extend.texi: Document patchable_function_entry attribute.
22375 * doc/invoke.texi: Document -fpatchable_function_entry
22376 command line option.
22377 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22379 * doc/tm.texi: Re-generate.
22381 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22384 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22385 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22387 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22389 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22391 2017-07-25 Richard Biener <rguenther@suse.de>
22393 PR tree-optimization/81455
22394 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22395 not walk in cycles when looking for guards.
22397 2017-07-25 Richard Biener <rguenther@suse.de>
22399 PR tree-optimization/81529
22400 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22401 when optimizing backedge uses.
22403 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22405 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22407 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22408 to dl_section_ref. On AIX, append an expression to subtract
22409 the size of the section length to dl_section_ref.
22411 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22413 * configure.ac: If any of the config.* scripts fail, exit 1.
22414 * configure: Regenerate.
22416 2017-07-25 Richard Biener <rguenther@suse.de>
22418 PR middle-end/81546
22419 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22420 of immediate uses, be more verbose on errors.
22422 2017-07-25 Richard Biener <rguenther@suse.de>
22424 PR tree-optimization/81510
22425 * tree-vect-loop.c (vect_is_simple_reduction): When the
22426 reduction stmt is not inside the loop bail out.
22428 2017-07-25 Richard Biener <rguenther@suse.de>
22430 PR tree-optimization/81303
22431 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22432 profitability check against LOOP_VINFO_NITERSM1.
22434 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22436 * domwalk.c (cmp_bb_postorder): Simplify.
22437 (sort_bbs_postorder): New function. Use it...
22438 (dom_walker::walk): ...here to optimize common cases.
22440 2017-07-25 Martin Liska <mliska@suse.cz>
22443 * ipa-visibility.c (function_and_variable_visibility): Make the
22444 redirection just on target that supports aliasing.
22445 Fix GNU coding style.
22447 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22450 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22452 * config/arm/rtems.h: Likewise.
22453 * config/bfin/rtems.h: Likewise.
22454 * config/i386/rtemself.h: Likewise.
22455 * config/lm32/rtems.h: Likewise.
22456 * config/m32c/rtems.h: Likewise.
22457 * config/m68k/rtemself.h: Likewise.
22458 * config/microblaze/rtems.h: Likewise.
22459 * config/mips/rtems.h: Likewise.
22460 * config/moxie/rtems.h: Likewise.
22461 * config/nios2/rtems.h: Likewise.
22462 * config/powerpcspe/rtems.h: Likewise.
22463 * config/rs6000/rtems.h: Likewise.
22464 * config/rtems.h: Likewise.
22465 * config/sh/rtems.h: Likewise.
22466 * config/sh/rtemself.h: Likewise.
22467 * config/sparc/rtemself.h: Likewise.
22469 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22472 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22473 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22474 * tree-ssa-structalias.c (alias_get_name): Same.
22476 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22479 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22480 instructions if no du chain is found.
22482 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22484 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22486 2017-07-25 Richard Biener <rguenther@suse.de>
22488 PR middle-end/81505
22489 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22492 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22494 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22495 upper-regs options.
22496 (ISA_2_7_MASKS_SERVER): Likewise.
22497 (ISA_3_0_MASKS_IEEE): Likewise.
22498 (OTHER_P8_VECTOR_MASKS): Likewise.
22499 (OTHER_VSX_VECTOR_MASKS): Likewise.
22500 (POWERPC_MASKS): Likewise.
22501 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22502 duplicate list of options.
22503 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22504 explicit -mupper-regs options.
22505 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22506 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22507 alias for -mupper-regs-df.
22508 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22509 (rs6000_init_hard_regno_mode_ok): Likewise.
22510 (rs6000_option_override_internal): Likewise.
22511 (rs6000_opt_masks): Likewise.
22512 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22513 options in terms of whether -mvsx or -mpower8-vector was used.
22514 (TARGET_UPPER_REGS_DI): Likewise.
22515 (TARGET_UPPER_REGS_SF): Likewise.
22516 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22517 -mupper-regs-* options.
22519 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22521 * passes.c (emergency_dump_function): Print some empty lines and a
22522 header before the RTL dump.
22524 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22526 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22528 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22531 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22532 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22534 2017-07-24 Carl Love <cel@us.ibm.com>
22536 * config/rs6000/rs6000-c.c: Add support for built-in functions
22537 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22538 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22539 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22540 vec_extract_fp_from_shortl): Add defines for the two builtins.
22541 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22542 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22544 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22545 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22546 * doc/extend.texi: Update the built-in documentation file for the
22547 new built-in function.
22549 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22552 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22554 * doc/generic.texi: Likewise.
22555 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22556 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22558 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22560 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22561 (aarch64_mls_elt_merge<mode>): Likewise.
22563 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22565 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22566 having __cxa_atexit.
22568 2017-07-23 Michael Collison <michael.collison@arm.com>
22570 * config/arm/arm.c (arm_option_override): Deprecate
22571 use of -mstructure-size-boundary.
22572 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22573 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22575 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22578 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22579 Reduce cost estimate for direct moves.
22581 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22584 * config/i386/i386.c (ix86_option_override_internal): Disable
22585 BMI, BMI2 and TBM instructions for -m16.
22587 2017-07-21 Carl Love <cel@us.ibm.com>
22589 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22590 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22591 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22592 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22593 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22594 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22595 VMULOSW): New enum "unspec" values.
22596 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22597 altivec_vmulosw): New patterns.
22598 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22599 VMULOSW): Add definitions.
22601 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22603 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22604 (qdf24xx): Likewise.
22605 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22606 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22607 (AARCH64_FL_V8_1): Renumber.
22608 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22609 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22610 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22611 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22612 rdma to feature modifiers list.
22614 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22616 PR middle-end/56727
22617 * ipa-visibility (function_and_variable_visibility): Convert
22618 recursive PLT call to direct call if appropriate.
22620 2017-07-21 Andrew Pinski <apinski@cavium.com>
22622 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22623 operand 1 to see if the types precision matches.
22624 * fold-const.c (operand_equal_p): Likewise.
22626 2017-07-21 Richard Biener <rguenther@suse.de>
22628 PR tree-optimization/81303
22629 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22630 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22631 (vect_peeling_hash_get_lowest_cost): Adjust.
22632 (vect_enhance_data_refs_alignment): Likewise. Use
22633 vect_get_peeling_costs_all_drs to compute the penalty for no
22634 peeling to match up costs.
22636 2017-07-21 Richard Biener <rguenther@suse.de>
22638 PR tree-optimization/81500
22639 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22640 we didn't identify a reduction path.
22642 2017-07-21 Tom de Vries <tom@codesourcery.com>
22643 Cesar Philippidis <cesar@codesourcery.com>
22645 PR gcov-profile/81442
22646 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22649 2017-07-21 Tom de Vries <tom@codesourcery.com>
22652 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22654 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22655 nvptx_override_options_after_change.
22657 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22659 * dwarf2out.c (output_file_names): Avoid double testing for
22660 dwarf_version >= 5.
22662 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22664 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22666 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22668 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22670 (try_crossjump_to_edge): Do not punt on partitioned functions.
22672 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22674 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22675 Put all BBs reachable only via paths crossing cold region to cold
22677 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22679 2016-07-21 Richard Biener <rguenther@suse.de>
22681 PR tree-optimization/81303
22682 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22683 into account prologue and epilogue iterations when raising
22684 min_profitable_iters to sth at least covering one vector iteration.
22686 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22688 * config/arm/arm.c (arm_test_cpu_arch_dat):
22691 2017-07-20 Nathan Sidwell <nathan@acm.org>
22693 Remove TYPE_METHODS.
22694 * tree.h (TYPE_METHODS): Delete.
22695 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22696 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22697 (dbxout_type_methods): Scan TYPE_FIELDS.
22698 (dbxout_type): Don't check TYPE_METHODS here.
22699 * function.c (use_register_for_decl): Always ignore register for
22700 class types when not optimizing.
22701 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22702 * tree.c (free_lang_data_in_type): Stitch out member functions and
22703 templates from TYPE_FIELDS.
22704 (build_distinct_type_copy, verify_type_variant,
22705 verify_type): Member fns are on TYPE_FIELDS.
22706 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22707 * tree-pretty-print.c (dump_generic_node): Likewise.
22709 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22712 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22713 V2TImode and V4TImode.
22714 (ix86_expand_vector_extract): Likewise.
22715 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22716 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22717 (ssescalarmode): Handle V4TImode and V2TImode.
22718 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22719 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22720 (VEXTRACTI128_MODE): New mode iterator.
22721 (splitter for *vec_extractv?ti first element): New.
22722 (VEC_INIT_MODE): New mode iterator.
22723 (vec_init<mode>): Consolidate 3 expanders into one using
22724 VEC_INIT_MODE mode iterator.
22726 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22728 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22729 non_spilled_static_chain_regno_p.
22731 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22733 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22735 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22737 * bb-reorder.c (connect_traces): Allow copying of blocks within
22740 2017-07-20 Richard Biener <rguenther@suse.de>
22742 * gimple.h (gimple_phi_result): Add gphi * overload.
22743 (gimple_phi_result_ptr): Likewise.
22744 (gimple_phi_arg): Likewise. Adjust index assert to only
22745 allow actual argument accesses rather than all slots available
22747 (gimple_phi_arg_def): Add gphi * overload.
22748 * tree-phinodes.c (make_phi_node): Initialize only actual
22750 (resize_phi_node): Clear memory not covered by old node,
22751 do not initialize excess argument slots.
22752 (reserve_phi_args_for_new_edge): Initialize new argument slot
22755 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22757 PR tree-optimization/81388
22759 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22761 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22762 by removing computation of may_be_zero.
22764 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22765 Tom de Vries <tom@codesourcery.com>
22767 PR middle-end/81030
22768 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22769 when gimple level profile disagrees with what RTL expander did.
22771 2017-07-20 Richard Biener <rguenther@suse.de>
22773 PR tree-optimization/61171
22774 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22775 (vect_analyze_stmt): Add slp instance parameter.
22776 (vectorizable_reduction): Likewise.
22777 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22778 (vect_is_simple_reduction): Deal with chains not detected
22779 as SLP reduction chain, specifically not properly associated
22780 chains containing a mix of plus/minus.
22781 (get_reduction_op): Remove.
22782 (get_initial_defs_for_reduction): Simplify, pass in whether
22783 this is a reduction chain, pass in the SLP node for the PHIs.
22784 (vect_create_epilog_for_reduction): Get the SLP instance as
22786 (vectorizable_reduction): Get the SLP instance as arg.
22787 During analysis remember the SLP node with the PHIs in the
22788 instance. Simplify getting at the vectorized reduction PHIs.
22789 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22790 through SLP instance.
22791 (vect_slp_analyze_operations): Likewise.
22792 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22793 (vect_transform_stmt): Likewise.
22795 2017-07-20 Tom de Vries <tom@codesourcery.com>
22797 PR tree-optimization/81489
22798 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22799 read of phi arg location to before loop that modifies phi.
22801 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22803 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22806 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22808 PR middle-end/81331
22809 * except.c (execute): Fix ordering issue.
22811 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22813 PR rtl-optimization/81423
22814 * combine.c (make_compound_operation_int): Don't try to optimize
22815 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22817 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22819 PR rtl-optimization/81423
22820 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22821 with a constant that is -1 in the truncated to mode.
22823 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22825 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22826 (determine_unlikely_bbs): ... here.
22827 * predict.h (propagate_unlikely_bbs_forward): Declare.
22828 * cfgexpand.c (pass_expand::execute): Use it.
22829 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22831 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22832 propagate_unlikely_bbs_forward.
22834 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22836 PR middle-end/81331
22837 * except.c (maybe_add_nop_after_section_switch): New function.
22840 2017-07-19 Tom de Vries <tom@codesourcery.com>
22842 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22844 2017-07-19 Tom de Vries <tom@codesourcery.com>
22846 * gimple.h (gimple_phi_arg): Make assert more strict.
22848 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22850 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22851 * config/rs6000/mmintrin.h: New file.
22852 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22854 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22856 PR tree-optimization/81346
22857 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22859 2017-07-19 Tom de Vries <tom@codesourcery.com>
22861 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22863 2017-07-19 Tom de Vries <tom@codesourcery.com>
22865 * config/nvptx/nvptx-modes.def: Add V2DImode.
22866 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22867 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22868 (nvptx_output_mov_insn): Handle lack of mov.b128.
22869 (nvptx_print_operand): Handle 'H' and 'L' codes.
22870 (nvptx_vector_mode_supported): Allow V2DImode.
22871 (nvptx_preferred_simd_mode): New function.
22872 (nvptx_data_alignment): New function.
22873 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22874 nvptx_preferred_simd_mode.
22875 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22877 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22879 2017-07-19 Tom de Vries <tom@codesourcery.com>
22881 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22882 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22883 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22884 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22885 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22886 (mov<VECIM>_insn): New define_insn.
22887 (define_expand "mov<VECIM>): New define_expand.
22889 2017-07-19 Tom de Vries <tom@codesourcery.com>
22891 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22893 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22895 PR tree-optimization/81346
22896 * fold-const.h (fold_div_compare, range_check_type): Declare.
22897 * fold-const.c (range_check_type): New function.
22898 (build_range_check): Use range_check_type.
22899 (fold_div_compare): No longer static, rewritten into
22900 a match.pd helper function.
22901 (fold_comparison): Don't call fold_div_compare here.
22902 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22903 as helper function.
22905 2017-07-19 Nathan Sidwell <nathan@acm.org>
22907 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22908 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22909 * tree.c (find_decls_types_r, verify_type): Use
22910 TYPE_{MIN,MAX}_VALUE_RAW.
22911 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22912 (hash_tree): Likewise.
22913 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22915 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22918 2017-07-18 Tom de Vries <tom@codesourcery.com>
22920 PR middle-end/81464
22921 * omp-expand.c (expand_omp_for_static_chunk): Handle
22922 equal-argument loop exit phi.
22924 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22927 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22928 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22929 operand 2 predicate.
22930 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22931 operand 2 predicate.
22932 (ror,rol -> rorx splitters): Use const_int_operand as
22933 operand 2 predicate.
22935 2017-06-18 Richard Biener <rguenther@suse.de>
22937 PR tree-optimization/81410
22938 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22939 the gap in the ! slp_perm SLP case after each group.
22941 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22943 PR middle-end/81463
22944 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22947 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22949 PR middle-end/81462
22950 * predict.c (set_even_probabilities): Cleanup; do not affect
22951 probabilities that are already known.
22952 (combine_predictions_for_bb): Call even when count is set.
22954 2017-07-18 Nathan Sidwell <nathan@acm.org>
22956 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22959 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22962 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22963 optimization for loop niter analysis.
22965 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22968 * config/avr/avr.c (avr_optimize_casesi): Don't use
22969 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22971 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22973 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22974 body_cost_vec from _vect_peel_extended_info.
22975 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22976 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22979 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22981 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22983 2017-07-18 Richard Biener <rguenther@suse.de>
22985 PR tree-optimization/80620
22986 PR tree-optimization/81403
22987 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22988 info when re-using a VN table entry.
22990 2017-07-18 Richard Biener <rguenther@suse.de>
22992 PR tree-optimization/81418
22993 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22994 vectype_in. Verify that with lane-reducing reduction operations
22995 we have a single def-use cycle.
22997 2017-07-17 Carl Love <cel@us.ibm.com>
22999 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23001 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23002 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23003 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23004 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23005 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23006 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23007 VMULOSW): New enum "unspec" values.
23008 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23009 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23010 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23011 altivec_vmulosw): New patterns.
23012 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23013 VMULOSW): Add definitions.
23015 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23017 * config/alpha/alpha.c: Include predict.h.
23019 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23021 * tree-vrp.c (compare_assert_loc): Fix comparison function
23022 to return predictable results.
23024 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23026 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23028 (subdi3): Likewise.
23029 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23030 * doc/invoke.texi (mexpand-adddi): Update text.
23032 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23034 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23035 that also clobbers the CC register. The old expand code is moved
23037 (*arc_clzsi2): ... here.
23038 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23039 the CC register. The old expand code is moved to ...
23040 (arc_ctzsi2): ... here.
23042 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23044 * config/arc/arc.opt (mindexed-loads): Use initial value
23045 TARGET_INDEXED_LOADS_DEFAULT.
23046 (mauto-modify-reg): Use initial value
23047 TARGET_AUTO_MODIFY_REG_DEFAULT.
23048 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23049 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23050 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23051 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23053 2017-07-17 Martin Liska <mliska@suse.cz>
23056 * opts.c (finish_options): Do not allow -fgnu-tm
23057 w/ -fsanitize={kernel-,}address. Say sorry.
23059 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23062 * tree-loop-distribution.c (classify_partition): Only assert on
23063 numer of iterations.
23064 (merge_dep_scc_partitions): Delete prameter. Update function call.
23065 (distribute_loop): Remove code handling loop with unknown niters.
23066 (pass_loop_distribution::execute): Skip loop with unknown niters.
23068 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23071 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23072 function sort_partitions_by_post_order.
23074 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23076 PR tree-optimization/81374
23077 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23078 the max index of basic blocks, rather than number of basic blocks.
23080 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23082 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23084 (arc_legitimate_pic_operand_p): Likewise.
23085 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23087 (arc_needs_pcl_p): Likewise.
23088 (arc_legitimate_pc_offset_p): Likewise.
23089 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23090 function is also used in constrains.md.
23091 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23092 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23093 PLUS. Only return true/false in known cases, otherwise assert.
23094 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23095 is already called in arc_legitimate_constant_p.
23096 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23098 (LEGITIMATE_PIC_OPERAND_P): Use
23099 arc_raw_symbolic_reference_mentioned_p function.
23100 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23105 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23106 Andrew Burgess <andrew.burgess@embecosm.com>
23108 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23109 (arc_return_address_register): New function.
23110 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23111 (arc_handle_fndecl_attribute): Add naked attribute.
23112 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23113 (TARGET_WARN_FUNC_RETURN): Likewise.
23114 (arc_allocate_stack_slots_for_args): New function.
23115 (arc_warn_func_return): Likewise.
23116 (machine_function): Change type fn_type.
23117 (arc_compute_function_type): Consider new naked function type,
23118 change function return type.
23119 (arc_must_save_register): Adapt to handle new
23120 arc_compute_function_type's return type.
23121 (arc_expand_prologue): Likewise.
23122 (arc_expand_epilogue): Likewise.
23123 (arc_return_address_regs): Delete.
23124 (arc_return_address_register): New function.
23125 (arc_epilogue_uses): Use above function.
23126 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23127 (arc_function_type): Change encoding, add naked type.
23128 (ARC_INTERRUPT_P): Change to handle the new encoding.
23129 (ARC_FAST_INTERRUPT_P): Likewise.
23130 (ARC_NORMAL_P): Define.
23131 (ARC_NAKED_P): Likewise.
23132 (arc_compute_function_type): Delete prototype.
23133 * config/arc/arc.md (in_ret_delay_slot): Use
23134 arc_return_address_register function.
23135 (simple_return): Likewise.
23136 (p_return_i): Likewise.
23138 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23140 PR tree-optimization/81428
23141 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23142 can't be built for those types.
23144 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23146 Remove stuff dead since r239246.
23148 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23149 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23150 (avr_inform_devices): Remove dead stuff.
23152 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23154 * config/arm/arm_neon.h: Fix softp typo.
23156 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23158 PR tree-optimization/81365
23159 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23160 aggregate moves onto bb predecessor edges, make sure there are no
23161 loads that could alias the lhs in between the start of bb and the
23164 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23167 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23168 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23169 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23171 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23173 PR tree-optimization/81396
23174 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23175 (init_symbolic_number): Initialize it to 1.
23176 (perform_symbolic_merge): Add n_ops from both operands into the new
23178 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23179 without base_addr as useless if they need more than one operation.
23180 (bswap_replace): Handle !bswap case for NULL base_addr.
23182 2017-07-17 Tom de Vries <tom@codesourcery.com>
23185 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23188 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23190 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23191 conditional builtin define __FIX_LEON3FT_B2BST.
23193 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23195 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23196 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23199 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23201 PR rtl-optimization/81424
23202 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23203 to remove potential trapping from operands if -fnon-call-exceptions.
23205 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23207 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23208 profile_proability for scalling.
23209 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23211 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23213 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23215 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23217 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23218 fixpoint arithmetics.
23220 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23222 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23223 fixpoint arithmetics.
23225 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23227 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23228 fixpoint arithmetics.
23230 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23232 * profile-count.h (profile_probability::from_reg_br_prob_note,
23233 profile_probability::to_reg_br_prob_note): New functions.
23234 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23235 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23236 * predict.c (probability_reliable_p): Update.
23237 (edge_probability_reliable_p): Update.
23238 (br_prob_note_reliable_p): Update.
23239 (invert_br_probabilities): Update.
23240 (add_reg_br_prob_note): New function.
23241 (combine_predictions_for_insn): Update.
23242 * asan.c (asan_clear_shadow): Update.
23243 * cfgbuild.c (compute_outgoing_frequencies): Update.
23244 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23245 (update_br_prob_note): Update.
23246 (rtl_verify_edges): Update.
23247 (purge_dead_edges): Update.
23248 (fixup_reorder_chain): Update.
23249 * emit-rtl.c (try_split): Update.
23250 * ifcvt.c (cond_exec_process_insns): Update.
23251 (cond_exec_process_if_block): Update.
23252 (dead_or_predicable): Update.
23253 * internal-fn.c (expand_addsub_overflow): Update.
23254 (expand_neg_overflow): Update.
23255 (expand_mul_overflow): Update.
23256 * loop-doloop.c (doloop_modify): Update.
23257 * loop-unroll.c (compare_and_jump_seq): Update.
23258 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23259 * predict.h: Update.
23260 * reorg.c (mostly_true_jump): Update.
23262 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23263 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23264 * config/arc/arc.c: (emit_unlikely_jump): Update.
23265 * config/arm/arm.c: (emit_unlikely_jump): Update.
23266 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23267 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23268 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23269 (ix86_print_operand): Update.
23270 (ix86_split_fp_branch): Update.
23271 (predict_jump): Update.
23272 * config/ia64/ia64.c (ia64_print_operand): Update.
23273 * config/mmix/mmix.c (mmix_print_operand): Update.
23274 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23275 (rs6000_expand_split_stack_prologue): Update.
23276 * config/rs6000/rs6000.c: Update.
23277 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23278 (s390_expand_vec_movstr): Update.
23279 (s390_expand_cs_tdsi): Update.
23280 (s390_expand_split_stack_prologue): Update.
23281 * config/sh/sh.c (sh_print_operand): Update.
23282 (expand_cbranchsi4): Update.
23283 (expand_cbranchdi4): Update.
23284 * config/sparc/sparc.c (output_v9branch): Update.
23285 * config/spu/spu.c (get_branch_target): Update.
23286 (ea_load_store_inline): Update.
23287 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23288 * config/tilepro/tilepro.c: Update.
23290 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23292 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23293 (gimplify_save_expr): Likewise.
23295 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23297 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23299 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23301 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23303 * timevar.def (TV_IPA_FNSUMMARY): Define.
23305 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23307 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23308 to back store errata sensitive sequence from being generated.
23309 (sqrtdf2_fix): Likewise.
23311 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23313 * tree-ssa-threadupdate.c (compute_path_counts,
23314 update_joiner_offpath_counts): Use profile_probability.
23316 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23319 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23321 * config/arm/arm-c.c (arm_cpu_builtins): Define
23322 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23324 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23326 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23327 array entries to represent __ieee128 versions of the
23328 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23329 scalar_extract_sig, and scalar_insert_exp built-in functions.
23330 (altivec_resolve_overloaded_builtin): Add special case handling
23331 for the __builtin_scalar_insert_exp function, as represented by
23332 the P9V_BUILTIN_VEC_VSIEDP constant.
23333 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23334 exponent support for __ieee128 argument.
23335 (VSESQP): Add scalar extract signature support for __ieee128
23337 (VSTDCNQP): Add scalar test negative support for __ieee128
23339 (VSIEQP): Add scalar insert exponent support for __int128 argument
23340 with __ieee128 result.
23341 (VSIEQPF): Add scalar insert exponent support for __ieee128
23342 argument with __ieee128 result.
23343 (VSTDCQP): Add scalar test data class support for __ieee128
23345 (VSTDCNQP): Add overload support for scalar test negative with
23346 __ieee128 argument.
23347 (VSTDCQP): Add overload support for scalar test data class
23348 __ieee128 argument.
23349 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23350 UNSPEC_VSX_SXSIGDP.
23351 (UNSPEC_VSX_SIEXPQP): New constant.
23352 (xsxexpqp): New insn for VSX scalar extract exponent quad
23354 (xsxsigqp): New insn for VSX scalar extract significand quad
23356 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23357 precision with floating point argument.
23358 (xststdcqp): New expand for VSX scalar test data class quad
23360 (xststdcnegqp): New expand for VSX scalar test negative quad
23362 (xststdcqp): New insn to match expansions for VSX scalar test data
23363 class quad precision and VSX scalar test negative quad precision.
23364 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23365 special case operand checking to enforce that second operand of
23366 VSX scalar test data class with quad precision argument is a 7-bit
23368 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23369 prototypes and descriptions of __ieee128 versions of
23370 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23371 scalar_test_data_class, and scalar_test_neg built-in functions.
23373 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23375 PR tree-optimization/81162
23376 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23377 replace a negate with an add.
23379 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23381 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23383 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23385 * config/arm/arm-c.c (arm_cpu_builtins): Define
23386 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23388 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23390 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23391 (armv8-r): Set ARM Cortex-R52 as default CPU.
23392 * config/arm/arm-tables.opt: Regenerate.
23393 * config/arm/arm-tune.md: Regenerate.
23394 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23396 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23397 extension for -mcpu=cortex-r52.
23399 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23401 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23402 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23403 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23404 (fp-armv8): Define it as FP_ARMv8 only.
23405 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23406 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23408 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23409 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23410 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23411 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23412 than TARGET_FPU_ARMV8.
23413 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23414 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23415 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23417 * config/arm/neon.md (neon_vrint): Likewise.
23418 (neon_vcvt): Likewise.
23419 (neon_<fmaxmin_op><mode>): Likewise.
23420 (<fmaxmin><mode>3): Likewise.
23421 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23422 * config/arm/predicates.md (arm_cond_move_operator): Check against
23423 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23425 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23427 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23428 to top of function.
23430 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23432 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23433 loop in comment with memset.
23435 2017-07-14 Martin Liska <mliska@suse.cz>
23437 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23438 * dwarf2out.c (is_java): Remove the function.
23439 (output_pubname): Remove usage of the function.
23440 (lower_bound_default): Remove usage of DW_LANG_Java.
23441 (gen_compile_unit_die): Likewise.
23442 * gcc.c: Remove compiler defaults for .java and .zip files.
23443 * gimple-expr.c (remove_suffix): Change as there's no longer
23444 extension than 4-letter one.
23445 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23446 (gimplify_save_expr): Likewise.
23447 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23448 as it's possible even for other languages than Java.
23449 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23450 * lto-opts.c (lto_write_options): Remove reference to Java.
23451 * opts.c (strip_off_ending): Update file extension handling.
23452 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23453 * tree-eh.c (lower_resx): Likewise.
23454 * tree.c (free_lang_data_in_type): Remove dead code.
23455 (find_decls_types_r): Likewise.
23456 (build_common_builtin_nodes): Remove Java from a comment.
23457 (verify_type): Remove dead code.
23458 * varasm.c (assemble_external): Remove Java from a comment.
23460 2017-07-14 Martin Liska <mliska@suse.cz>
23462 * opts.c (finish_options): Add quotes.
23463 (common_handle_option): Likewise.
23465 2017-07-14 Martin Liska <mliska@suse.cz>
23467 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23468 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23469 Remove N_SO_PASCAL.
23470 * dwarf2out.c (lower_bound_default): Do not handle
23472 (gen_compile_unit_die): Likewise.
23473 * gcc.c: Remove default extension binding for GNU Pascal.
23474 * stmt.c: Remove Pascal language from a comment.
23475 * xcoffout.c: Likewise.
23477 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23480 * diagnostic-show-locus.c (fixit_cmp): New function.
23481 (layout::layout): Sort m_fixit_hints.
23482 (column_range::column_range): Assert that the values are valid.
23483 (struct char_span): New struct.
23484 (correction::overwrite): New method.
23485 (struct source_line): New struct.
23486 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23487 calls in terms of classes source_line and char_span, and
23488 correction::overwrite.
23489 (selftest::test_overlapped_fixit_printing_2): New function.
23490 (selftest::diagnostic_show_locus_c_tests): Call it.
23492 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23494 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23495 early if there is no lhs.
23497 2017-07-13 Martin Liska <mliska@suse.cz>
23499 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23500 (gen_reference_type_die): Likewise.
23501 * stor-layout.c: Remove Pascal-related comment.
23503 2017-07-13 Martin Liska <mliska@suse.cz>
23505 * opts.c (finish_options): Add quotes to error messages.
23506 (parse_sanitizer_options): Likewise.
23508 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23510 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23512 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23514 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23516 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23518 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23520 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23522 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23525 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23526 provides the hardware capability bits, define the macro
23527 __BUILTIN_CPU_SUPPORTS__.
23528 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23529 if GLIBC does not provide the hardware capability bits. Add a
23530 gcc_unreachable call if the built-in cpu function is neither
23531 __builtin_cpu_is nor __builtin_cpu_supports.
23532 (rs6000_get_function_versions_dispatcher): Change the warning
23533 that an old GLIBC is used which does not export the capability
23534 bits to be an error.
23535 * doc/extend.texi (target_clones attribute): Document the
23536 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23537 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23538 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23539 the macros defined by GCC if the newer GLIBC is available.
23541 2017-07-12 Jeff Law <law@redhat.com>
23543 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23544 remaining includes slightly.
23545 * config/riscv/riscv-builtins.c: Include profile-count.h.
23547 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23550 * config/avr/avr.c (avr_set_current_function): In diagnostic
23551 messages: Quote keywords and (parts of) identifiers.
23552 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23555 2017-07-12 Carl Love <cel@us.ibm.com>
23557 * config/rs6000/rs6000-c.c: Add support for built-in functions
23558 vector bool char vec_revb (vector bool char);
23559 vector bool short vec_revb (vector short char);
23560 vector bool int vec_revb (vector bool int);
23561 vector bool long long vec_revb (vector bool long long);
23562 * doc/extend.texi: Update the built-in documentation file for the
23563 new built-in functions.
23565 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23567 * config/s390/s390.md: Remove movcc splitter.
23569 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23571 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23572 load/store on condition.
23574 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23577 * config/avr/avr.c (avr_encode_section_info)
23578 [progmem && !TREE_READONLY]: Error if progmem object needs
23581 2017-07-11 Michael Collison <michael.collison@arm.com>
23583 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23586 2017-07-11 Carl Love <cel@us.ibm.com>
23588 * config/rs6000/rs6000-c.c: Add support for builtins
23589 vector unsigned int vec_parity_lsbb (vector signed int);
23590 vector unsigned int vec_parity_lsbb (vector unsigned int);
23591 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23592 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23593 vector unsigned long long vec_parity_lsbb (vector signed long long);
23594 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23595 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23596 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23597 * doc/extend.texi: Update the built-in documentation file for the
23598 new built-in functions.
23600 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23602 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23603 (layout::m_primary_loc): New field.
23604 (layout::layout): Initialize new field. Move location filtering
23605 logic from here to...
23606 (layout::maybe_add_location_range): ...this new method. Add
23607 support for filtering to just the lines already specified by other
23609 (layout::will_show_line_p): New method.
23610 (gcc_rich_location::add_location_if_nearby): New method.
23611 (selftest::test_add_location_if_nearby): New test function.
23612 (selftest::diagnostic_show_locus_c_tests): Call it.
23613 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23616 2017-07-11 Tom de Vries <tom@codesourcery.com>
23618 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23619 (bb_first_real_insn): New function.
23620 (nvptx_single): Add extra initialization of broadcasted condition
23623 2017-07-11 Nathan Sidwell <nathan@acm.org>
23625 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23627 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23629 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23630 Binutils doc as TEXI will mess them up.
23631 * doc/invoke.texi (AVR Options): Same here.
23633 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23635 * config/sparc/sparc.opt (mfix-ut700): New option.
23636 (mfix-gr712rc): Likewise.
23637 (sparc_fix_b2bst): New variable.
23638 * doc/invoke.texi (SPARC options): Document them.
23639 (ARM options): Fix warnings.
23640 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23641 instructions to prevent sequences that can trigger the store-store
23642 errata for certain LEON3FT processors.
23643 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23644 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23645 * config/sparc/sparc.md (fix_b2bst): New attribute.
23646 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23648 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23651 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23653 (*rsqrtsf2_sse): Ditto.
23655 (div<mode>3): Macroize insn from divdf3 and divsf3
23656 using MODEF mode iterator.
23658 2017-07-10 Martin Sebor <msebor@redhat.com>
23660 PR tree-optimization/80397
23661 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23662 instead of testing for equality to INTEGER_TYPE.
23664 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23666 * config.gcc: Remove uclibc from arc target spec.
23668 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23670 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23672 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23675 * lto-wrapper.c (remove_option): New function.
23676 (merge_and_complain): Merge PIC/PIE options more realistically.
23678 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23680 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23684 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23685 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23686 * config.in: Regenerate.
23687 * configure: Regenerate.
23688 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23689 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23690 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23691 (TARGET_GASISR_PROLOGUES): ...target mask.
23692 * common/config/avr/avr-common.c
23693 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23694 Set -mgas-isr-prologues.
23695 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23696 INSERT_PASS_BEFORE for it.
23697 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23698 * config/avr/avr.c (avr_option_override)
23699 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23700 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23701 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23702 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23703 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23705 (make_avr_pass_pre_proep): New function.
23706 (emit_push_sfr) <treg>: Add argument to function and use it
23707 instead of TMP_REG.
23708 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23709 and set machine->gasisr.yes.
23710 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23711 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23712 __gcc_isr.n_pushed to .L__stack_usage.
23713 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23714 (avr_asm_final_postscan_insn): ...this new static function.
23715 * config/avr/avr.h (machine_function)
23716 <is_no_gccisr, use_L__stack_usage>: New fields.
23717 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23718 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23719 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23720 (gasisr, *gasisr): New expander and insn.
23721 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23722 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23723 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23725 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23727 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23730 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23732 Move jump-tables out of .text again.
23735 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23736 (ASM_OUTPUT_ADDR_VEC): New function.
23737 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23738 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23739 INSN_ADDRESSes as asm comment.
23740 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23741 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23742 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23743 * config/avr/avr.md (*tablejump): Adjust comment.
23744 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23745 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23747 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23748 (avr_output_addr_vec): New proto.
23749 (avr_log_t) <insn_addresses>: New field.
23751 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23754 * config/i386/i386.c (ix86_function_arg_advance): Set
23755 outgoing_args_on_stack to true if there are outgoing arguments
23757 (ix86_function_arg): Likewise.
23758 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23759 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23760 * config/i386/i386.h (machine_function): Add
23761 outgoing_args_on_stack.
23763 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23765 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23766 supporting pthreds.
23767 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23769 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23771 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23772 (REAL_H): Remove $(MACHMODE_H).
23773 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23775 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23776 $(MACHMODE_H) and double-int.h.
23777 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23779 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23782 2017-07-07 Andrew Pinski <apinski@cavium.com>
23784 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23785 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23787 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23789 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23790 Add warning if GCC was not configured to link against a GLIBC that
23791 exports the hardware capability bits.
23792 (make_resolver_func): Make resolver function private and not a
23793 COMDAT function. Create the name with clone_function_name instead
23794 of make_unique_name.
23797 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23798 correct operand in doing the split.
23800 2017-07-07 Carl Love <cel@us.ibm.com>
23802 * config/rs6000/rs6000-c: Add support for built-in function
23803 vector unsigned short vec_pack_to_short_fp32 (vector float,
23805 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23806 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23807 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23808 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23809 (convert_4f32_8i16): Add define_expand.
23810 * doc/extend.texi: Update the built-in documentation file for the
23811 new built-in function.
23813 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23815 * config/sparc/m8.md: New file.
23816 * config/sparc/sparc.md: Include m8.md.
23818 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23820 * config/sparc/sparc.opt: New option -mvis4b.
23821 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23822 (sparc_option_override): Handle VIS4B.
23823 (enum sparc_builtins): Define
23824 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23825 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23826 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23827 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23828 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23829 (check_constant_argument): New function.
23830 (sparc_vis_init_builtins): Define builtins
23831 __builtin_vis_dictunpack{8,16,32},
23832 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23833 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23834 __builtin_vis_fpcmpde{8,16,32}shl and
23835 __builtin_vis_fpcmpur{8,16,32}shl.
23836 (sparc_expand_builtin): Check that the constant operands to
23837 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23838 constant and in range.
23839 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23841 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23842 (SPARC_IMM5_P): Likewise.
23843 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23844 (enabled): Handle vis4b.
23845 (UNSPEC_DICTUNPACK): New unspec.
23846 (UNSPEC_FPCMPSHL): Likewise.
23847 (UNSPEC_FPUCMPSHL): Likewise.
23848 (UNSPEC_FPCMPDESHL): Likewise.
23849 (UNSPEC_FPCMPURSHL): Likewise.
23850 (cpu_feature): New CPU feature `vis4b'.
23851 (dictunpack{8,16,32}): New insns.
23852 (FPCSMODE): New mode iterator.
23853 (fpcscond): New code iterator.
23854 (fpcsucond): Likewise.
23855 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23856 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23857 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23858 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23859 * config/sparc/constraints.md: Define constraints `q' for unsigned
23860 2-bit integer constants and `t' for unsigned 5-bit integer
23862 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23864 (imm5_operand_dictunpack16): Likewise.
23865 (imm5_operand_dictunpack32): Likewise.
23866 (imm2_operand): Likewise.
23867 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23868 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23869 ditunpack* and fpcmp*shl builtins.
23871 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23873 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23874 * config.in: Add HAVE_AS_SPARC6 define.
23875 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23877 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23879 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23880 (CPP_CPU_SPEC): Handle m8.
23881 (ASM_CPU_SPEC): Likewise.
23882 * config/sparc/sparc-opts.h (enum processor_type): Add
23884 * config/sparc/sparc.c (m8_costs): New struct.
23885 (sparc_option_override): Handle TARGET_CPU_m8.
23886 (sparc32_initialize_trampoline): Likewise.
23887 (sparc64_initialize_trampoline): Likewise.
23888 (sparc_issue_rate): Likewise.
23889 (sparc_register_move_cost): Likewise.
23890 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23891 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23892 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23893 (CPP_CPU_SPEC): Handle M8.
23894 (ASM_CPU_SPEC): Likewise.
23895 (AS_M8_FLAG): Define.
23896 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23897 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23898 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23900 * configure: Regenerate.
23901 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23904 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23906 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23908 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23909 ("*movdi_insn_sp32"): Do not set v3pipe.
23910 ("*movsi_insn"): Likewise.
23911 ("*movdi_insn_sp64"): Likewise.
23912 ("*movsf_insn"): Likewise.
23913 ("*movdf_insn_sp32"): Likewise.
23914 ("*movdf_insn_sp64"): Likewise.
23915 ("*zero_extendsidi2_insn_sp64"): Likewise.
23916 ("*sign_extendsidi2_insn"): Likewise.
23917 ("*mov<VM32:mode>_insn"): Likewise.
23918 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23919 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23920 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23921 ("<vlop:code><VL:mode>3"): Likewise.
23922 ("*not_<vlop:code><VL:mode>3"): Likewise.
23923 ("*nand<VL:mode>_vis"): Likewise.
23924 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23925 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23926 ("one_cmpl<VL:mode>2"): Likewise.
23927 ("faligndata<VM64:mode>_vis"): Likewise.
23928 ("alignaddrsi_vis"): Likewise.
23929 ("alignaddrdi_vis"): Likweise.
23930 ("alignaddrlsi_vis"): Likewise.
23931 ("alignaddrldi_vis"): Likewise.
23932 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23933 ("bmaskdi_vis"): Likewise.
23934 ("bmasksi_vis"): Likewise.
23935 ("bshuffle<VM64:mode>_vis"): Likewise.
23936 ("cmask8<P:mode>_vis"): Likewise.
23937 ("cmask16<P:mode>_vis"): Likewise.
23938 ("cmask32<P:mode>_vis"): Likewise.
23939 ("pdistn<P:mode>_vis"): Likewise.
23940 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23942 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23944 * config/sparc/sparc.md ("subtype"): New insn attribute.
23945 ("*wrgsr_sp64"): Set insn subtype.
23946 ("*rdgsr_sp64"): Likewise.
23947 ("alignaddrsi_vis"): Likewise.
23948 ("alignaddrdi_vis"): Likewise.
23949 ("alignaddrlsi_vis"): Likewise.
23950 ("alignaddrldi_vis"): Likewise.
23951 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23952 ("fexpand_vis"): Likewise.
23953 ("fpmerge_vis"): Likewise.
23954 ("faligndata<VM64:mode>_vis"): Likewise.
23955 ("bshuffle<VM64:mode>_vis"): Likewise.
23956 ("cmask8<P:mode>_vis"): Likewise.
23957 ("cmask16<P:mode>_vis"): Likewise.
23958 ("cmask32<P:mode>_vis"): Likewise.
23959 ("fchksm16_vis"): Likewise.
23960 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23961 ("fmean16_vis"): Likewise.
23962 ("fp<plusminus_insn>64_vis"): Likewise.
23963 ("<plusminus_insn>v8qi3"): Likewise.
23964 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23965 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23966 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23967 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23968 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23969 ("*movqi_insn"): Likewise.
23970 ("*movhi_insn"): Likewise.
23971 ("*movsi_insn"): Likewise.
23972 ("movsi_pic_gotdata_op"): Likewise.
23973 ("*movdi_insn_sp32"): Likewise.
23974 ("*movdi_insn_sp64"): Likewise.
23975 ("movdi_pic_gotdata_op"): Likewise.
23976 ("*movsf_insn"): Likewise.
23977 ("*movdf_insn_sp32"): Likewise.
23978 ("*movdf_insn_sp64"): Likewise.
23979 ("*zero_extendhisi2_insn"): Likewise.
23980 ("*zero_extendqihi2_insn"): Likewise.
23981 ("*zero_extendqisi2_insn"): Likewise.
23982 ("*zero_extendqidi2_insn"): Likewise.
23983 ("*zero_extendhidi2_insn"): Likewise.
23984 ("*zero_extendsidi2_insn_sp64"): Likewise.
23985 ("ldfsr"): Likewise.
23986 ("prefetch_64"): Likewise.
23987 ("prefetch_32"): Likewise.
23988 ("tie_ld32"): Likewise.
23989 ("tie_ld64"): Likewise.
23990 ("*tldo_ldub_sp32"): Likewise.
23991 ("*tldo_ldub1_sp32"): Likewise.
23992 ("*tldo_ldub2_sp32"): Likewise.
23993 ("*tldo_ldub_sp64"): Likewise.
23994 ("*tldo_ldub1_sp64"): Likewise.
23995 ("*tldo_ldub2_sp64"): Likewise.
23996 ("*tldo_ldub3_sp64"): Likewise.
23997 ("*tldo_lduh_sp32"): Likewise.
23998 ("*tldo_lduh1_sp32"): Likewise.
23999 ("*tldo_lduh_sp64"): Likewise.
24000 ("*tldo_lduh1_sp64"): Likewise.
24001 ("*tldo_lduh2_sp64"): Likewise.
24002 ("*tldo_lduw_sp32"): Likewise.
24003 ("*tldo_lduw_sp64"): Likewise.
24004 ("*tldo_lduw1_sp64"): Likewise.
24005 ("*tldo_ldx_sp64"): Likewise.
24006 ("*mov<VM32:mode>_insn"): Likewise.
24007 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24008 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24010 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24012 * config/sparc/sparc.md ("type"): New insn type viscmp.
24013 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24015 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24016 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24017 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24018 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24020 ("n7_vis_logical_11cycle"): Likewise.
24021 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24022 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24023 * config/sparc/niagara.md ("niag_vis"): Likewise.
24024 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24025 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24027 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24029 * config/sparc/sparc.md: New instruction type `bmask'.
24030 (bmaskdi_vis): Use the `bmask' type.
24031 (bmasksi_vis): Likewise.
24032 * config/sparc/ultra3.md (us3_array): Likewise.
24033 * config/sparc/niagara7.md (n7_array): Likewise.
24034 * config/sparc/niagara4.md (n4_array): Likewise.
24035 * config/sparc/niagara2.md (niag2_vis): Likewise.
24036 (niag3_vis): Likewise.
24037 * config/sparc/niagara.md (niag_vis): Likewise.
24039 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24041 * ipa-comdats.c: Remove optimize check from gate.
24042 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24043 for functions not optimized.
24044 (ipa_fn_summary_read): Skip optimize check.
24045 (ipa_fn_summary_write): Likewise.
24046 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24048 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24050 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24051 (check_callers): Give up on caller which is not optimized.
24052 (inline_small_functions): Likewise.
24053 (ipa_inline): Do not give up when not optimizing.
24054 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24055 away unoptimizes cdtors.
24056 (whole_program_function_and_variable_visibility): Do
24057 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24058 * ipa.c (process_references): Do not check optimize.
24059 (symbol_table::remove_unreachable_nodes): Update optimize check.
24060 (set_writeonly_bit): Update optimize check.
24061 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24062 (pass_ipa_single_use::gate): Remove.
24064 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24066 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24067 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24068 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24069 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24070 permute_load, permute_store, adjust_extract, adjust_splat,
24071 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24072 replace_swap_with_copy, dump_swap_insn_table,
24073 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24074 recombine_lvx_pattern, recombine_stvx_pattern,
24075 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24076 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24077 to file rs6000-p8swap.c.
24078 * config/rs6000/rs6000-p8swap.c: New file.
24079 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24080 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24081 and rs6000*-*-* targets.
24083 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24085 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24087 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24089 * lto-wrapper.c (merge_and_complain): Do not merge
24090 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24091 fsigned_zeros, ftrapping_math, fwrapv.
24092 (append_compiler_options): Do not track these options.
24093 (append_linker_options): Likewie
24095 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24097 * cgraphunit.c (cgraph_node::finalize_function): When
24098 !flag_toplevel_reorde set no_reorder flag.
24099 (varpool_node::finalize_decl): Likewise.
24100 (symbol_table::compile): Drop no toplevel reorder path.
24102 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24104 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24105 edges; zero probability is not better than uninitialized.
24107 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24109 * asan.h (asan_sanitize_allocas_p): Declare.
24110 * asan.c (asan_sanitize_allocas_p): New function.
24111 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24112 (handle_builtin_alloca): Likewise.
24113 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24114 if !asan_sanitize_allocas_p.
24115 * params.def (asan-instrument-allocas): Add new option.
24116 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24117 * opts.c (common_handle_option): Disable allocas sanitization for
24120 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24122 * asan.c: Include gimple-fold.h.
24123 (get_last_alloca_addr): New function.
24124 (handle_builtin_stackrestore): Likewise.
24125 (handle_builtin_alloca): Likewise.
24126 (asan_emit_allocas_unpoison): Likewise.
24127 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24128 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24129 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24130 (instrument_builtin_call): Pass gimple iterator to
24131 get_mem_refs_of_builtin_call.
24132 (last_alloca_addr): New global.
24133 * asan.h (asan_emit_allocas_unpoison): Declare.
24134 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24135 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24136 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24137 if function calls alloca.
24138 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24139 * gimple-fold.h (replace_call_with_value): Declare.
24140 * internal-fn.c: Include asan.h.
24141 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24142 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24144 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24146 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24147 (C_SELFTEST_FLAGS): New.
24148 (CPP_SELFTEST_FLAGS): New.
24149 (SELFTEST_DEPS): New, from deps of s-selftest.
24150 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24151 (CPP_SELFTEST_DEPS): New.
24152 (selftest): Add dependency on s-selftest-c++.
24153 (s-selftest): Rename to...
24154 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24155 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24156 than SELFTEST_FLAGS.
24157 (selftest-gdb): Rename to...
24158 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24160 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24161 (selftest-valgrind): Rename to...
24162 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24164 (selftest-valgrind): Reintroduce as an alias for
24165 selftest-c-valgrind.
24166 (s-selftest-c++): New.
24167 (selftest-c++-gdb): New.
24168 (selftest-c++-valgrind): New.
24170 2017-07-06 Olivier Hainque <hainque@adacore.com>
24172 * gcc.c (process_command): When deciding if undefined variables
24173 should be ignored when processing specs, accept "gcc -v" as well.
24175 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24177 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24178 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24180 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24182 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24183 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24184 * config/arm/arm-tables.opt: Regenerate.
24185 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24187 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24189 2017-07-06 Carl Love <cel@us.ibm.com>
24191 * ChangeLog: Clean up from mid air collision
24193 2017-07-06 Carl Love <cel@us.ibm.com>
24195 * config/rs6000/rs6000-c.c: Add support for built-in functions
24196 vector signed int vec_subc (vector signed int, vector signed int);
24197 vector signed __int128 vec_subc (vector signed __int128,
24198 vector signed __int128);
24199 vector unsigned __int128 vec_subc (vector unsigned __int128,
24200 vector unsigned __int128);
24201 vector signed int vec_sube (vector signed int, vector signed int,
24202 vector signed int);
24203 vector unsigned int vec_sube (vector unsigned int,
24204 vector unsigned int,
24205 vector unsigned int);
24206 vector signed __int128 vec_sube (vector signed __int128,
24207 vector signed __int128,
24208 vector signed__int128);
24209 vector unsigned __int128 vec_sube (vector unsigned __int128,
24210 vector unsigned __int128,
24211 vector unsigned __int128);
24212 vector signed int vec_subec (vector signed int, vector signed int,
24213 vector signed int);
24214 vector unsigned int vec_subec (vector unsigned int,
24215 vector unsigned int,
24216 vector unsigned int);
24217 vector signed __int128 vec_subec (vector signed __int128,
24218 vector signed __int128,
24219 vector signed__int128);
24220 vector unsigned __int128 vec_subec (vector unsigned __int128,
24221 vector unsigned __int128,
24222 vector unsigned __int128);
24223 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24224 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24225 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24226 BU_ALTIVEC_OVERLOAD_X definitions.
24227 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24228 * doc/extend.texi: Update the built-in documentation file for the new
24229 built-in functions.
24231 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24234 * diagnostic-show-locus.c (layout::layout): Use start and finish
24235 spelling location for the start and finish of each range.
24236 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24237 Add unused aspect param.
24238 * input.c (expand_location_1): Add "aspect" param, and use it
24239 to access the correct part of the location.
24240 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24242 (expand_location_to_spelling_point): Likewise.
24243 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24244 param, and pass it to expand_location_1.
24246 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24248 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24249 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24250 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24251 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24252 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24253 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24254 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24255 _mm_maskz_getmant_ss): New intrinsics.
24256 (__builtin_ia32_getexpss128_mask): Changed to ...
24257 __builtin_ia32_getexpss128_round ... this.
24258 (__builtin_ia32_getexpsd128_mask): Changed to ...
24259 __builtin_ia32_getexpsd128_round ... this.
24260 * config/i386/i386-builtin-types.def
24261 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24262 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24263 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24264 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24265 __builtin_ia32_getmantss_mask_round): New builtins.
24266 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24267 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24268 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24269 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24270 * config/i386/sse.md
24271 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24272 avx512f_sgetexp<mode><mask_scalar_name>
24273 <round_saeonly_scalar_name> ... this.
24274 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24275 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24276 vgetexp<ssescalarmodesuffix>
24277 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24278 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24279 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24280 avx512f_vgetmant<mode><mask_scalar_name>
24281 <round_saeonly_scalar_name> ... this.
24282 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24283 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24284 vgetmant<ssescalarmodesuffix>
24285 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24286 %0<mask_scalar_operand4>, %1, %2
24287 <round_saeonly_scalar_mask_op4>, %3} ... this.
24288 * config/i386/subst.md (mask_scalar_operand4,
24289 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24290 round_saeonly_scalar_nimm_predicate): New subst attributes.
24292 2017-07-06 Julia Koval <julia.koval@intel.com>
24294 * config/i386/i386.c (ix86_erase_embedded_rounding):
24295 Remove code for old rounding pattern.
24297 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24299 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24301 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24303 * doc/sourcebuild.texi (Test Directives, Variants of
24304 dg-require-support): Add documentation for dg-require-stack-check.
24306 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24308 * config/i386/subst.md (mask_scalar, round_scalar,
24309 round_saeonly_scalar): New meta-templates.
24310 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24311 round_scalar_mask_operand3, round_scalar_mask_op3,
24312 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24313 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24314 round_saeonly_scalar_constraint,
24315 round_saeonly_scalar_prefix): New subst attribute.
24316 * config/i386/sse.md
24317 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24318 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24319 <round_scalar_name> ... this.
24320 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24321 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24322 <round_scalar_name> ... this.
24323 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24324 <sse>_vm<code><mode>3<mask_scalar_name>
24325 <round_saeonly_scalar_name> ... this.
24326 (v<plusminus_mnemonic><ssescalarmodesuffix>
24327 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24328 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24329 v<plusminus_mnemonic><ssescalarmodesuffix>
24330 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24331 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24332 (v<multdiv_mnemonic><ssescalarmodesuffix>
24333 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24334 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24335 v<multdiv_mnemonic><ssescalarmodesuffix>
24336 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24337 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24338 (v<maxmin_float><ssescalarmodesuffix>
24339 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24340 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24341 v<maxmin_float><ssescalarmodesuffix>
24342 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24343 %0<mask_scalar_operand3>, %1, %<iptr>2
24344 <round_saeonly_scalar_mask_op3>} ... this.
24346 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24348 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24349 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24351 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24352 Alan Hayward <alan.hayward@arm.com>
24353 David Sherwood <david.sherwood@arm.com>
24355 * combine.c (simplify_if_then_else): Remove "enum" before
24357 * compare-elim.c (can_eliminate_compare): Likewise.
24358 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24360 (aarch64_lookup_simd_builtin_type): Likewise.
24361 (aarch64_simd_builtin_type): Likewise.
24362 (aarch64_init_simd_builtin_types): Likewise.
24363 (aarch64_simd_expand_args): Likewise.
24364 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24366 (aarch64_reverse_mask): Likewise.
24367 (aarch64_simd_emit_reg_reg_move): Likewise.
24368 (aarch64_gen_adjusted_ldpstp): Likewise.
24369 (aarch64_ccmp_mode_to_code): Likewise.
24370 (aarch64_operands_ok_for_ldpstp): Likewise.
24371 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24372 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24374 (aarch64_min_divisions_for_recip_mul): Likewise.
24375 (aarch64_reassociation_width): Likewise.
24376 (aarch64_get_condition_code_1): Likewise.
24377 (aarch64_simd_emit_reg_reg_move): Likewise.
24378 (aarch64_simd_attr_length_rglist): Likewise.
24379 (aarch64_reverse_mask): Likewise.
24380 (aarch64_operands_ok_for_ldpstp): Likewise.
24381 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24382 (aarch64_gen_adjusted_ldpstp): Likewise.
24383 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24385 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24386 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24387 (arm_lookup_simd_builtin_type): Likewise.
24388 (arm_simd_builtin_type): Likewise.
24389 (arm_init_simd_builtin_types): Likewise.
24390 (arm_expand_builtin_args): Likewise.
24391 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24392 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24393 (ft32_setup_incoming_varargs): Likewise.
24394 (ft32_function_arg): Likewise.
24395 (ft32_function_arg_advance): Likewise.
24396 (ft32_pass_by_reference): Likewise.
24397 (ft32_arg_partial_bytes): Likewise.
24398 (ft32_valid_pointer_mode): Likewise.
24399 (ft32_addr_space_pointer_mode): Likewise.
24400 (ft32_addr_space_legitimate_address_p): Likewise.
24401 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24403 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24404 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24405 (iamcu_alignment): Likewise.
24406 (canonicalize_vector_int_perm): Likewise.
24407 (ix86_noce_conversion_profitable_p): Likewise.
24408 (ix86_mpx_bound_mode): Likewise.
24409 (ix86_operands_ok_for_move_multiple): Likewise.
24410 * config/microblaze/microblaze-protos.h
24411 (microblaze_expand_conditional_branch_reg): Likewise.
24412 * config/microblaze/microblaze.c
24413 (microblaze_expand_conditional_branch_reg): Likewise.
24414 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24416 (rs6000_reassociation_width): Likewise.
24417 (rs6000_invalid_binary_op): Likewise.
24418 (fusion_p9_p): Likewise.
24419 (emit_fusion_p9_load): Likewise.
24420 (emit_fusion_p9_store): Likewise.
24421 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24423 (riscv_hard_regno_mode_ok_p): Likewise.
24424 (riscv_address_insns): Likewise.
24425 (riscv_split_symbol): Likewise.
24426 (riscv_legitimize_move): Likewise.
24427 (riscv_function_value): Likewise.
24428 (riscv_hard_regno_nregs): Likewise.
24429 (riscv_expand_builtin): Likewise.
24430 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24431 (riscv_build_integer): Likewise.
24432 (riscv_split_integer): Likewise.
24433 (riscv_legitimate_constant_p): Likewise.
24434 (riscv_cannot_force_const_mem): Likewise.
24435 (riscv_regno_mode_ok_for_base_p): Likewise.
24436 (riscv_valid_base_register_p): Likewise.
24437 (riscv_valid_offset_p): Likewise.
24438 (riscv_valid_lo_sum_p): Likewise.
24439 (riscv_classify_address): Likewise.
24440 (riscv_legitimate_address_p): Likewise.
24441 (riscv_address_insns): Likewise.
24442 (riscv_load_store_insns): Likewise.
24443 (riscv_force_binary): Likewise.
24444 (riscv_split_symbol): Likewise.
24445 (riscv_force_address): Likewise.
24446 (riscv_legitimize_address): Likewise.
24447 (riscv_move_integer): Likewise.
24448 (riscv_legitimize_const_move): Likewise.
24449 (riscv_legitimize_move): Likewise.
24450 (riscv_address_cost): Likewise.
24451 (riscv_subword): Likewise.
24452 (riscv_output_move): Likewise.
24453 (riscv_canonicalize_int_order_test): Likewise.
24454 (riscv_emit_int_order_test): Likewise.
24455 (riscv_function_arg_boundary): Likewise.
24456 (riscv_pass_mode_in_fpr_p): Likewise.
24457 (riscv_pass_fpr_single): Likewise.
24458 (riscv_pass_fpr_pair): Likewise.
24459 (riscv_get_arg_info): Likewise.
24460 (riscv_function_arg): Likewise.
24461 (riscv_function_arg_advance): Likewise.
24462 (riscv_arg_partial_bytes): Likewise.
24463 (riscv_function_value): Likewise.
24464 (riscv_pass_by_reference): Likewise.
24465 (riscv_setup_incoming_varargs): Likewise.
24466 (riscv_print_operand): Likewise.
24467 (riscv_elf_select_rtx_section): Likewise.
24468 (riscv_save_restore_reg): Likewise.
24469 (riscv_for_each_saved_reg): Likewise.
24470 (riscv_register_move_cost): Likewise.
24471 (riscv_hard_regno_mode_ok_p): Likewise.
24472 (riscv_hard_regno_nregs): Likewise.
24473 (riscv_class_max_nregs): Likewise.
24474 (riscv_memory_move_cost): Likewise.
24475 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24476 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24477 (rl78_addr_space_address_mode): Likewise.
24478 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24480 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24481 (rs6000_reassociation_width): Likewise.
24482 (rs6000_invalid_binary_op): Likewise.
24483 (fusion_p9_p): Likewise.
24484 (emit_fusion_p9_load): Likewise.
24485 (emit_fusion_p9_store): Likewise.
24486 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24487 (ok_for_simple_move_operands): Likewise.
24488 (ok_for_simple_move_strict_operands): Likewise.
24489 (ok_for_simple_arith_logic_operands): Likewise.
24490 (visium_legitimize_reload_address): Likewise.
24491 (visium_select_cc_mode): Likewise.
24492 (output_cbranch): Likewise.
24493 (visium_split_double_move): Likewise.
24494 (visium_expand_copysign): Likewise.
24495 (visium_expand_int_cstore): Likewise.
24496 (visium_expand_fp_cstore): Likewise.
24497 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24498 (visium_function_arg): Likewise.
24499 (visium_function_arg_advance): Likewise.
24500 (visium_libcall_value): Likewise.
24501 (visium_setup_incoming_varargs): Likewise.
24502 (visium_legitimate_constant_p): Likewise.
24503 (visium_legitimate_address_p): Likewise.
24504 (visium_legitimize_address): Likewise.
24505 (visium_secondary_reload): Likewise.
24506 (visium_register_move_cost): Likewise.
24507 (visium_memory_move_cost): Likewise.
24508 (prepare_move_operands): Likewise.
24509 (ok_for_simple_move_operands): Likewise.
24510 (ok_for_simple_move_strict_operands): Likewise.
24511 (ok_for_simple_arith_logic_operands): Likewise.
24512 (visium_function_value_1): Likewise.
24513 (rtx_ok_for_offset_p): Likewise.
24514 (visium_legitimize_reload_address): Likewise.
24515 (visium_split_double_move): Likewise.
24516 (visium_expand_copysign): Likewise.
24517 (visium_expand_int_cstore): Likewise.
24518 (visium_expand_fp_cstore): Likewise.
24519 (visium_split_cstore): Likewise.
24520 (visium_select_cc_mode): Likewise.
24521 (visium_split_cbranch): Likewise.
24522 (output_cbranch): Likewise.
24523 (visium_print_operand_address): Likewise.
24524 * expmed.c (flip_storage_order): Likewise.
24525 * expmed.h (emit_cstore): Likewise.
24526 (flip_storage_order): Likewise.
24527 * genrecog.c (validate_pattern): Likewise.
24528 * hsa-gen.c (gen_hsa_addr): Likewise.
24529 * internal-fn.c (expand_arith_overflow): Likewise.
24530 * ira-color.c (allocno_copy_cost_saving): Likewise.
24531 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24532 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24533 (process_invariant_for_inheritance): Likewise.
24534 * lra-eliminations.c (move_plus_up): Likewise.
24535 * omp-low.c (lower_oacc_reductions): Likewise.
24536 * simplify-rtx.c (simplify_subreg): Likewise.
24537 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24538 (TARGET_CHKP_BOUND_MODE): Likewise..
24539 * targhooks.c (default_chkp_bound_mode): Likewise.
24540 (default_setup_incoming_vararg_bounds): Likewise.
24541 * targhooks.h (default_chkp_bound_mode): Likewise.
24542 (default_setup_incoming_vararg_bounds): Likewise.
24543 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24544 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24545 (have_whole_vector_shift): Likewise.
24546 * tree-vect-stmts.c (vectorizable_load): Likewise.
24547 * doc/tm.texi: Regenerate.
24549 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24551 Graceful degrade if Binutils PR21472 is not available.
24554 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24555 .rodata in flash test fails.
24556 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24557 * confgure: Regenerate.
24558 * config.in: Regenerate.
24559 * config/avr/avr.c (avr_asm_named_section)
24560 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24561 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24562 (avr_asm_init_sections): Same.
24564 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24566 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24567 (fma<VH:mode>4_intrinsic): Likewise.
24568 (*fmsub<VCVTF:mode>4): Likewise.
24569 (*fmsub<VH:mode>4_intrinsic): Likewise.
24571 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24574 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24575 Don't depend on "optimize > 0".
24576 (out_movhi_r_mr, out_movqi_mr_r): Same.
24577 (out_movhi_mr_r, out_movqi_r_mr): Same.
24578 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24579 io_address_operand on "optimize > 0".
24581 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24583 * tree-loop-distribution.c: Add general explanantion on the pass.
24584 (generate_loops_for_partition): Mark distributed loop.
24585 (pg_add_dependence_edges): New parameter. Handle alias data
24586 dependence specially and record it in the parameter if asked.
24587 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24588 (init_partition_graph_vertices, add_partition_graph_edge): New.
24589 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24590 (free_partition_graph_vdata, build_partition_graph): New.
24591 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24592 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24593 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24594 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24595 (version_for_distribution_p, finalize_partitions): New.
24596 (distribute_loop): Handle alias data dependence specially. Factor
24597 out loop fusion code as functions and call these functions.
24599 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24601 * tree-loop-distribution.c (classify_partition): New parameter and
24602 better handle reduction statement.
24603 (rdg_build_partitions): Revise comment.
24604 (distribute_loop): Compute statements in all partitions and pass it
24605 to classify_partition.
24607 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24609 * tree-loop-distribution.c (enum partition_type): New.
24610 (struct partition): New field type.
24611 (partition_merge_into): Add parameter. Update partition type.
24612 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24613 (build_rdg_partition_for_vertex): Compute partition type.
24614 (rdg_build_partitions): Dump partition type.
24615 (distribute_loop): Update calls to partition_merge_into.
24617 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24619 * tree-loop-distribution.c (struct ddr_hasher): New.
24620 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24622 (classify_partition): Call get_data_dependence.
24623 (pg_add_dependence_edges): Ditto.
24624 (distribute_loop): Release data dependence hash table.
24626 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24628 * tree-loop-distribution.c (ref_base_address): Delete.
24629 (similar_memory_accesses): Rename ...
24630 (share_memory_accesses): ... to this. Check if partitions access
24631 the same memory reference.
24632 (distribute_loop): Call share_memory_accesses.
24634 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24636 * tree-loop-distribution.c (struct partition): New field recording
24637 its data reference.
24638 (partition_alloc, partition_free): Init and release data refs.
24639 (partition_merge_into): Merge data refs.
24640 (build_rdg_partition_for_vertex): Collect data refs for partition.
24641 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24643 (distribute_loop): Remve data refs from vertice data of partition
24646 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24648 * tree-loop-distribution.c (params.h): Include header file.
24649 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24650 (datarefs_vec): New global var.
24651 (create_rdg_vertices): Use datarefs_vec directly.
24652 (free_rdg): Don't free data references.
24653 (build_rdg): Update use. Don't free data references.
24654 (distribute_loop): Compute global variable for data references.
24655 Bail out if there are too many data references.
24657 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24659 * tree-loop-distribution.c (loop_nest): New global var.
24660 (build_rdg): Use loop directly, rather than loop nest.
24661 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24663 (distribute_loop): Compute global variable loop nest. Update use.
24665 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24667 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24668 (partition_merge_into): New parameter. Dump reason for fusion.
24669 (distribute_loop): Update use of partition_merge_into.
24671 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24673 * tree-loop-distribution.c (bb_top_order_index): New.
24674 (bb_top_order_index_size, bb_top_order_cmp): New.
24675 (stmts_from_loop): Use topological order.
24676 (pass_loop_distribution::execute): Compute and release topological
24677 order for basic blocks.
24679 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24681 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24684 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24686 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24687 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24688 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24689 * internal-fn.def (LOOP_DIST_ALIAS): New.
24690 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24691 (fold_loop_internal_call): ... this.
24692 (vect_loop_dist_alias_call): New function.
24693 (set_uid_loop_bbs): Call fold_loop_internal_call.
24694 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24697 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24700 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24701 Require dead FLAGS_REG at the beginning of a peephole.
24703 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24706 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24707 arguments in the call to __builtin_ia32_sbb_u32.
24708 (_subborrow_u64): Swap _X and _Y arguments in the call to
24709 __builtin_ia32_sbb_u64.
24711 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24714 * tree-vrp.c (compare_assert_loc): Turn into a function template
24715 with stable template parameter. Only test if a->e is NULL,
24716 !a->e == !b->e has been verified already. Use e == NULL or
24717 e != NULL instead of e or ! e tests. If stable is true, don't use
24718 iterative_hash_expr, on the other side allow a or b or both NULL
24719 and sort the NULLs last.
24720 (process_assert_insertions): Sort using compare_assert_loc<false>
24721 instead of compare_assert_loc, later sort using
24722 compare_assert_loc<true> before calling process_assert_insertions_for
24723 in a loop. Use break instead of continue once seen NULL pointer.
24725 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24727 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24728 Cortex-R7 and Cortex-R8 processors.
24730 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24732 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24733 uninitialized while src is not.
24735 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24737 * common/config/arm/arm-common.c: Adjust include path for
24739 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24740 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24741 (arm-cpu-data.h): Likewise.
24742 (arm-cpu-cdata.h): Likewise.
24743 * config/arm/arm-cpu.h: Delete.
24744 * config/arm/arm-cpu-cdata.h: Delete.
24745 * config/arm/arm-cpu-data.h: Delete.
24747 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24749 * config/arm/arm-cpus.in (cortex-a55): New.
24750 (cortex-a75): Likewise.
24751 (cortex-a75.cortex-a55): Likewise.
24752 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24754 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24755 * config/arm/arm-cpu-cdata.h: Regenerate.
24756 * config/arm/arm-cpu-data.h: Regenerate.
24757 * config/arm/arm-cpu.h: Regenerate.
24758 * config/arm/arm-tables.opt: Regenerate.
24759 * config/arm/arm-tune.md: Regenerate.
24761 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24763 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24765 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24767 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24770 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24772 PR tree-optimization/81292
24773 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24774 full_string_p, also call adjust_related_strinfos if the adjustment
24775 is simple, otherwise invalidate related strinfos.
24777 2017-07-04 Martin Liska <mliska@suse.cz>
24780 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24781 newly created variable as DECL_IGNORED_P.
24783 2017-07-04 Martin Liska <mliska@suse.cz>
24786 * ipa-inline.c (inline_small_functions):
24787 Use xstrdup_for_dump.
24789 2017-07-04 Tom de Vries <tom@codesourcery.com>
24791 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24793 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24796 * config/darwin.c (darwin_function_switched_text_sections):
24799 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24801 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24803 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24805 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24807 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24809 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24810 min_profitable_iters, and th as inclusive lower bounds.
24811 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24812 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24813 for min_profitable_iters and min_profitable_estimate.
24814 (vect_transform_loop): Treat th as an inclusive lower bound.
24815 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24817 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24820 * config/darwin.c (darwin_function_switched_text_sections):
24821 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24822 in two pieces, and suppress the use of buf.
24824 2017-07-03 Nathan Sidwell <nathan@acm.org>
24826 * hash-table.h (hash_table_mod1): Fix indentation.
24828 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24830 PR middle-end/81290
24831 * predict.c (force_edge_cold): Be more careful about propagation
24833 * profile-count.h (profile_probability::guessed,
24834 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24836 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24838 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24840 * doc/invoke.texi (rcpc architecture extension): Document it.
24842 2017-07-03 Richard Biener <rguenther@suse.de>
24844 PR tree-optimization/60510
24845 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24846 the scalar reduction PHI and use it.
24847 (vectorizable_reduction): Properly guard the single_defuse_cycle
24848 path for non-SLP reduction chains where we cannot use it.
24849 Rework reduc_def/index and vector type deduction. Rework
24850 vector operand gathering during reduction op code-gen.
24851 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24852 chains dissolve the chain and leave it to non-SLP reduction
24855 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24857 * tree-data-ref.h (dr_alignment): Declare.
24858 * tree-data-ref.c (dr_alignment): New function.
24859 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24860 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24862 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24864 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24866 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24867 and base_misalignment fields.
24868 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24869 * tree-data-ref.c: Include builtins.h.
24870 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24871 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24872 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24873 * tree-vect-data-refs.c: Include tree-cfg.h.
24874 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24875 fields instead of calculating an alignment here.
24876 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24877 innermost_loop_behavior fields.
24879 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24881 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24883 (DR_STEP_ALIGNMENT): New macro.
24884 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24885 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24886 (create_data_ref): Print it.
24887 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24888 to tell whether the step preserves vector (mis)alignment.
24889 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24890 Move the check for an integer step and generalise to all INTEGER_CST.
24891 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24892 Print the outer step alignment.
24894 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24896 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24897 with offset_alignment.
24898 (DR_ALIGNED_TO): Delete.
24899 (DR_OFFSET_ALIGNMENT): New macro.
24900 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24901 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24902 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24903 (create_data_ref): Likewise.
24904 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24905 (vect_analyze_data_refs): Likewise.
24906 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24907 creating dummy innermost behavior.
24909 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24911 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24912 with a "innermost_loop_behavior *" and refeence tree.
24913 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24914 (create_data_ref): Update call accordingly.
24915 * tree-predcom.c (find_looparound_phi): Likewise.
24917 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24919 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24920 fields with dr_wrt_vec_loop.
24921 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24922 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24923 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24924 (vect_dr_behavior): New function.
24925 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24926 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24927 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24928 track whether the step preserves the misalignment.
24929 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24930 Use vect_dr_behavior.
24931 (vect_setup_realignment): Update call accordingly.
24932 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24933 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24934 call to vect_create_addr_base_for_vector_ref.
24935 (vect_create_cond_for_align_checks): Likewise.
24936 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24937 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24938 (vect_recog_mask_conversion_pattern): Likewise.
24939 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24940 (new_stmt_vec_info): Remove redundant zeroing.
24942 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24944 * common/config/arm/arm-common.c (arm_be8_option): New function.
24945 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24946 (ISA_ARMv6): Add isa_bit_be8.
24947 * config/arm/arm.h (arm_be8_option): Add prototype.
24948 (BE8_SPEC_FUNCTION): New define.
24949 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24950 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24951 (mlittle-endian): Similarly.
24952 (mbe8, mbe32): New options.
24953 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24954 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24956 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24958 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24960 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24962 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24963 (cleanup_tree_cfg_bb): Use it.
24964 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24966 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24968 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24971 * loop-doloop.c (add_test): Update profile.
24973 2017-07-03 Martin Liska <mliska@suse.cz>
24976 * sanopt.c (rewrite_usage_of_param): New function.
24977 (sanitize_rewrite_addressable_params): Likewise.
24978 (pass_sanopt::execute): Call rewrite_usage_of_param.
24980 2017-07-03 Richard Biener <rguenther@suse.de>
24982 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24983 back to using VIEW_CONVERT_EXPR.
24985 2017-07-03 Martin Liska <mliska@suse.cz>
24988 * doc/extend.texi: Document when a resolver function is
24989 generated for target_clones.
24991 2017-07-03 Martin Liska <mliska@suse.cz>
24993 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24994 and shadow memory of auto variables which are subject of
24995 use-after-scope sanitization.
24996 (asan_expand_mark_ifn): Add do set only when is_poison.
24998 2016-07-03 Richard Biener <rguenther@suse.de>
25000 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25002 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25003 (vectorizable_reduction): Perform reduction PHI creation when
25004 visiting a reduction PHI and adjust and simplify code generation
25005 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25006 (vect_transform_loop): Visit reduction PHIs.
25007 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25008 defs into the SLP tree.
25009 (vect_build_slp_tree): Reduction defs terminate the recursion.
25010 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25012 (vect_get_vec_defs_for_stmt_copy): Export.
25013 (vect_get_vec_defs): Likewise.
25014 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25016 (vect_get_vec_defs_for_stmt_copy): Declare.
25017 (vect_get_vec_defs): Likewise.
25019 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25021 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25022 parameter with a "loop" parameter and use it instead of the
25023 loop containing DR_STMT. Don't check simple_iv when doing
25024 BB analysis. Describe the two analysis modes in the comment.
25026 2017-07-03 Tom de Vries <tom@codesourcery.com>
25028 PR tree-optimization/69468
25029 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25030 (find_same_succ_bb): Handle ignore_edge_flags.
25032 2017-07-03 Tom de Vries <tom@codesourcery.com>
25034 PR tree-optimization/81192
25035 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25037 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25039 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25041 2017-07-03 Tom de Vries <tom@codesourcery.com>
25043 PR tree-optimization/81192
25044 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25045 BB_SAME_SUCC (bb) == NULL.
25047 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25049 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25052 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25054 * dumpfile.c: Include profile-count.h
25055 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25057 (insert_cond_bb): Update profile.
25058 * tree-cfg.h (insert_cond_bb): Update prototype.
25059 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25060 * tree-dump.c: Do not include tree-cfg.
25062 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25064 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25066 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25068 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25071 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25073 * tree-complex.c (expand_complex_div_wide): update profile.
25075 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25076 Alan Hayward <alan.hayward@arm.com>
25077 David Sherwood <david.sherwood@arm.com>
25079 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25080 (CORETYPES_H): New define.
25081 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25082 (insn-modes-inline.h, s-modes-inline-h): New rules.
25083 (generated_files): Add insn-modes-inline.h.
25084 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25085 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25086 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25087 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25088 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25089 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25090 (build/gencodes.o, build/genconditions.o): Likewise.
25091 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25092 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25093 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25094 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25095 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25096 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25097 * coretypes.h: Include everything up to real.h for generators.
25098 Include insn-modes.h first. Include wide-int-print.h after
25099 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25100 * machmode.h: Don't include insn-modes.h here.
25101 * function-tests.c: Remove includes of signop.h, machmode.h,
25102 double-int.h and wide-int.h.
25104 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25106 * optc-save-gen.awk: Likewise.
25107 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25108 * godump.c: Remove include of wide-int-print.h.
25109 * pretty-print.h: Likewise.
25110 * wide-int-print.cc: Likewise.
25111 * wide-int.cc: Likewise.
25112 * hash-map-tests.c: Remove include of signop.h.
25113 * hash-set-tests.c: Likewise.
25114 * rtl-tests.c: Likewise.
25115 * mkconfig.sh: Remove include of machmode.h.
25116 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25118 (emit_insn_modes_inline_h): ...this new function. Emit the code
25119 into an insn-modes-inline.h header file, adding appropriate
25120 include guards and end comments.
25121 (emit_insn_modes_c_header): Remove include of machmode.h.
25122 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25124 (main): Handle -i flag and call emit_insn_modes_inline_h when
25127 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25129 * tree-ssa-strlen.c (strinfo): Rename the length field to
25130 nonzero_chars. Add a full_string_p field.
25131 (compare_nonzero_chars, zero_length_string_p): New functions.
25132 (get_addr_stridx): Add an offset_out parameter.
25133 Use compare_nonzero_chars.
25134 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25135 (new_strinfo): Update after above changes to strinfo.
25136 (set_endptr_and_length): Set full_string_p.
25137 (get_string_length): Update after above changes to strinfo.
25138 (unshare_strinfo): Update call to new_strinfo.
25139 (maybe_invalidate): Likewise.
25140 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25141 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25142 as a uhwi instead of an shwi. Update after above changes to
25143 strinfo and new_strinfo.
25144 (zero_length_string): Assert that chainsi contains full strings.
25145 Use zero_length_string_p. Update call to new_strinfo.
25146 (adjust_related_strinfos): Update after above changes to strinfo.
25147 Copy full_string_p from origsi.
25148 (adjust_last_stmt): Use zero_length_string_p.
25149 (handle_builtin_strlen): Update after above changes to strinfo and
25150 new_strinfo. Install the lhs as the string length if the previous
25151 entry didn't describe a full string.
25152 (handle_builtin_strchr): Update after above changes to strinfo
25154 (handle_builtin_strcpy): Likewise.
25155 (handle_builtin_strcat): Likewise.
25156 (handle_builtin_malloc): Likewise.
25157 (handle_pointer_plus): Likewise.
25158 (handle_builtin_memcpy): Likewise. Track nonzero characters
25159 that aren't necessarily followed by a nul terminator.
25160 (handle_char_store): Likewise.
25162 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25164 PR tree-optimization/80769
25165 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25166 for malloc and calloc. Document the new invariant that all related
25167 strinfos have delayed lengths or none do.
25168 (verify_related_strinfos): Move earlier in file.
25169 (set_endptr_and_length): New function, split out from...
25170 (get_string_length): ...here. Also set the lengths of related
25172 (zero_length_string): Assert that chainsi has known (rather than
25174 (adjust_related_strinfos): Likewise.
25176 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25178 PR tree-optimization/81136
25179 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25180 assert that two references with the same misalignment have the same
25181 compile-time misalignment if those compile-time misalignments
25184 2017-07-01 Andi Kleen <ak@linux.intel.com>
25186 * print-tree.c (print_node): Print all attributes.
25188 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25190 * cfg.c (scale_bbs_frequencies): New function.
25191 * cfg.h (scale_bbs_frequencies): Declare it.
25192 * cfgloopanal.c (single_likely_exit): Cleanup.
25193 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25195 (scale_loop_profile): Likewise.
25196 (loop_version): Likewise.
25197 (create_empty_loop_on_edge): Update.
25198 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25199 scale_loop_frequencies, scale_loop_profile, loopify,
25200 loop_version): Update prototypes.
25201 * modulo-sched.c (sms_schedule): Update.
25202 * predict.c (unlikely_executed_edge_p): Also check probability.
25203 (probably_never_executed_edge_p): Fix typo.
25204 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25205 * tree-parloops.c (gen_parallel_loop): Update.
25206 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25207 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25208 * tree-ssa-loop-split.c (split_loop): Update.
25209 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25210 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25211 (vect_loop_versioning): Update.
25212 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25214 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25216 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25218 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25220 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25221 to keep profile consistent.
25223 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25225 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25226 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25227 * profile-count.h (max_safe_multiplier): Make unsigned.
25228 (profile_count::guessed_zero): New.
25230 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25232 * bb-reorder.c (fix_up_crossing_landing_pad,
25233 fix_crossing_conditional_branches): Use make_single_succ_edge
25234 to keep profile consistent.
25236 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25238 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25241 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25244 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25245 the right scopes, make sure cond_jump isn't preserved between multiple
25246 iterations. Search for fallthru edge whenever there are 3+ edges and
25247 use find_fallthru_edge for it.
25249 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25251 Patch by Alexander Monakov <amonakov@ispras.ru>
25252 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25253 probabilities consistently.
25255 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25257 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25258 emit_cmp_and_jump_insns.
25260 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25263 * tree-inline.c (expand_call_inline): Combine profile statuses.
25265 2017-06-30 Andrew Pinski <apinski@cavium.com>
25267 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25268 fold_stmt returned true.
25270 2017-06-30 Nathan Sidwell <nathan@acm.org>
25272 * ggc.h (empty_string): Delete.
25273 * cfgexpand.c (expand_asm_stmt): Use plain "".
25274 * optabs.c (expand_asm_memory_barrier): Likewise.
25275 * stringpool.c (empty_string): Delete.
25276 (digit_vector, digit_string): Delete.
25277 (ggc_alloc_string): Use plain "", don't optimize single digit
25278 strings. Use ggc_alloc_atomic.
25280 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25282 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25283 comparison set and one other set, use the cost of the non-comparison
25286 2017-06-30 Nathan Sidwell <nathan@acm.org>
25288 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25291 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25293 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25294 loops. Remove now unneeded calls to gimple_switch_set_label() that
25295 just set removed labels to NULL_TREE.
25297 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25299 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25300 (set_range_info): ...here. Only call set_range_info_raw if domain
25302 (set_nonzero_bits): Call set_range_info_raw.
25303 * tree-ssanames.h (set_range_info_raw): New.
25305 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25308 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25309 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25310 of nonimmediate_operand and <store_mask_constraint> instead of m
25311 for the input operand. For V8FI iterator, always split if input
25312 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25313 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25314 <store_mask_predicate> instead of register_operand and
25315 <store_mask_constraint> instead of v for the input operand. Make
25316 sure both operands aren't MEMs for if not <mask_applied>.
25318 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25320 * lto-wrapper.c (copy_file) Close both file descriptors before
25323 2017-06-30 Martin Liska <mliska@suse.cz>
25326 * multiple_target.c (create_dispatcher_calls): Make ifunc
25327 also for function that don't have calls or are not referenced.
25329 2017-06-30 Richard Biener <rguenther@suse.de>
25331 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25332 analyze the first scalar stmt. Move vector type computation
25333 for the BB case here from ...
25334 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25335 live operation processing in the SLP case properly.
25337 2017-06-30 Richard Biener <rguenther@suse.de>
25339 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25341 2017-06-30 Martin Liska <mliska@suse.cz>
25344 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25345 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25347 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25349 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25352 2017-06-30 Richard Biener <rguenther@suse.de>
25354 PR tree-optimization/81249
25355 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25356 condition reduction result to original scalar type.
25358 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25360 * profile-count.h (enum profile_quality): Fix typos and whitespace
25363 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25365 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25366 type for branch probabilities.
25368 2017-06-29 Julian Brown <julian@codesourcery.com>
25369 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25371 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25372 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25373 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25374 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25376 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25378 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25379 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25380 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25381 CC usage from generic code to here.
25382 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25383 CC usage into the target macros.
25385 2017-06-29 Maya Rashish <coypu@sdf.org>
25387 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25390 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25392 * arm/arm-builtins.c: Include profile-count.h
25393 * except.c (sjlj_emit_function_enter): Use
25394 profile_probability::unlikely.
25396 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25398 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25399 and tocrel_offset be pointer args rather than implicitly using
25401 (legitimate_constant_pool_address_p, rs6000_emit_move,
25402 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25403 tocrel_offset and use in toc_relative_expr_p call.
25404 (print_operand, print_operand_address): Use static tocrel_base_oac
25405 and tocrel_offset_oac.
25406 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25409 2017-06-29 Maya Rashish <coypu@sdf.org>
25411 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25413 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25415 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25416 objects, take into account only the alignment of 'op0' and 'mode1' if
25419 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25421 * ccmp.c (ccmp_tree_comparison_p): New function.
25422 (ccmp_candidate_p): Update to use above function.
25423 (get_compare_parts): New function.
25424 (expand_ccmp_next): Update to use new functions.
25425 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25427 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25428 take mode as argument.
25429 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25430 * expr.c (expand_expr_real_1): Pass mode as argument.
25432 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25434 * combine.c (combine_instructions): Print insns to dump_file, together
25437 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25439 * asan.c (asan_emit_stack_protection): Update.
25440 (create_cond_insert_point): Update.
25441 * auto-profile.c (afdo_propagate_circuit): Update.
25442 * basic-block.h (struct edge_def): Turn probability to
25443 profile_probability.
25444 (EDGE_FREQUENCY): Update.
25445 * bb-reorder.c (find_traces_1_round): Update.
25446 (better_edge_p): Update.
25447 (sanitize_hot_paths): Update.
25448 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25449 (make_single_succ_edge): Update.
25450 (check_bb_profile): Update.
25451 (dump_edge_info): Update.
25452 (update_bb_profile_for_threading): Update.
25453 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25454 probabilitycount to 0.
25455 * cfgbuild.c (compute_outgoing_frequencies): Update.
25456 * cfgcleanup.c (try_forward_edges): Update.
25457 (outgoing_edges_match): Update.
25458 (try_crossjump_to_edge): Update.
25459 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25460 (expand_gimple_tailcall): Update.
25461 (construct_init_block): Use make_single_succ_edge.
25462 (construct_exit_block): Use make_single_succ_edge.
25463 * cfghooks.c (verify_flow_info): Update.
25464 (redirect_edge_succ_nodup): Update.
25465 (split_edge): Update.
25466 (account_profile_record): Update.
25467 * cfgloopanal.c (single_likely_exit): Update.
25468 * cfgloopmanip.c (scale_loop_profile): Update.
25469 (set_zero_probability): Remove.
25470 (duplicate_loop_to_header_edge): Update.
25471 * cfgloopmanip.h (loop_version): Update prototype.
25472 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25473 (force_nonfallthru_and_redirect): Update.
25474 (update_br_prob_note): Update.
25475 (rtl_verify_edges): Update.
25476 (purge_dead_edges): Update.
25477 (rtl_lv_add_condition_to_bb): Update.
25478 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25479 * cgraphunit.c (init_lowered_empty_function): Update.
25480 (cgraph_node::expand_thunk): Update.
25481 * cilk-common.c: Include profile-count.h
25482 * dojump.c (inv): Remove.
25483 (jumpifnot): Update.
25484 (jumpifnot_1): Update.
25485 (do_jump_1): Update.
25487 (do_jump_by_parts_greater_rtx): Update.
25488 (do_compare_rtx_and_jump): Update.
25489 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25490 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25491 * dwarf2cfi.c: Include profile-count.h
25492 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25493 (sjlj_emit_dispatch_table): Likewise.
25494 * explow.c: Include profile-count.h
25495 * expmed.c (emit_store_flag_force): Update.
25496 (do_cmp_and_jump): Update.
25497 * expr.c (compare_by_pieces_d::generate): Update.
25498 (compare_by_pieces_d::finish_mode): Update.
25499 (emit_block_move_via_loop): Update.
25500 (store_expr_with_bounds): Update.
25501 (store_constructor): Update.
25502 (expand_expr_real_2): Update.
25503 (expand_expr_real_1): Update.
25504 * expr.h (try_casesi, try_tablejump): Update prototypes.
25505 * gimple-pretty-print.c (dump_probability): Update.
25506 (dump_profile): New.
25507 (dump_gimple_label): Update.
25508 (dump_gimple_bb_header): Update.
25509 * graph.c (draw_cfg_node_succ_edges): Update.
25510 * hsa-gen.c (convert_switch_statements): Update.
25511 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25512 (find_if_case_1): Update.
25513 (find_if_case_2): Update.
25514 * internal-fn.c (expand_arith_overflow_result_store): Update.
25515 (expand_addsub_overflow): Update.
25516 (expand_neg_overflow): Update.
25517 (expand_mul_overflow): Update.
25518 (expand_vector_ubsan_overflow): Update.
25519 * ipa-cp.c (good_cloning_opportunity_p): Update.
25520 * ipa-split.c (split_function): Use make_single_succ_edge.
25521 * ipa-utils.c (ipa_merge_profiles): Update.
25522 * loop-doloop.c (add_test): Update.
25523 (doloop_modify): Update.
25524 * loop-unroll.c (compare_and_jump_seq): Update.
25525 (unroll_loop_runtime_iterations): Update.
25526 * lra-constraints.c (lra_inheritance): Update.
25527 * lto-streamer-in.c (input_cfg): Update.
25528 * lto-streamer-out.c (output_cfg): Update.
25529 * mcf.c (adjust_cfg_counts): Update.
25530 * modulo-sched.c (sms_schedule): Update.
25531 * omp-expand.c (expand_omp_for_init_counts): Update.
25532 (extract_omp_for_update_vars): Update.
25533 (expand_omp_ordered_sink): Update.
25534 (expand_omp_for_ordered_loops): Update.
25535 (expand_omp_for_generic): Update.
25536 (expand_omp_for_static_nochunk): Update.
25537 (expand_omp_for_static_chunk): Update.
25538 (expand_cilk_for): Update.
25539 (expand_omp_simd): Update.
25540 (expand_omp_taskloop_for_outer): Update.
25541 (expand_omp_taskloop_for_inner): Update.
25542 * omp-simd-clone.c (simd_clone_adjust): Update.
25543 * optabs.c (expand_doubleword_shift): Update.
25544 (expand_abs): Update.
25545 (emit_cmp_and_jump_insn_1): Update.
25546 (expand_compare_and_swap_loop): Update.
25547 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25548 * predict.c (predictable_edge_p): Update.
25549 (edge_probability_reliable_p): Update.
25550 (set_even_probabilities): Update.
25551 (combine_predictions_for_insn): Update.
25552 (combine_predictions_for_bb): Update.
25553 (propagate_freq): Update.
25554 (estimate_bb_frequencies): Update.
25555 (force_edge_cold): Update.
25556 * profile-count.c (profile_count::dump): Add missing space into dump.
25557 (profile_count::debug): Add newline.
25558 (profile_count::differs_from_p): Explicitly convert to unsigned.
25559 (profile_count::stream_in): Update.
25560 (profile_probability::dump): New member function.
25561 (profile_probability::debug): New member function.
25562 (profile_probability::differs_from_p): New member function.
25563 (profile_probability::differs_lot_from_p): New member function.
25564 (profile_probability::stream_in): New member function.
25565 (profile_probability::stream_out): New member function.
25566 * profile-count.h (profile_count_quality): Rename to ...
25567 (profile_quality): ... this one.
25568 (profile_probability): New.
25569 (profile_count): Update.
25570 * profile.c (compute_branch_probabilities): Update.
25571 * recog.c (peep2_attempt): Update.
25572 * sched-ebb.c (schedule_ebbs): Update.
25573 * sched-rgn.c (find_single_block_region): Update.
25574 (compute_dom_prob_ps): Update.
25575 (schedule_region): Update.
25576 * sel-sched-ir.c (compute_succs_info): Update.
25577 * stmt.c (struct case_node): Update.
25578 (do_jump_if_equal): Update.
25579 (get_outgoing_edge_probs): Update.
25580 (conditional_probability): Update.
25581 (emit_case_dispatch_table): Update.
25582 (expand_case): Update.
25583 (expand_sjlj_dispatch_table): Update.
25584 (emit_case_nodes): Update.
25585 * targhooks.c: Update.
25586 * tracer.c (better_p): Update.
25587 (find_best_successor): Update.
25588 * trans-mem.c (expand_transaction): Update.
25589 * tree-call-cdce.c: Update.
25590 * tree-cfg.c (gimple_split_edge): Upate.
25591 (move_sese_region_to_fn): Upate.
25592 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25593 * tree-eh.c (lower_resx): Upate.
25594 (cleanup_empty_eh_move_lp): Upate.
25595 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25596 * tree-inline.c (copy_edges_for_bb): Update.
25597 (copy_cfg_body): Update.
25598 * tree-parloops.c (gen_parallel_loop): Update.
25599 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25600 (gimple_gen_time_profiler): Update.
25601 * tree-ssa-dce.c (remove_dead_stmt): Update.
25602 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25603 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25604 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25605 (unloop_loops): Update.
25606 (try_peel_loop): Update.
25607 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25608 * tree-ssa-loop-split.c (connect_loops): Update.
25609 (split_loop): Update.
25610 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25611 (hoist_guard): Update.
25612 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25613 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25614 (value_replacement): Update.
25615 * tree-ssa-reassoc.c (branch_fixup): Update.
25616 * tree-ssa-tail-merge.c (replace_block_by): Update.
25617 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25618 (create_edge_and_update_destination_phis): Update.
25619 (compute_path_counts): Update.
25620 (recompute_probabilities): Update.
25621 (update_joiner_offpath_counts): Update.
25622 (freqs_to_counts_path): Update.
25623 (duplicate_thread_path): Update.
25624 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25625 (struct switch_conv_info): Update.
25626 (gen_inbound_check): Update.
25627 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25628 (vect_do_peeling): Update.
25629 (vect_loop_versioning): Update.
25630 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25631 (optimize_mask_stores): Update.
25632 * ubsan.c (ubsan_expand_null_ifn): Update.
25633 * value-prof.c (gimple_divmod_fixed_value): Update.
25634 (gimple_divmod_fixed_value_transform): Update.
25635 (gimple_mod_pow2): Update.
25636 (gimple_mod_pow2_value_transform): Update.
25637 (gimple_mod_subtract): Update.
25638 (gimple_mod_subtract_transform): Update.
25639 (gimple_ic): Update.
25640 (gimple_stringop_fixed_value): Update.
25641 (gimple_stringops_transform): Update.
25642 * value-prof.h: Update.
25644 2017-06-29 Carl Love <cel@us.ibm.com>
25646 * config/rs6000/rs6000-c.c: Add support for built-in functions
25647 vector signed int vec_signed (vector float);
25648 vector signed long long vec_signed (vector double);
25649 vector signed int vec_signed2 (vector double, vector double);
25650 vector signed int vec_signede (vector double);
25651 vector signed int vec_signedo (vector double);
25652 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25653 instruction generator.
25654 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25655 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25656 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25658 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25659 vunsignede_v2df): Add define_expands.
25660 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25661 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25662 VEC_UNSIGNEDO): Add definitions.
25663 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25664 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25665 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25666 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25667 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25668 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25669 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25670 vec_unsignede, vec_unsignedo): Add builtin defines.
25671 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25673 * doc/extend.texi: Update the built-in documentation file for the
25674 new built-in functions.
25676 2017-06-29 Richard Biener <rguenther@suse.de>
25678 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25679 reduction chains to LOOP_VINFO_REDUCTIONS.
25680 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25681 SLP reductions after processing reduction chains.
25683 2017-06-29 Nathan Sidwell <nathan@acm.org>
25685 * builtins.c (fold_builtin_FUNCTION): Use
25686 lang_hooks.decl_printable_name.
25688 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25690 PR middle-end/81194
25691 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25692 with only one label.
25693 * stmt.c (expand_case): Assert NCASES is greater than one.
25695 2017-06-29 Richard Biener <rguenther@suse.de>
25697 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25699 (group_case_labels): Likewise.
25700 (find_taken_edge): Push sanity checking on val to workers...
25701 (find_taken_edge_cond_expr): ... here
25702 (find_taken_edge_switch_expr): ... and here, handle cases
25703 with just a default label.
25704 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25705 (group_case_labels): Likewise.
25706 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25707 group_case_labels does anything cleanup the CFG again.
25709 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25711 PR tree-optimization/81196
25712 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25713 exit condition comparing two IVs.
25715 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25717 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25718 profile to the dummy entry at the end of the list of architectures.
25719 * config/arm/arm-cpu-cdata.h: Regenerated.
25721 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25722 Michael Collison <michael.collison@arm.com>
25725 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25727 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25728 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25729 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25730 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25731 of shift when the shift amount is masked with constant equal to
25732 the size of the mode.
25733 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25736 2017-06-29 Martin Liska <mliska@suse.cz>
25738 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25740 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25743 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25746 (edge_growth_cache_entry::edge_growth_cache_entry): New
25748 (reset_edge_growth_cache): Update to use constructor.
25750 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25752 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25753 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25754 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25756 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25758 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25759 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25761 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25763 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25764 (*-linux-uclibc*): Add t-uclibc tmake_file.
25765 * config/t-musl: New.
25766 * config/t-uclibc: New.
25768 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25770 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25772 (gen_comm_data): Emit architectural setting of arch_prof.
25773 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25775 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25776 (armv8-m.base, armv8-m.main): Likewise.
25777 * arm-protos.h (arm_build_target): Add profile field.
25778 (arch_option): Likewise.
25779 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25781 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25782 arm_active_target.profile.
25784 2017-06-28 Richard Biener <rguenther@suse.de>
25786 PR middle-end/81227
25787 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25788 TYPE_OVERFLOW_WRAPS.
25789 * match.pd (negate_expr_p): Likewise.
25790 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25791 fold_build2, not fold_binary.
25793 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25795 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25796 Convert memory address to Pmode.
25797 (aarch64_print_operand): Assert MEM operands are always Pmode.
25799 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25802 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25803 Remove redundant if.
25804 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25805 * config/arm/aarch-common-protos.h
25806 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25807 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25809 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25812 * multiple_target.c (create_dispatcher_calls): Set the default
25813 clone to be static, not public.
25815 2017-06-28 Richard Biener <rguenther@suse.de>
25817 * tree-vect-loop.c (vectorizable_reduction): Move special
25818 cond reduction IV var creation ...
25819 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25820 parameter. Use STMT_VINFO_VECTYPE.
25821 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25824 2017-06-28 Martin Liska <mliska@suse.cz>
25827 * ipa-visibility.c (non_local_p): Handle visibility.
25829 2017-06-28 Martin Liska <mliska@suse.cz>
25832 * common.opt: Add IntegerRange to various options.
25833 * opt-functions.awk (integer_range_info): New function.
25834 * optc-gen.awk: Add integer_range_info to cl_options struct.
25835 * opts-common.c (decode_cmdline_option): Handle
25836 CL_ERR_INT_RANGE_ARG.
25837 (cmdline_handle_error): Likewise.
25838 * opts.c (print_filtered_help): Show valid interval in
25839 when --help is provided.
25840 * opts.h (struct cl_option): Add range_min and range_max fields.
25841 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25843 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25845 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25846 (x * C EQ/NE y * C): New transformation.
25848 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25850 * genmultilib (combination_space): Accept '+' in option names.
25852 2017-06-28 Martin Liska <mliska@suse.cz>
25855 * asan.c (instrument_derefs): Bail out inner references
25856 that are hard register variables.
25858 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25861 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25862 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25864 2017-06-28 Richard Biener <rguenther@suse.de>
25866 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25867 (vect_get_slp_defs): Adjust.
25868 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25870 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25872 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25873 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25874 (vectorizable_reduction): Adjust.
25875 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25877 (vect_get_slp_defs): Likewise.
25878 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25879 (vectorizable_bswap): Adjust.
25880 (vectorizable_call): Likewise.
25881 (vectorizable_conversion): Likewise.
25882 (vectorizable_assignment): Likewise.
25883 (vectorizable_shift): Likewise.
25884 (vectorizable_operation): Likewise.
25885 (vectorizable_store): Likewise.
25886 (vectorizable_condition): Likewise.
25887 (vectorizable_comparison): Likewise.
25889 2017-06-28 Michael Collison <michael.collison@arm.com>
25892 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25894 (arm_gen_movmemqi): Removed unused variable 'i'.
25895 Convert 'for' loop into 'while' loop.
25896 (arm_expand_prologue): Remove last unnecessary set of insn.
25897 (thumb_pop): Remove unused variable 'pushed_words'.
25898 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25900 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25902 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25903 * config/s390/s390-protos.h: Add prototype of
25904 s390_rel_address_ok_p.
25905 * config/s390/s390.c (s390_got_symbol): New function.
25906 (s390_rel_address_ok_p): New function.
25907 (legitimize_pic_address): Use s390_rel_address_ok_p.
25908 (s390_load_got): Use s390_got_symbol.
25909 (s390_option_override): Issue error if
25910 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25911 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25913 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25915 2017-06-27 Andrew Pinski <apinski@cavium.com>
25917 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25918 (X * copysign (1.0, X)): New pattern.
25919 (X * copysign (1.0, -X)): New pattern.
25920 (copysign (-1.0, CST)): New pattern.
25922 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25924 * genmultilib (combination_space): Remove variable.
25925 Validate reuse rules against regular expression for any sequence
25926 of multilib options in any order.
25928 2017-06-27 Michael Collison <michael.collison@arm.com>
25930 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25931 call aarch64_split_simd_combine.
25932 * (aarch64_combine_internal<mode>): Delete pattern.
25933 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25934 Allow register and subreg operands.
25936 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25938 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25939 specific need, just fallback on defaults.
25940 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25942 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25943 Olivier Hainque <hainque@adacore.com>
25945 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25947 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25948 targets. Pick a default if no particular attempt applied.
25949 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25952 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25954 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25955 (x86_64-wrs-vxworks7): Likewise.
25957 2017-06-27 Marek Polacek <polacek@redhat.com>
25960 * ubsan.c (instrument_null): Check get_base_address's result for null.
25962 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25964 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25966 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25968 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25969 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25970 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25971 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25972 New function types.
25973 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25974 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25975 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25976 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25977 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25978 BUILT_IN_FEUPDATEENV): New builtins.
25979 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25980 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25981 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25982 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25984 (builtin_structptr_types): Adjust size.
25985 * tree.c (builtin_structptr_types): Add four entries.
25987 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25988 Olivier Hainque <hainque@adacore.com>
25990 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25991 (TLS_SYM): New local macro, forcing reference to __tls__ on
25992 link command lines for VxWorks 7 RTPs, triggering initialization
25994 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25995 OS features TLS support, true for RTPs on VxWorks 7.
25996 * config/vxworks.c (vxworks_override_options): Setup emutls
25999 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26001 * predict.c (test_prediction_value_range): Use -1U instead of -1
26002 to avoid narrowing conversion warning.
26003 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26004 to avoid narrowing conversion warning.
26005 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26007 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26009 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26011 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26012 64bit configurations.
26013 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26014 (SIZE_TYPE): Likewise.
26015 * config/vxworks.c (vxworks_emutls_var_fields): Use
26016 long_unsigned_type_node instead of unsigned_type_node as the offset
26017 field type, which is "pointer" mode in emutls.c.
26019 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26022 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26024 PR middle-end/81207
26025 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26026 gimple_vuse copying separately from gimple_vdef copying.
26028 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26030 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26033 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26034 Olivier Hainque <hainque@adacore.com>
26036 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26037 for all vxworks7 targets.
26038 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26039 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26040 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26041 variations for VX6/VX7 and 32/64bits later on in ...
26042 (VXWORKS_LIB_SPEC): Leverage new macros.
26043 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26044 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26046 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26048 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26049 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26051 2017-06-26 Carl Love <cel@us.ibm.com>
26053 * config/rs6000/rs6000-c.c: Add support for built-in functions
26054 vector bool char vec_reve (vector bool char);
26055 vector signed char vec_reve (vector signed char);
26056 vector unsigned char vec_reve (vector unsigned char);
26057 vector bool int vec_reve (vector bool int);
26058 vector signed int vec_reve (vector signed int);
26059 vector unsigned int vec_reve (vector unsigned int);
26060 vector bool long long vec_reve (vector bool long long);
26061 vector signed long long vec_reve (vector signed long long);
26062 vector unsigned long long vec_reve (vector unsigned long long);
26063 vector bool short vec_reve (vector bool short);
26064 vector signed short vec_reve (vector signed short);
26065 vector double vec_reve (vector double);
26066 vector float vec_reve (vector float);
26067 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26068 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26069 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26070 (altivec_vreve): New pattern.
26071 * config/rs6000/altivec.h (vec_reve): New define.
26072 * doc/extend.texi (vec_rev): Update the built-in documentation file
26073 for the new built-in functions.
26075 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26077 PR tree-optimization/71815
26078 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26080 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26082 (slsr_process_phi): Likewise.
26083 (replace_uncond_cands_and_profitable_phis): Don't replace a
26084 multiply candidate with a stride of 1 (copy or cast).
26085 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26087 (lowest_cost_path): Likewise.
26088 (total_savings): Likewise.
26090 2017-06-26 Richard Biener <rguenther@suse.de>
26093 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26094 Use def_builtin_pure for all gather builtins.
26096 2017-06-26 Richard Biener <rguenther@suse.de>
26098 PR tree-optimization/81203
26099 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26100 non-dominating BBs.
26102 2017-06-26 Marek Polacek <polacek@redhat.com>
26105 * doc/invoke.texi: Document -Wmultistatement-macros.
26107 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26109 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26110 arm_neon_ok_no_float_abi effective target.
26112 2017-06-26 Richard Biener <rguenther@suse.de>
26114 PR tree-optimization/80928
26115 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26116 (copy_bbs): Set BB_DUPLICATED flag early.
26117 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26119 (execute_on_shrinking_pred): Likewise.
26120 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26121 BB_DUPLICATED blocks.
26122 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26123 iterate over all PHIs considering removal of *gsi.
26125 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26127 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26130 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26132 * config/rs6000/rs6000-string.c: (expand_block_clear,
26133 do_load_for_compare, select_block_compare_mode,
26134 compute_current_alignment, expand_block_compare,
26135 expand_strncmp_align_check, expand_strn_compare,
26136 expand_block_move, rs6000_output_load_multiple)
26137 Move functions related to string/block move/compare
26138 to a separate file.
26139 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26140 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26141 for this function which is now used in two files.
26142 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26143 * config.gcc: Add rs6000-string.o to extra_objs for
26144 targets powerpc*-*-* and rs6000*-*-*.
26146 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26149 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26150 32-bit, since indexed is not valid for DImode.
26151 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26152 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26153 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26154 (define_peephole2 for Altivec d-form store): Likewise.
26157 * multiple_target.c (create_dispatcher_calls): Only create the
26158 dispatcher call if the function is the default clone of a
26159 versioned function.
26161 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26163 PR middle-end/80902
26164 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26165 a call, force the call to not be a tail call.
26167 2017-06-23 Jeff Law <law@redhat.com>
26169 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26172 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26174 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26175 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26176 handling for early expansion of vector shifts (sl,sr,sra,rl).
26177 (builtin_function_type): Add vector shift right instructions
26178 to the unsigned argument list.
26180 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26182 rtl-optimizatoin/79286
26183 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26184 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26185 trap. PIC register plus a const unspec without offset can never trap.
26187 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26189 * tree.h (builtin_structptr_type): New type.
26190 (builtin_structptr_types): Declare new array.
26191 * tree.c (builtin_structptr_types): New array.
26192 (free_lang_data, build_common_tree_nodes): Use it.
26194 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26197 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26200 2017-06-22 Matt Turner <mattst88@gmail.com>
26202 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26203 Lake models to skylake case. Assume skylake for unknown
26204 models with clflushopt.
26206 2017-06-22 Jeff Law <law@redhat.com>
26208 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26209 frame sizes that do not satisfy aarch64_uimm12_shift.
26211 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26213 * profile-count.h (apply_probability,
26214 apply_scale, probability_in): Fix checks for zero.
26216 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26218 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26219 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26221 2016-06-22 Richard Biener <rguenther@suse.de>
26223 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26224 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26225 REDUC_MAX_EXPR support.
26226 (vectorizable_reduction): Likewise.
26227 (vect_create_epilog_for_reduction): Likewise.
26229 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26231 * match.pd (A / (1 << B) -> A >> B): New.
26232 * generic-match-head.c: Include optabs-tree.h.
26233 * gimple-match-head.c: Likewise.
26234 * optabs-tree.h (target_supports_op_p): New.
26235 * optabs-tree.c (target_supports_op_p): New.
26237 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26239 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26240 $gcc_cv_ld --help output.
26241 (gcc_cv_ld_demangle): Likewise.
26242 (gcc_cv_ld_eh_frame_hdr): Likewise.
26243 (gcc_cv_ld_pie): Likewise.
26244 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26245 (gcc_cv_ld_buildid): Likewise.
26246 (gcc_cv_ld_sysroot): Likewise.
26247 (ld_bndplt_support): Likewise.
26248 (ld_pushpopstate_support): Likewise.
26249 * configure: Regenerate.
26250 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26252 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26255 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26256 operands indexes to avoid gap between operands and match_dups.
26258 2017-06-21 Andrew Pinski <apinski@cavium.com>
26260 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26261 Increment Arith_shift and Arith_shift_reg by 1.
26262 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26264 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26265 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26266 (aarch64_strip_extend): Add new argument and test for it.
26267 (aarch64_cheap_mult_shift_p): New function.
26268 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26269 add a cost if it is true.
26270 Update calls to aarch64_strip_extend.
26271 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26273 2017-06-21 Andrew Pinski <apinski@cavium.com>
26275 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26277 (thunderxt88): Likewise.
26278 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26279 (thunderx_prefetch_tune): New variable.
26280 (thunderx2t99_prefetch_tune): Update for the correct values.
26281 (thunderxt88_tunings): New variable.
26282 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26283 generic_prefetch_tune.
26284 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26286 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26288 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26289 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26290 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26291 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26292 (aarch64_atomic_cas<mode>, GPI): Likewise.
26294 2017-06-21 Martin Liska <mliska@suse.cz>
26296 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26297 statements on cold and hot labels.
26298 * predict.c (tree_estimate_probability_bb): Remove the
26299 prediction from this place.
26301 2017-06-21 Martin Liska <mliska@suse.cz>
26303 PR tree-optimization/79489
26304 * gimplify.c (maybe_add_early_return_predict_stmt): New
26306 (gimplify_return_expr): Call the function.
26307 * predict.c (tree_estimate_probability_bb): Remove handling
26309 * predict.def: Update comment about early return predictor.
26310 * gimple-predict.h (is_gimple_predict): New function.
26311 * predict.def: Change default value of early return to 66.
26312 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26314 * passes.def: Put pass_strip_predict_hints to the beginning of
26317 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26319 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26320 FUNCTION_DECL declarations.
26321 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26323 (dwaf2out_decl): Likewise.
26324 * godump.c (go_early_global_decl): Skip call to the real debug hook
26325 for FUNCTION_DECL declarations.
26326 * passes.c (rest_of_decl_compilation): Skip call to the
26327 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26328 -fdump-go-spec is passed.
26330 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26332 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26333 Reorder for compactness.
26334 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26335 (def_builtin_pure, def_builtin_pure2): New functions.
26336 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26338 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26340 * match.pd (nop_convert): New predicate.
26341 ((A +- CST1) +- CST2): Allow some NOP conversions.
26343 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26346 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26347 with ctors/dtors if GOVD_SHARED is set.
26349 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26351 * config/aarch64/aarch64.md (movti_aarch64):
26352 Emit mov rather than orr.
26353 (movtf_aarch64): Likewise.
26354 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26355 Emit mov rather than orr.
26357 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26359 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26360 Swap alternatives, make integer dup more expensive.
26362 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26364 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26365 Return true for non-tls symbols.
26367 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26369 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26370 (cortex-a75): Likewise.
26371 (cortex-a75.cortex-a55): Likewise.
26372 * config/aarch64/aarch64-tune.md: Regenerate.
26373 * doc/invoke.texi (-mtune): Document new values for -mtune.
26375 2017-06-21 Tom de Vries <tom@codesourcery.com>
26377 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26378 stack_size feature.
26379 (Effective-Target Keywords, Other attributes): Suggest using
26380 dg-add-options stack_size feature to get stack limit in stack_size
26381 effective target documentation.
26383 2017-06-21 Julian Brown <julian@codesourcery.com>
26384 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26386 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26387 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26388 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26390 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26391 attribute type list for neon_multiply.
26392 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26393 attribute type list for neon_multiply.
26394 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26395 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26396 attribute type list for neon_multiply.
26397 * config/arm/types.md (crypto_pmull): Add.
26398 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26399 attribute type list.
26401 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26403 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26406 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26408 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26409 to make sure not to dereference a NULL cost_classes_ptr pointer.
26411 2017-06-20 Carl Love <cel@us.ibm.com>
26413 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26414 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26415 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26416 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26417 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26418 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26419 VMULOSW): New enum "unspec" values.
26420 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26421 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26422 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26423 altivec_vmulosw): New patterns.
26424 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26425 VMULOSW): Add definitions.
26427 2017-06-20 Julia Koval <julia.koval@intel.com>
26429 * config/i386/i386.c: Fix rounding expand for new pattern.
26430 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26432 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26434 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26435 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26437 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26439 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26442 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26444 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26445 architecture, then by implementer ID.
26446 * config/aarch64/aarch64-tune.md: Regenerate.
26448 2017-06-20 Richard Biener <rguenther@suse.de>
26450 PR middle-end/81097
26451 * fold-const.c (split_tree): Fold to type before negating.
26453 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26455 * diagnostic-show-locus.c
26456 (selftest::test_fixit_deletion_affecting_newline): New function.
26457 (selftest::diagnostic_show_locus_c_tests): Call it.
26459 2017-06-20 Andreas Schwab <schwab@suse.de>
26462 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26465 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26467 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26468 __ARM_FEATURE_COPROC according to support.
26470 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26472 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26473 Rewritten to avoid overflow for > 32-bit pointers.
26476 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26477 by removing enum keyword.
26478 (ubsan_type_descriptor): Likewise. Formatting fix.
26481 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26482 splitter): Require TARGET_SSE2 in the condition.
26484 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26487 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26488 for doing vector set of SFmode on ISA 3.0.
26489 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26490 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26492 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26493 SFmode value into a V4SF variable that was extracted from another
26494 V4SF variable without converting the element to double precision
26495 and back to single precision vector format.
26496 (vsx_insert_extract_v4sf_p9_2): Likewise.
26498 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26500 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26501 in UWHI to avoid undefined overflow.
26504 * ubsan.h (enum ubsan_encode_value_phase): New.
26505 (ubsan_encode_value): Change second argument to
26506 enum ubsan_encode_value_phase with default value of
26507 UBSAN_ENCODE_VALUE_GENERIC.
26508 * ubsan.c (ubsan_encode_value): Change second argument to
26509 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26510 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26511 create_tmp_var_raw instead of create_tmp_var and use a
26513 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26514 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26515 ubsan_encode_value callers.
26518 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26519 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26520 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26522 2017-06-19 Richard Biener <rguenther@suse.de>
26524 PR middle-end/81118
26525 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26526 estimates if we changed anything.
26528 2017-06-19 Richard Biener <rguenther@suse.de>
26530 PR tree-optimization/80887
26531 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26532 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26533 simplified lookups, then reset mprts_hook.
26534 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26536 (try_to_simplify): Likewise.
26538 2017-06-19 Martin Liska <mliska@suse.cz>
26541 * gimplify.c (gimplify_switch_expr):
26542 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26544 2017-06-19 Martin Liska <mliska@suse.cz>
26546 * doc/install.texi: Document that PGO runs in 4 stages.
26548 2017-06-19 Martin Liska <mliska@suse.cz>
26551 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26552 to dispatcher function name.
26553 * multiple_target.c (replace_function_decl): New function.
26554 (create_dispatcher_calls): Redirect both edges and references.
26556 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26558 * profile-count.c (profile_count::dump): Dump quality.
26559 (profile_count::differs_from_p): Update for unsigned val.
26560 * profile-count.h (profile_count_quality): New enum.
26561 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26563 2017-06-19 Richard Biener <rguenther@suse.de>
26565 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26566 struct function as arg.
26567 (estimate_numbers_of_iterations): Export overload with loop arg.
26568 (free_numbers_of_iterations_estimates_loop): Use an overload of
26569 free_numbers_of_iterations_estimates instead.
26570 * tree-cfg.c (remove_bb): Adjust.
26571 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26572 * tree-parloops.c (gen_parallel_loop): Likewise.
26573 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26575 (tree_unroll_loops_completely): Likewise.
26576 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26577 Use an overload instead and export.
26578 (estimated_loop_iterations): Adjust.
26579 (max_loop_iterations): Likewise.
26580 (likely_max_loop_iterations): Likewise.
26581 (estimate_numbers_of_iterations): Take struct function as arg
26583 (loop_exits_before_overflow): Adjust.
26584 (free_numbers_of_iterations_estimates_loop): Use an overload.
26585 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26586 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26588 2017-06-19 Richard Biener <rguenther@suse.de>
26591 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26592 RANGE_EXPR conservatively.
26594 2017-06-16 Carl Love <cel@us.ibm.com>
26596 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26597 definitions for vec_float, vec_float2, vec_floato,
26598 vec_floate built-ins.
26599 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26600 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26602 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26603 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26604 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26605 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26606 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26607 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26608 vec_floato): Add builtin defines.
26609 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26610 Update the built-in documentation file for the new built-in
26613 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26615 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26616 (mthumb): Mark as the negative of -marm.
26618 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26620 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26622 (ARM Options, -mtune): Document that this accepts the same
26623 extension options as -mcpu.
26624 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26626 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26628 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26629 permitted extensions.
26631 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26633 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26634 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26635 (armv8-m.main): Add option +nodsp.
26636 * config/arm/arm-cpu-cdata.h: Regenerated.
26638 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26640 * config/arm/t-fuchsia: New file.
26641 * config.gcc (arm*-*-fuchsia*): Use it.
26643 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26645 * config/arm/t-symbian: Rewrite for new option infrastructure.
26647 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26649 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26650 (MULTILIB_REQUIRED): Likewise.
26652 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26654 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26655 (MULTILIB_RESUE): Likewise.
26656 (MULTILIB_MATCHES): Likewise.
26657 (MULTLIB_REQUIRED): Likewise.
26659 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26661 * config/arm/t-rtems: Rewrite for new option framework.
26663 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26665 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26666 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26667 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26668 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26669 * config/arm/t-multilib: ... here.
26670 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26671 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26672 armv7-a and armv8*-a when A-profile libraries have not been built.
26673 * config/arm/t-rmprofile: Rewrite.
26675 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26677 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26678 with a backslash. Remove the backslash after substituting unescaped
26680 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26682 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26684 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26685 the driver to pass through the default setting of -mfloat-abi.
26686 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26688 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26689 (all_feat_combs): New rule.
26690 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26692 * config/arm/t-aprofile: Rewrite.
26694 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26696 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26697 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26698 fpu is not specified by the user/command-line.
26699 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26700 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26701 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26702 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26703 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26704 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26706 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26708 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26709 * config/arm/t-arm-elf: Rewritten.
26711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26713 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26714 have some floating-point instructions.
26715 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26716 (TARGET_MAYBE_HARD_FLOAT): New macro.
26717 * config/arm/arm-builtins.c (arm_init_builtins): Use
26718 TARGET_MAYBE_HARD_FLOAT.
26719 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26721 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26723 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26724 (configargs.h): Include it.
26725 (arm_print_hint_for_fpu_option): New function.
26726 (arm_parse_fpu_option): New function.
26727 (candidate_extension): New class.
26728 (arm_canon_for_multilib): New function.
26729 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26730 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26731 (ARCH_CANONICAL_SPECS): New macro.
26732 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26734 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26736 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26737 are set after handling multilib fragments. Set target_cpu_default2
26740 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26742 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26744 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26745 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26747 (check_arch): Likewise.
26748 * config/arm/arm.c (arm_configure_build_target): Handle
26749 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26750 options in the default.
26752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26754 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26755 when an option is an alias of another.
26756 * config/arm/parsecpu.awk (optalias): New parser token.
26757 (gen_comm_data): Mark non-alias options as such. Emit entries
26758 for extension aliases.
26759 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26760 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26761 (armv6kz, armv6zk, armv6t2): Likewise.
26762 (armv7): Make vfpv3-d16 an alias.
26763 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26765 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26766 Sort in canonical order.
26767 (armv8-a): Sort in canonical order.
26768 (armv8.1-a, armv8.2-a): Likewise.
26769 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26771 (cortex-a9): Sort in canonical order.
26772 * config/arm/arm.c (selftests.h): Include it.
26773 (arm_test_cpu_arch_data): New function.
26774 (arm_run_self_tests): New function.
26775 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26776 (targetm): Move declaration to the end of the file.
26777 * arm-cpu-cdata.h: Regenerated.
26779 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26781 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26782 call to target_mode_check describing the type of option passed.
26783 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26784 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26785 arm_parse_cpu_option_name to match parameters against list of
26787 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26788 arm_arch_core_flags data structure.
26789 * config/arm/arm-cpu_cdata.h: Regenerated.
26791 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26793 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26795 (arm_print_hint_for_cpu_option): Likewise.
26796 (arm_print_hint_for_arch_option): Likewise.
26797 (arm_parse_cpu_option_name): Likewise.
26798 (arm_parse_arch_option_name): Likewise.
26799 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26800 of entries in the all_fpus list.
26801 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26802 (arm_parse_cpu_option_name): Declare.
26803 (arm_parse_arch_option_name): Declare.
26804 (arm_parse_option_features): Declare.
26805 (arm_intialize_isa): Declare.
26806 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26808 (gen_comm_data): ... here. Make definitions non-static.
26809 * config/arm/arm-cpu-data.h: Regenerated.
26810 * config/arm/arm-cpu-cdata.h: Regenerated.
26812 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26814 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26815 (cpu_arch_extension): New structure.
26816 (cpu_arch_option, arch_option, cpu_option): New structures.
26817 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26818 architecture types.
26819 (gen_data): Generate new format data tables.
26820 * config/arm/arm.c (cpu_tune): New structure.
26821 (cpu_option, processors): Delete.
26822 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26823 (arm_print_hint_for_cpu_option): ... this and ...
26824 (arm_print_hint_for_arch_option): ... this.
26825 (arm_parse_arch_cpu_name): Delete. Replace with ...
26826 (arm_parse_cpu_option_name): ... this and ...
26827 (arm_parse_arch_option_name): ... this.
26828 (arm_unrecognized_feature): Change type of target parameter to
26830 (arm_parse_arch_cpu_features): Delete. Replace with ...
26831 (arm_parse_option_features): ... this.
26832 (arm_configure_build_target): Rework to use new configuration data
26834 (arm_print_tune_info): Rework for new configuration data tables.
26835 * config/arm/arm-cpu-data.h: Regenerated.
26836 * config/arm/arm-cpu.h: Regenerated.
26838 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26840 * Makefile.in (OBJS): Move sbitmap.o from here ...
26841 (OBJS-libcommon): ... to here.
26843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26845 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26846 (ISA_ALL_CRYPTO): New macro.
26847 (ISA_ALL_SIMD): New macro
26848 (ISA_ALL_FP): New macro.
26849 * config/arm/arm.c (fpu_bitlist): Update initializer.
26850 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26852 (arm9e): Add fpu. Add option for nofp
26853 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26854 (arm926ej-s, arm1026ej-s): Likewise.
26855 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26856 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26857 neon-fp16, neon-vfpv4, nofp and nosimd.
26858 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26859 (cortex-a8): Add fpu. Add option for nofp.
26860 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26861 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26862 (cortex-r4f): Add fpu.
26863 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26864 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26866 (cortex-r8): Likewise.
26867 (cortex-m4): Add fpu. Add option for nofp.
26868 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26869 (cortex-a17.cortex-a7): Likewise.
26870 (cortex-a32): Add fpu. Add options for crypto and nofp.
26871 (cortex-a35, cortex-a53): Likewise.
26872 (cortex-a57): Add fpu. Add option for crypto.
26873 (cortex-a72, cortex-a73): Likewise.
26874 (exynos-m1): Likewise.
26875 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26876 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26877 (cortex-m33): Add fpu. Add option for nofp.
26878 * config/arm/arm-cpu-cdata.h: Regenerated
26879 * config/arm/arm-cpu-data.h: Regenerated.
26881 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26883 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26884 (armv5te, armv5tej): Likewise.
26885 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26886 (armv7): Add options fp and vfpv3-d16.
26887 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26888 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26890 (armv7ve): Likewise.
26891 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26892 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26893 (armv8-a): Add nocrypto option.
26894 (armv8.1-a, armv8.2-a): Likewise.
26895 (armv8-m.main): add options fp, fp.dp and nofp.
26897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26899 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26901 (armv8-a+crc): Delete.
26902 (armv8.1-a): Add options simd, crypto and nofp.
26903 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26904 (armv8.2-a+fp16): Delete.
26905 (armv8-m.main): Add option dsp.
26906 (armv8-m.main+dsp): Delete.
26907 (cortex-a8): Add fpu. Add nofp option.
26908 (cortex-a9): Add fpu. Add nofp and nosimd options.
26909 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26910 link to main cpu and architecture data structures.
26911 (gen_comm_data): Only put isa attributes from the main architecture
26913 (option): New statement for architecture and CPU entries.
26914 * arm.c (struct cpu_option): New structure.
26915 (struct processors): Add entry for options.
26916 (arm_unrecognized_feature): New function.
26917 (arm_parse_arch_cpu_name): Ignore any characters after the first
26919 (arm_parse_arch_cpu_feature): New function.
26920 (arm_configure_build_target): Separate out any CPU and architecture
26921 features and parse separately. Don't error out if -mfpu=auto is
26922 used with only an architecture string.
26923 (arm_print_asm_arch_directives): New function.
26924 (arm_file_start): Call it.
26925 * config/arm/arm-cpu-cdata.h: Regenerated.
26926 * config/arm/arm-cpu-data.h: Likewise.
26927 * config/arm/arm-tables.opt: Likewise.
26929 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26931 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26932 assembler when it is not -mfpu=auto.
26934 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26936 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26937 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26938 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26939 (ASM_CPU_SPEC): Rewrite.
26940 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26941 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26942 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26943 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26944 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26945 copied string is NUL-terminated. Also strip any characters prefixed
26947 (arm_rewrite_selected_arch): New function.
26948 (arm_rewrite_march): New function.
26950 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26952 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26953 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26954 (march, mcpu, mtune): Convert to string-based options.
26955 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26956 (arm_parse_arch_cpu_name): New function.
26957 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26958 identify selected architecture or CPU.
26959 (arm_option_save): New function.
26960 (TARGET_OPTION_SAVE): Redefine.
26961 (arm_option_restore): Restore string options.
26962 (arm_option_print): Print string options.
26964 2017-06-16 Martin Sebor <msebor@redhat.com>
26966 PR tree-optimization/80933
26967 PR tree-optimization/80934
26968 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26969 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26970 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26971 (gimple_fold_builtin): Call them.
26973 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26975 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26976 as unlikely; update profile.
26978 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26980 * predict.c (force_edge_cold): Handle declaring edges impossible
26983 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26985 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26987 (try_unroll_loop_completely): Fix reporting.
26989 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26991 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26993 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26996 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26997 if given a non-constant argument for an intrinsic which requires a
27000 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27002 * profile.c (compare_freqs): New function.
27003 (branch_prob): Sort edge list.
27004 (find_spanning_tree): Assume that the list is priority sorted.
27006 2017-06-16 Richard Biener <rguenther@suse.de>
27008 PR tree-optimization/81090
27009 * passes.def (pass_record_bounds): Remove.
27010 * tree-pass.h (make_pass_record_bounds): Likewise.
27011 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27012 make_pass_record_bounds): Likewise.
27013 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27014 not free niter estimates at the beginning but at the end.
27015 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27017 2017-06-16 Richard Biener <rguenther@suse.de>
27019 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27020 initializer to workaround ICE in host GCC 4.8.
27022 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27024 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27026 (clone_inlined_nodes): Update.
27028 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27030 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27031 prefetch settings, and enable prefetching by default at -O3.
27033 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27035 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27036 Set flag_prefetch_loop_arrays according to tuning data.
27038 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27040 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27041 New tune structure.
27042 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27043 [Unrelated to main purpose of the patch] Place the pointer field last
27044 to enable type checking errors when tune structure are wrongly merged.
27045 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27046 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27047 (thunderx2t99_prefetch_tune): New tune constants.
27048 (tune_params *_tunings): Update all tunings (no functional change).
27049 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27050 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27051 from tunings structures.
27053 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27056 * ubsan.c (instrument_null): Add T argument, use it instead
27057 of computing it based on IS_LHS.
27058 (instrument_object_size): Likewise.
27059 (pass_ubsan::execute): Adjust instrument_null and
27060 instrument_object_size callers to pass gimple_get_lhs or
27061 gimple_assign_rhs1 result to it. Use instrument_null instead of
27062 calling get_base_address and instrument_mem_ref. Handle
27063 aggregate call arguments for object-size sanitization.
27065 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27067 PR tree-optimization/81089
27068 * tree-vrp.c (is_masked_range_test): Validate operands of
27071 2017-06-15 Martin Sebor <msebor@redhat.com>
27074 * dumpfile.c (dump_register): Avoid calling memset to initialize
27075 a class with a default ctor.
27076 * gcc.c (struct compiler): Remove const qualification.
27077 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27078 * hash-table.h: Ditto.
27079 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27081 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27082 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27084 * params.h (struct param_info): Make struct members non-const.
27085 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27086 with default initialization.
27087 * vec.h (vec_copy_construct, vec_default_construct): New helper
27089 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27090 with vec_copy_construct.
27091 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27092 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27093 * doc/invoke.texi (-Wclass-memaccess): Document.
27095 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27097 * emit-rtl.h (is_leaf): Update comment about local
27098 register allocator.
27100 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27103 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27104 for a variable to have a section before checking if the section has a
27106 Set section to.persistent if persistent attribute is set.
27107 Warn if .persistent attribute is used on an automatic variable.
27109 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27111 PR rtl-optimization/80474
27112 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27114 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27116 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27119 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27121 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27122 (casesi): Emit bounds checking as RTL.
27123 (casesi_internal_mips16_<mode>): Remove bounds checking.
27125 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27127 * config/xtensa/xtensa.c (xtensa_option_override): Append
27128 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27129 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27130 xtensa_doloop_hooks): Define unconditionally.
27131 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27133 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27134 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27135 for it in xtensa_option_override.
27136 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27137 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27139 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27141 * doc/cppopts.texi: Document '-' special value to -MF.
27143 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27145 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27146 (cortex_a53_fconst): Likewise.
27147 (cortex_a53_fpmul): Likewise.
27148 (cortex_a53_f_load_64): Likewise.
27149 (cortex_a53_f_load_many): Likewise.
27150 (cortex_a53_advsimd_alu): Likewise.
27151 (cortex_a53_advsimd_alu_q): Likewise.
27152 (cortex_a53_advsimd_mul): Likewise.
27153 (cortex_a53_advsimd_mul_q): Likewise.
27154 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27155 Add missing fmul, r2f_cvt and fconst cases.
27157 2017-06-14 Richard Biener <rguenther@suse.de>
27159 PR middle-end/81088
27160 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27162 (fold_binary_loc): When associating do not treat pre-existing
27163 TREE_OVERFLOW on literal constants as a reason to allow
27164 TREE_OVERFLOW on associated literal constants.
27166 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27168 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27169 (MASK_FEATURES): New macro.
27170 * config/sparc/sparc.c (sparc_option_override): Remove the special
27171 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27173 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27175 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27176 a division of 0 if non-call exceptions are enabled.
27178 2017-06-14 Andrew Pinski <apinski@cavium.com>
27179 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27182 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27183 Improve vector initialization code gen for only variable case.
27185 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27187 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27189 2017-06-14 Richard Biener <rguenther@suse.de>
27191 PR tree-optimization/81083
27192 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27195 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27197 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27198 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27199 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27200 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27201 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27202 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27204 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27206 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27207 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27209 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27211 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27213 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27215 * config/rs6000/t-rtems: Don't handle SPE.
27217 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27219 * config/rs6000/t-linux: Don't handle SPE.
27221 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27223 * config/rs6000/eabispe.h: Delete file.
27225 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27227 * config/rs6000/t-spe: Delete file.
27229 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27231 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27232 (rs6000_legitimate_offset_address_p): Return false for anything in
27233 V2SImode or V2SFmode.
27235 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27237 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27238 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27239 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27241 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27242 (rs6000_legitimate_offset_address_p): Ditto.
27243 (rs6000_emit_move): Ditto.
27244 (rs6000_init_builtins): Remove V4HI_type_node.
27246 2017-06-13 Martin Liska <mliska@suse.cz>
27249 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27250 (gate_asan): Likewise.
27251 * asan.h (asan_no_sanitize_address_p): Remove the function.
27252 (sanitize_flags_p): New function.
27253 * builtins.def: Fix coding style.
27254 * common.opt: Use renamed enum value.
27255 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27256 * doc/extend.texi: Document no_sanitize attribute.
27257 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27258 to SANITIZE_UNDEFINED_NONDEFAULT.
27259 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27260 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27261 Use sanitize_flags_p.
27262 * gimplify.c (gimplify_function_tree): Likewise.
27263 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27264 * opts.c (parse_no_sanitize_attribute): New function.
27265 (common_handle_option): Use renamed enum value.
27266 * opts.h (parse_no_sanitize_attribute): Declare.
27267 * tree.c (sanitize_flags_p): New function.
27268 * tree.h: Declared here.
27269 * tsan.c: Use sanitize_flags_p.
27270 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27271 (instrument_mem_ref): Likewise.
27272 (instrument_bool_enum_load): Likewise.
27273 (do_ubsan_in_current_function): Remove the function.
27274 (pass_ubsan::execute): Use sanitize_flags_p.
27275 * ubsan.h: Remove do_ubsan_in_current_function
27276 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27277 (dump_function_to_file): Use it here.
27279 2017-06-13 Martin Jambor <mjambor@suse.cz>
27281 PR tree-optimization/80803
27282 PR tree-optimization/81063
27283 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27284 (propagate_subaccesses_across_link): Enqueue subtree whenever
27285 necessary instead of relying on the caller.
27287 2017-06-13 Martin Jambor <mjambor@suse.cz>
27289 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27290 that have a first_link.
27291 (sort_and_splice_var_accesses): Do not check first_link before
27293 (subtree_mark_written_and_enqueue): Likewise.
27294 (propagate_all_subaccesses): Likewise and do not stop at first
27295 parent with a first_link.
27297 2017-06-13 Martin Jambor <mjambor@suse.cz>
27299 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27302 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27304 * match.pd: New pattern.
27306 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27308 * tree-vrp.c (is_masked_range_test): New function.
27309 (register_edge_assert_for): Determine ranges for
27312 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27314 PR tree-optimization/67328
27315 * fold-const.c (maskable_range_p): New function.
27316 (build_range_check): Generate bittests if possible.
27318 2017-06-13 Martin Liska <mliska@suse.cz>
27320 * gimple-pretty-print.c (dump_probability): Add new argument.
27321 (dump_edge_probability): Dump both probability and count.
27322 (dump_gimple_label): Likewise.
27323 (dump_gimple_bb_header): Likewise.
27325 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27328 * config/avr/avr-devices.c: Fix indentation.
27329 * config/avr/gen-avr-mmcu-specs.c: Dito.
27331 2017-06-13 Richard Biener <rguenther@suse.de>
27333 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27334 instead get vector type from stmt_info.
27335 (vectorizable_reduction): Adjust. Remove dead code.
27337 2017-06-13 Richard Biener <rguenther@suse.de>
27339 PR middle-end/81065
27340 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27341 case of C * (x * C2 + C3).
27342 (fold_addr_of_array_ref_difference): Properly fold index difference.
27344 2017-06-12 David S. Miller <davem@davemloft.net>
27347 * config/sparc/sparc.md (return expander): Emit frame blockage if
27348 function uses alloca.
27350 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27352 * combine.c (make_field_assignment): Check len rather than the mode
27353 precision when calling force_to_mode.
27355 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27357 Support multilibs and devices that see flash in RAM address range.
27360 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27361 (avr_mcu_t) <flash_pm_offset>: New field.
27362 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27363 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27364 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27365 (AVR_TINY_PM_OFFSET): Remove macro.
27366 * config/avr/avr.opt (-mshort-calls): New option.
27367 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27368 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27369 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27370 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27371 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27372 instead of avr_arch->have_jmp_call.
27373 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27374 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27375 avr_arch->flash_pm_offset to define.
27376 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27377 new field flash_pm_offset. Add entry for avrxmega3.
27378 (avr_texinfo): Add entry for avrxmega3.
27379 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27380 attiny212, attiny214,
27381 attiny412, attiny414, attiny416, attiny417,
27382 attiny814, attiny816, attiny817,
27383 attiny1614, attiny1616, attiny1617,
27384 attiny3214, attiny3216, attiny3217.
27385 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27386 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27387 (avr_print_operand_address) [AVR_TINY]: Same.
27388 (avr_asm_init_sections) <readonly_data_section>: Only patch
27389 callback if avr_arch->flash_pm_offset = 0.
27390 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27391 for rodata if avr_arch->flash_pm_offset != 0.
27392 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27393 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27394 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27395 (m_options): Append opt_rcall.
27396 (m_dirnames): Append dir_rcall.
27397 * config/avr/t-multilib: Regenerate.
27399 * configure.ac [target=avr]: Check whether avrxmega3 default
27400 linker description file works as needed.
27401 * configure: Regenerate.
27402 * doc/avr-mmcu.texi: Regenerate.
27403 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27404 <__AVR_ARCH__>: Document avrxmega3 and 103.
27405 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27406 <__AVR_SHORT_CALLS__>: Document it.
27407 <__AVR_PM_BASE_ADDRESS__>: Document it.
27408 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27409 (AVR Variable Attributes) <progmem>: Document this is
27410 not needed for avrxmega3.
27411 (AVR Named Address Spaces) <__flash>: Dito.
27413 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27415 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27417 2017-06-12 Doug Rupp <rupp@adacore.com>
27419 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27420 Append vxworks-stdint.h to the tm_file list.
27421 * config/vxworks-stdint.h: New file.
27423 2017-06-12 Martin Liska <mliska@suse.cz>
27425 PR tree-optimization/81041
27426 * tree-profile.c (gimple_gen_ic_func_profiler):
27427 Create an extra BB in profile-generate
27428 (gimple_gen_time_profiler): Likewise.
27430 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27432 PR tree-optimization/81003
27433 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27434 (update_range_test): Use it instead of force_gimple_operand_gsi.
27436 2017-06-12 Richard Biener <rguenther@suse.de>
27438 PR tree-optimization/81053
27439 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27440 with backedge value not defined in loop. Simplify def stmt
27443 2017-06-11 Tom de Vries <tom@codesourcery.com>
27446 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27448 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27449 nvptx_cannot_force_const_mem.
27451 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27453 * opts.c (finish_options): Move test for flag_split_stack after
27454 it has been initialized.
27456 2017-06-11 Jason Merrill <jason@redhat.com>
27458 * tree.h (id_equal): New.
27459 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27460 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27461 instead of strcmp of IDENTIFIER_POINTER.
27463 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27465 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27466 (mark_all_inlined_calls_cdtor): Fix formating.
27467 (inline_transform): Rescale profile before inlining.
27469 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27471 * cgraph.h (cgraph_edge::clone): Update prototype.
27472 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27473 (cgraph_node::create_clone): Update.
27474 (cgraph_node::create_version_clone): Update.
27475 * tree-inline.c (copy_bb): Update.
27476 (expand_call_inline): Update.
27478 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27480 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27481 factored out from ...
27482 (rs6000_emit_prologue): ... here.
27484 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27486 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27487 factored out from ...
27488 (rs6000_emit_prologue): ... here.
27490 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27492 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27494 (handle_missing_profiles): Fix computation of tp_first_run.
27495 (counts_to_freqs): Do not touch freqs when count is 0.
27497 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27499 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27502 2017-06-10 Tom de Vries <tom@codesourcery.com>
27504 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27505 attributes): Document signal effective target.
27507 2017-06-10 Tom de Vries <tom@codesourcery.com>
27509 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27510 Document effective target stack_size.
27512 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27514 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27515 to the edit_context if they can be auto-applied.
27517 2017-06-9 Ian Lance Taylor <iant@golang.org>
27519 * opts.c (finish_options): If -fsplit-stack, disable implicit
27520 -forder-blocks-and-partition.
27521 * doc/invoke.texi (Optimize Options): Document that when using
27522 -fsplit-stack -forder-blocks-and-partition is not implicitly
27525 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27527 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27528 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27529 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27530 * builtins.def (abort, trap, unreachable): Declare cold.
27531 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27532 * tree-core.h (ECF_COLD): New.
27533 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27534 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27536 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27538 * predict.c (unlikely_executed_stmt_p): Cleanup.
27540 2017-06-09 Richard Biener <rguenther@suse.de>
27542 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27543 model if the ref is always written to.
27545 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27547 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27549 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27551 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27554 2017-06-09 Tom de Vries <tom@codesourcery.com>
27557 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27558 "target cannot support label values" when encountering LABEL_REF.
27560 2017-06-09 Martin Liska <mliska@suse.cz>
27562 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27563 (gimple_gen_ic_func_profiler): Emit direct comparison
27564 of __gcov_indirect_call_callee with NULL.
27565 (gimple_gen_time_profiler): Change probability from
27566 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27568 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27570 * profile.c (edge_gcov_counts): Turn to pointer.
27571 (compute_branch_probabilities, compute_branch_probabilities): Update.
27572 (branch_prob): Do not clear edge_gcov_count.
27573 * profile.h (edge_gcov_counts): Turn to pointer.
27574 (edge_gcov_count): Update.
27576 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27578 * gimple.h (gimple_check_failed): Mark cold.
27580 2017-06-09 Richard Biener <rguenther@suse.de>
27582 PR tree-optimization/66623
27583 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27584 refactor check_reduction into two parts, properly computing
27585 whether we have to check reduction validity for outer loop
27588 2017-06-09 Richard Biener <rguenther@suse.de>
27590 PR tree-optimization/79483
27591 * graphite-scop-detection.c (order): New global.
27592 (get_order): Compute bb to order mapping that satisfies code
27593 generation constraints.
27594 (cmp_pbbs): New helper.
27595 (build_scops): Start domwalk at entry block, sort generated
27598 2017-06-09 Richard Biener <rguenther@suse.de>
27600 PR middle-end/81007
27601 * ipa-polymorphic-call.c
27602 (ipa_polymorphic_call_context::restrict_to_inner_class):
27603 Skip FIELD_DECLs with error_mark_node type.
27604 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27607 2017-06-09 Martin Liska <mliska@suse.cz>
27609 * predict.c (struct branch_predictor): New struct.
27610 (test_prediction_value_range): New test.
27611 (predict_c_tests): New function.
27612 * selftest-run-tests.c (selftest::run_tests): Run the function.
27613 * selftest.h: Declare new tests.
27615 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27618 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27619 gen_add3_insn did not fail.
27620 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27621 r0, construct that number in a temporary reg and add that reg to r0.
27622 If asked to put the result in r0 as well, fail.
27624 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27626 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27627 for early expansion of vec_eqv.
27629 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27631 PR middle-end/81005
27632 * ubsan.c (instrument_null): Avoid pointless code temporary.
27633 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27635 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27639 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27642 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27645 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27647 * predict.c (unlikely_executed_edge_p): Move ahead.
27648 (probably_never_executed_edge_p): Use it.
27650 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27652 PR middle-end/79988
27653 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27654 gimple_call_builtin_p call.
27656 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27658 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27659 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27660 rtl_check_failed_type2, rtl_check_failed_code1,
27661 rtl_check_failed_code2, rtl_check_failed_code_mode,
27662 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27663 rtvec_check_failed_bounds, rtl_check_failed_flag,
27664 _fatal_insn_not_found, _fatal_insn): Likewise.
27665 * tree.h (tree_contains_struct_check_failed,
27666 tree_check_failed, tree_not_check_failed,
27667 tree_class_check_failed, tree_range_check_failed,
27668 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27669 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27670 tree_operand_check_failed, omp_clause_check_failed,
27671 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27674 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27676 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27677 flag_branch_probabilities.
27678 * ipa-inline.c (edge_badness): Likewise.
27679 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27680 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27681 * predict.c (maybe_hot_frequency_p): Likewise.
27682 (probably_never_executed): Likewise.
27683 * sched-ebb.c (schedule_ebbs): Likewise.
27684 * sched-rgn.c (find_single_block_region): Likewise.
27685 * tracer.c (tail_duplicate): Likewise.
27687 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27689 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27690 longer requires x_flag_profile_use.
27692 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27694 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27695 instead of flag_reorder_blocks_and_partition.
27696 * dbxout.c (dbxout_function_end): Likewise.
27697 * dwarf2out.c (gen_subprogram_die): Likewise.
27698 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27699 * hw-doloop.c (reorg_loops): Likewise.
27700 * varasm.c (assemble_start_function,
27701 assemble_end_function): Likewise.
27702 (decide_function_section): Do not check for
27703 flag_reorder_blocks_and_partition.
27705 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27707 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27709 (chkp_get_hard_register_fake_addr_expr): Ditto.
27710 (chkp_build_addr_expr): Add check for hard reg case.
27711 (chkp_parse_array_and_component_ref): Ditto.
27712 (chkp_find_bounds_1): Ditto.
27713 (chkp_process_stmt): Don't generate bounds store for
27716 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27718 * predict.c (maybe_hot_bb_p): Do not check profile status.
27719 (maybe_hot_edge_p): Likewise.
27720 (probably_never_executed): Check for zero counts even if profile
27722 (unlikely_executed_edge_p): New function.
27723 (unlikely_executed_stmt_p): New function.
27724 (unlikely_executed_bb_p): New function.
27725 (set_even_probabilities): Use unlikely predicates.
27726 (combine_predictions_for_bb): Likewise.
27727 (predict_paths_for_bb): Likewise.
27728 (predict_paths_leading_to_edge): Likewise.
27729 (determine_unlikely_bbs): New function.
27730 (estimate_bb_frequencies): Use it.
27731 (compute_function_frequency): Use zero counts even if profile is
27733 * profile-count.h: Fix typo.
27735 2017-08-08 Julia Koval <julia.koval@intel.com>
27737 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27738 _mm512_mask_cvtsepi16_storeu_epi8,
27739 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27740 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27741 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27742 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27743 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27744 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27745 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27746 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27747 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27748 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27749 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27750 __builtin_ia32_pmovuswb256mem_mask,
27751 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27752 __builtin_ia32_pmovwb512mem_mask): New builtins.
27754 2017-08-08 Julia Koval <julia.koval@intel.com>
27756 PR target/73350,80862
27757 * config/i386/subst.md (round): Fix round pattern.
27758 * config/i386/i386.c (ix86_erase_embedded_rounding):
27759 Fix erasing rounding for the fixed pattern.
27761 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27763 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27765 2017-06-08 Martin Liska <mliska@suse.cz>
27767 PR gcov-profile/80911
27768 * gcov.c (block_info::block_info): New constructor.
27770 2017-06-07 Carl Love <cel@us.ibm.com>
27772 * config/rs6000/rs6000-c: The return type of the following
27773 built-in functions was implemented as int not long long. Fix sign
27774 of return value for the unsigned version of vec_mulo and vec_mule.
27775 vector unsigned long long vec_bperm (vector unsigned long long,
27776 vector unsigned char)
27777 vector signed long long vec_mule (vector signed int,
27779 vector unsigned long long vec_mule (vector unsigned int,
27780 vector unsigned int)
27781 vector signed long long vec_mulo (vector signed int,
27783 vector unsigned long long vec_mulo (vector unsigned int,
27784 vector unsigned int)
27785 * doc/extend.texi: Fix the documentation for the built-in
27788 2017-06-07 Carl Love <cel@us.ibm.com>
27791 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27794 2017-06-07 Carl Love <cel@us.ibm.com>
27796 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27797 support, Generate doublehv for signed int/float for BE case only.
27799 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27801 * doc/invoke.texi (mcx16): Rewrite.
27803 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27805 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27806 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27807 *mov<mode>_softfloat, and an anonymous splitter): Use
27808 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27810 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27812 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27814 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27815 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27816 (rs6000_debug_reg_global): Adjust.
27817 (rs6000_init_hard_regno_mode_ok): Adjust.
27818 (rs6000_dbx_register_number): Adjust.
27819 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27820 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27821 Remove SPE_ACC and SPEFSCR.
27822 (REG_ALLOC_ORDER): Ditto.
27823 (FRAME_POINTER_REGNUM): Change to 111.
27824 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27825 (REG_CLASS_NAMES): Ditto.
27826 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27827 (REGISTER_NAMES): Ditto.
27828 (ADDITIONAL_REG_NAMES): Ditto.
27829 (rs6000_reg_names): Ditto.
27830 * config/rs6000/rs6000.md: Renumber some register number
27833 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27835 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27837 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27838 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27840 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27841 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27842 Delete the SPE high registers.
27843 (REG_ALLOC_ORDER): Ditto.
27844 (enum reg_class): Remove SPE_HIGH_REGS.
27845 (REG_CLASS_NAMES): Ditto.
27846 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27847 (REGISTER_NAMES): Ditto.
27848 (rs6000_reg_names): Ditto.
27849 * doc/tm.texi.in: Remove SPE as example.
27850 * doc/tm.texi: Regenerate.
27852 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27854 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27855 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27856 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27857 * config/rs6000/rs6000.md (type): Remove "brinc".
27859 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27861 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27862 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27863 * config/rs6000/linuxspe.h: Delete file.
27864 * config/rs6000/rs6000.md: Don't include spe.md.
27865 * config/rs6000/spe.h: Delete file.
27866 * config/rs6000/spe.md: Delete file.
27867 * config/rs6000/t-rs6000: Remove spe.md.
27869 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27871 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27872 (reg_or_none500mem_operand): Delete.
27873 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27874 instead of reg_or_none500mem_operand.
27876 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27878 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27879 handling of SPE flags.
27880 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27882 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27884 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27886 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27887 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27888 paired_divv2sf3): Similar.
27889 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27890 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27891 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27893 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27894 Rename the paired_* instruction patterns.
27895 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27897 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27898 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27899 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27900 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27901 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27902 PAIRED_VECTOR_MODE.
27903 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27904 (spe_func_has_64bit_regs_p): Delete.
27905 (spe_expand_predicate_builtin): Delete.
27906 (spe_expand_evsel_builtin): Delete.
27907 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27908 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27909 (invalid_e500_subreg): Delete.
27910 (rs6000_legitimize_address): Always force_reg op2 as well, for
27911 paired single memory accesses.
27912 (rs6000_member_type_forces_blk): Delete.
27913 (rs6000_spe_function_arg): Delete.
27914 (rs6000_expand_unop_builtin): Delete SPE handling.
27915 (rs6000_expand_binop_builtin): Ditto.
27916 (spe_expand_stv_builtin): Delete.
27917 (bdesc_2arg_spe): Delete.
27918 (spe_expand_builtin): Delete.
27919 (spe_expand_predicate_builtin): Delete.
27920 (spe_expand_evsel_builtin): Delete.
27921 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27922 (spe_init_builtins): Delete.
27923 (spe_func_has_64bit_regs_p): Delete.
27924 (savres_routine_name): Delete "info" parameter. Adjust callers.
27925 (rs6000_emit_stack_reset): Ditto.
27926 (rs6000_dwarf_register_span): Delete.
27927 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27928 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27929 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27931 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27933 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27934 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27935 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27936 mulv2sf3, divv2sf3): Delete expanders.
27938 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27940 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27942 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27944 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27945 * config/rs6000/rs6000.c: Ditto.
27947 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27949 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27950 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27951 comparison_operator.
27953 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27955 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27956 * config/rs6000/rs6000.opt: Ditto.
27957 * config/rs6000/t-rtems: Ditto.
27959 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27961 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27962 TARGET_E500_SINGLE by 0, simplify.
27963 * config/rs6000/rs6000.c: Ditto.
27964 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27965 (spe_build_register_parallel): Delete.
27966 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27967 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27968 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27969 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27970 (E500_CONVERT): Delete.
27971 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27973 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27975 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27976 * config/rs6000/dfp.md: Ditto.
27977 (negdd2, *negdd2_fpr): Merge.
27978 (absdd2, *absdd2_fpr): Merge.
27979 (negtd2, *negtd2_fpr): Merge.
27980 (abstd2, *abstd2_fpr): Merge.
27981 * config/rs6000/e500.h: Delete file.
27982 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27983 TARGET_FPRS by 1 and simplify.
27984 * config/rs6000/rs6000-c.c: Ditto.
27985 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27986 TARGET_DF_SPE by 0.
27987 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27989 * config/rs6000/rs6000.md: Ditto.
27990 (floatdidf2, *floatdidf2_fpr): Merge.
27991 (move_from_CR_gt_bit): Delete.
27992 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27993 (E500_CR_IOR_COMPARE): Delete.
27994 (All patterns that require !TARGET_FPRS): Delete.
27995 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27997 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27999 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28001 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28003 * graphds.c (add_edge): Intitialize edge's attached data.
28004 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28005 pointer parameter. Call pointed function on each edge during
28006 graph traversing. Skip traversing the edge when the function
28008 (graphds_dfs, graphds_scc): Ditto.
28009 (for_each_edge): New parameter. Pass the new parameter to callback
28011 * graphds.h (skip_edge_callback): New function pointer type.
28012 (graphds_dfs, graphds_scc): New function pointer parameter.
28013 (graphds_edge_callback, for_each_edge): New parameter.
28015 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28017 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28018 out code checking if runtime alias check is possible to below ...
28019 Call the new function.
28020 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28021 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28023 2017-06-07 Marek Polacek <polacek@redhat.com>
28026 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28027 TYPE_OVERFLOW_WRAPS check.
28029 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28031 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28032 if versioning is required.
28033 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28034 peeling with the check for versioning.
28036 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28038 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28039 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28040 Set true to new parameter if new ssa variable is defined.
28041 (vect_gen_vector_loop_niters): Refactor. Set range information
28042 for the new vector loop bound variable.
28043 (vect_do_peeling): Ditto.
28045 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28047 * tree-affine.c (ssa.h): Include header file.
28048 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28049 has wrapping overflow behavior.
28051 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28053 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28055 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28057 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28058 (tree_to_aff_combination): ... here.
28060 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28062 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28063 reg_pressure model function.
28064 (ivopts_global_cost_for_size): Delete.
28065 (determine_set_costs, iv_ca_recount_cost): Call new model function
28066 ivopts_estimate_reg_pressure.
28068 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28070 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28071 expensive than udiv. Remove floating point cases from mod.
28073 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28075 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28076 Increase idiv cost.
28078 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28080 * config/aarch64/aarch64.md
28081 (copysignsf3): Fix mask generation.
28083 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28085 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28087 (class dump_manager): Add register_dumps method.
28088 * dumpfile.c: Include langhooks.h.
28089 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28090 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28091 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28092 (dump_manager::dump_register): Start with 512 entries instead of 32.
28093 (dump_manager::register_dumps): New method.
28094 * toplev.c (general_init): Instead of invoking register_dumps
28095 langhook, invoke register_dumps method on the dump manager.
28096 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28099 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28101 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28102 Say that # requires an associated define_split to exist, and that
28103 the define_split must be suitable for use after register allocation.
28105 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28107 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28108 (compute_outgoing_frequencies): Also initialize zero counts.
28109 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28110 around loops; preserve more of profile when nothing changes.
28112 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28114 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28116 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28117 * config/arm/arm-cpu-cdata.h: Regenerate.
28118 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28119 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28120 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28121 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28122 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28124 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28126 * config/arm/t-rmprofile: Likewise.
28127 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28129 2017-06-06 David S. Miller <davem@davemloft.net>
28132 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28133 blockage if function uses alloca.
28135 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28137 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28138 New "uid" fields to hold pretty-print IDs of group and ref.
28139 Memory references are now identified as <group_id>:<ref_id>
28140 instead of using [random] addresses.
28141 (dump_mem_details): Simplify, no functional change.
28142 (dump_mem_ref): Simplify and make output more concise.
28143 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28144 (find_or_create_group): Initialize group uid.
28145 (record_ref): Initialize ref uid. Improve debug output.
28146 (prune_group_by_reuse, should_issue_prefetch_p,)
28147 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28148 (mark_nontemporal_store, determine_loop_nest_reuse):
28149 Improve debug output.
28151 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28153 * dbgcnt.def (prefetch): New debug counter.
28154 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28155 (schedule_prefetches): Stop issueing prefetches if debug counter
28158 2017-06-06 Tom de Vries <tom@codesourcery.com>
28160 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28161 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28163 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28165 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28166 Use aarch64_reg_or_zero predicate for operand 4.
28167 (aarch64_compare_and_swap<mode> define_insn_and_split):
28168 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28169 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28171 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28173 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28174 (arm_compute_save_core_reg_mask): This.
28175 (thumb1_compute_save_reg_mask): Rename into ...
28176 (thumb1_compute_save_core_reg_mask): This.
28177 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28178 (arm_compute_frame_layout): Likewise.
28180 2017-06-06 Richard Biener <rguenther@suse.de>
28182 PR tree-optimization/80974
28183 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28184 keep or clear leaders SSA info.
28186 2017-06-06 Tom de Vries <tom@codesourcery.com>
28188 * config/nvptx/nvptx.c (split_mode_p): New function.
28189 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28191 2017-06-06 Tom de Vries <tom@codesourcery.com>
28193 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28195 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28198 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28201 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28203 * shrink-wrap.c (handle_simple_exit): Update profile.
28204 (try_shrink_wrapping): Upate profile.
28206 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28208 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28209 (tree_guess_outgoing_edge_probabilities): New.
28210 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28211 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28213 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28215 * ipa-split.c (split_function): Initialize return bb profile.
28217 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28219 * profile.c (compute_branch_probabilities): Also initialize
28220 EXIT_BLOCK profile.
28222 2017-06-06 Richard Biener <rguenther@suse.de>
28224 PR tree-optimization/80928
28225 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28226 (vect_analyze_loop_operations): Properly guard analysis for
28228 (vect_transform_loop): Likewise.
28229 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28230 (vect_model_induction_cost): Do not cost for pure SLP.
28231 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28232 of induction in inner loop vectorization.
28233 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28234 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28235 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28237 (vect_analyze_slp_cost_1): Cost induction.
28238 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28239 (vect_get_slp_vect_defs): Likewise.
28240 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28241 (vect_transform_stmt): Handle SLP reductions.
28242 * tree-vectorizer.h (vectorizable_induction): Adjust.
28244 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28246 * config/rs6000/rs6000.c (make_resolver_func): Update
28247 init_lowered_empty_function call.
28249 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28251 * doc/invoke.texi: Document the -fprofile-abs-path option.
28252 * common.opt (fprofile-abs-path): New option.
28253 * gcov-io.h (gcov_write_filename): Declare.
28254 * gcov-io.c (gcov_write_filename): New function.
28255 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28256 * profile.c (output_location): Likewise.
28258 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28260 * shring-wrap.c: Revert accidental commit.
28262 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28264 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28266 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28268 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28270 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28271 profile in callgraph edge.
28272 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28273 (apply_scale): Likewise.
28274 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28275 Also scale profile when inlining function with zero profile.
28276 (initialize_cfun): Update exit block profile even when it is zero.
28277 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28278 when profile is read.
28280 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28282 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28283 (CLONE_*): New constants to define the processors we can generate
28284 code for with the target_clone attribute.
28285 (rs6000_clone_map): New array to identify which clone processors
28286 the current program is running on.
28287 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28288 target_clone attribute.
28289 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28290 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28291 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28292 (cpu_expand_builtin): Add support for target_clone attribute.
28293 (rs6000_valid_attribute_p): Allow "default" attribute.
28294 (get_decl_name): New debug function to simplify printing the
28295 current function name in debugging statements.
28296 (rs6000_clone_priority): New functions to support the target_clone
28297 attribute, and be able to generate code to switch between ISA 2.05
28298 through ISA 3.0 (power6 through power9).
28299 (rs6000_compare_version_priority): Likewise.
28300 (rs6000_get_function_versions_dispatcher): Likewise.
28301 (make_resolver_func): Likewise.
28302 (add_condition_to_bb): Likewise.
28303 (dispatch_function_versions): Likewise.
28304 (rs6000_generate_version_dispatcher_body): Likewise.
28305 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28306 (fusion_gpr_load_p): Fix a spacing issue.
28307 * doc/extend.texi (Common Function Attributes): Document that the
28308 PowerPC supports the target_clone attribute.
28310 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28312 * config/arm/arm.h: explain F symbol found in description of ARM
28313 register allocation in its legend.
28315 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28317 * config/mips/frame-header-opt.c: Include profile-count.h.
28318 * config/riscv/riscv.c: Include profile-count.h
28320 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28322 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28324 (sm_set_flag_if_changed): Add bbs field.
28325 (execute_sm_if_changed_flag_set): Pass BBS.
28326 (execute_sm): Update.
28328 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28330 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28333 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28335 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28336 (peephole2): New peephole2 to emit the above.
28337 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28339 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28341 * config/aarch64/aarch64.c (define_peephole2 above
28342 *sub_<shift>_<mode>): New peephole.
28344 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28346 * config/i386/i386.c (make_resolver_func): Update.
28347 * Makefile.in: Add profile-count.h and profile-count.o
28348 * auto-profile.c (afdo_indirect_call): Update to new API.
28349 (afdo_set_bb_count): Update.
28350 (afdo_propagate_edge): Update.
28351 (afdo_propagate_circuit): Update.
28352 (afdo_calculate_branch_prob): Update.
28353 (afdo_annotate_cfg): Update.
28354 * basic-block.h: Include profile-count.h
28355 (struct edge_def): Turn count to profile_count.
28356 (struct basic_block_def): Likewie.
28357 (REG_BR_PROB_BASE): Move to profile-count.h
28358 (RDIV): Move to profile-count.h
28359 * bb-reorder.c (max_entry_count): Turn to profile_count.
28360 (find_traces): Update.
28361 (rotate_loop):Update.
28362 (connect_traces):Update.
28363 (sanitize_hot_paths):Update.
28364 * bt-load.c (migrate_btr_defs): Update.
28365 * cfg.c (RDIV): Remove.
28366 (init_flow): Use alloc_block.
28367 (alloc_block): Uninitialize count.
28368 (unchecked_make_edge): Uninitialize count.
28369 (check_bb_profile): Update.
28370 (dump_edge_info): Update.
28371 (dump_bb_info): Update.
28372 (update_bb_profile_for_threading): Update.
28373 (scale_bbs_frequencies_int): Update.
28374 (scale_bbs_frequencies_gcov_type): Update.
28375 (scale_bbs_frequencies_profile_count): New.
28376 * cfg.h (update_bb_profile_for_threading): Update.
28377 (scale_bbs_frequencies_profile_count): Declare.
28378 * cfgbuild.c (compute_outgoing_frequencies): Update.
28379 (find_many_sub_basic_blocks): Update.
28380 * cfgcleanup.c (try_forward_edges): Update.
28381 (try_crossjump_to_edge): Update.
28382 * cfgexpand.c (expand_gimple_tailcall): Update.
28383 (construct_exit_block): Update.
28384 * cfghooks.c (verify_flow_info): Update.
28385 (dump_bb_for_graph): Update.
28386 (split_edge): Update.
28387 (make_forwarder_block): Update.
28388 (duplicate_block): Update.
28389 (account_profile_record): Update.
28390 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28391 (get_estimated_loop_iterations): Update.
28392 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28393 (single_likely_exit): Update.
28394 * cfgloopmanip.c (scale_loop_profile): Update.
28396 (set_zero_probability): Update.
28397 (lv_adjust_loop_entry_edge): Update.
28398 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28399 (purge_dead_edges): Update.
28400 (rtl_account_profile_record): Update.
28401 * cgraph.c (cgraph_node::create): Uninitialize count.
28402 (symbol_table::create_edge): Uninitialize count.
28403 (cgraph_update_edges_for_call_stmt_node): Update.
28404 (cgraph_edge::dump_edge_flags): Update.
28405 (cgraph_node::dump): Update.
28406 (cgraph_edge::maybe_hot_p): Update.
28407 * cgraph.h: Include profile-count.h
28408 (create_clone), create_edge, create_indirect_edge): Update.
28409 (cgraph_node): Turn count to profile_count.
28410 (cgraph_edge0: Likewise.
28411 (make_speculative, clone): Update.
28412 (create_edge): Update.
28413 (init_lowered_empty_function): Update.
28414 * cgraphclones.c (cgraph_edge::clone): Update.
28415 (duplicate_thunk_for_node): Update.
28416 (cgraph_node::create_clone): Update.
28417 * cgraphunit.c (cgraph_node::analyze): Update.
28418 (cgraph_node::expand_thunk): Update.
28419 * final.c (dump_basic_block_info): Update.
28420 * gimple-streamer-in.c (input_bb): Update.
28421 * gimple-streamer-out.c (output_bb): Update.
28422 * graphite.c (print_global_statistics): Update.
28423 (print_graphite_scop_statistics): Update.
28424 * hsa-brig.c: Include basic-block.h.
28425 * hsa-dump.c: Include basic-block.h.
28426 * hsa-gen.c (T sum_slice): Update.
28427 (convert_switch_statements):Update.
28428 * hsa-regalloc.c: Include basic-block.h.
28429 * ipa-chkp.c (chkp_produce_thunks): Update.
28430 * ipa-cp.c (struct caller_statistics): Update.
28431 (init_caller_stats): Update.
28432 (gather_caller_stats): Update.
28433 (ipcp_cloning_candidate_p): Update.
28434 (good_cloning_opportunity_p): Update.
28435 (get_info_about_necessary_edges): Update.
28436 (dump_profile_updates): Update.
28437 (update_profiling_info): Update.
28438 (update_specialized_profile): Update.
28439 (perhaps_add_new_callers): Update.
28440 (decide_about_value): Update.
28441 (ipa_cp_c_finalize): Update.
28442 * ipa-devirt.c (struct odr_type_warn_count): Update.
28443 (struct decl_warn_count): Update.
28444 (struct final_warning_record): Update.
28445 (possible_polymorphic_call_targets): Update.
28446 (ipa_devirt): Update.
28447 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28448 * ipa-icf.c (sem_function::merge): Update.
28449 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28450 * ipa-inline.c (compute_uninlined_call_time): Update.
28451 (compute_inlined_call_time): Update.
28452 (want_inline_small_function_p): Update.
28453 (want_inline_self_recursive_call_p): Update.
28454 (edge_badness): Update.
28455 (lookup_recursive_calls): Update.
28456 (recursive_inlining): Update.
28457 (inline_small_functions): Update.
28458 (dump_overall_stats): Update.
28459 (dump_inline_stats): Update.
28460 * ipa-profile.c (ipa_profile_generate_summary): Update.
28461 (ipa_propagate_frequency): Update.
28462 (ipa_profile): Update.
28463 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28464 * ipa-utils.c (ipa_merge_profiles): Update.
28465 * loop-doloop.c (doloop_modify): Update.
28466 * loop-unroll.c (report_unroll): Update.
28467 (unroll_loop_runtime_iterations): Update.
28468 * lto-cgraph.c (lto_output_edge): Update.
28469 (lto_output_node): Update.
28470 (input_node): Update.
28471 (input_edge): Update.
28472 (merge_profile_summaries): Update.
28473 * lto-streamer-in.c (input_cfg): Update.
28474 * lto-streamer-out.c (output_cfg): Update.
28475 * mcf.c (create_fixup_graph): Update.
28476 (adjust_cfg_counts): Update.
28477 (sum_edge_counts): Update.
28478 * modulo-sched.c (sms_schedule): Update.
28479 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28480 * predict.c (maybe_hot_count_p): Update.
28481 (probably_never_executed): Update.
28482 (dump_prediction): Update.
28483 (combine_predictions_for_bb): Update.
28484 (propagate_freq): Update.
28485 (handle_missing_profiles): Update.
28486 (counts_to_freqs): Update.
28487 (rebuild_frequencies): Update.
28488 (force_edge_cold): Update.
28489 * predict.h: Include profile-count.h
28490 (maybe_hot_count_p, counts_to_freqs): UPdate.
28491 * print-rtl-function.c: Do not include cfg.h
28492 * print-rtl.c: Include basic-block.h
28493 * profile-count.c: New file.
28494 * profile-count.h: New file.
28495 * profile.c (is_edge_inconsistent): Update.
28496 (correct_negative_edge_counts): Update.
28497 (is_inconsistent): Update.
28498 (set_bb_counts): Update.
28499 (read_profile_edge_counts): Update.
28500 (compute_frequency_overlap): Update.
28501 (compute_branch_probabilities): Update; Initialize and deinitialize
28503 (branch_prob): Update.
28504 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28505 (edge_gcov_count): New.
28506 (bb_gcov_count): New.
28507 * shrink-wrap.c (try_shrink_wrapping): Update.
28508 * tracer.c (better_p): Update.
28509 * trans-mem.c (expand_transaction): Update.
28510 (ipa_tm_insert_irr_call): Update.
28511 (ipa_tm_insert_gettmclone_call): Update.
28512 * tree-call-cdce.c: Update.
28513 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28514 (gimple_duplicate_sese_tail): Update.
28515 (gimple_account_profile_record): Update.
28516 (execute_fixup_cfg): Update.
28517 * tree-inline.c (copy_bb): Update.
28518 (copy_edges_for_bb): Update.
28519 (initialize_cfun): Update.
28520 (freqs_to_counts): Update.
28521 (copy_cfg_body): Update.
28522 (expand_call_inline): Update.
28523 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28524 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28525 (try_unroll_loop_completely): Update.
28526 (try_peel_loop): Update.
28527 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28528 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28529 * tree-ssa-loop-split.c (connect_loops): Update.
28530 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28531 * tree-ssa-reassoc.c (branch_fixup): Update.
28532 * tree-ssa-tail-merge.c (replace_block_by): Update.
28533 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28534 (compute_path_counts): Update.
28535 (update_profile): Update.
28536 (recompute_probabilities): Update.
28537 (update_joiner_offpath_counts): Update.
28538 (estimated_freqs_path): Update.
28539 (freqs_to_counts_path): Update.
28540 (clear_counts_path): Update.
28541 (ssa_fix_duplicate_block_edges): Update.
28542 (duplicate_thread_path): Update.
28543 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28544 (struct switch_conv_info): Update.
28545 * tree-tailcall.c (decrease_profile): Update.
28546 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28547 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28548 * value-prof.c (check_counter): Update.
28549 (gimple_divmod_fixed_value): Update.
28550 (gimple_mod_pow2): Update.
28551 (gimple_mod_subtract): Update.
28552 (gimple_ic_transform): Update.
28553 (gimple_stringop_fixed_value): Update.
28554 * value-prof.h (gimple_ic): Update.
28556 2017-06-02 Carl Love <cel@us.ibm.com>
28558 * config/rs6000/rs6000-c: Add support for built-in functions
28559 vector double vec_doublee (vector signed int);
28560 vector double vec_doublee (vector unsigned int);
28561 vector double vec_doublee (vector float);
28562 vector double vec_doubleh (vector signed int);
28563 vector double vec_doubleh (vector unsigned int);
28564 vector double vec_doubleh (vector float);
28565 vector double vec_doublel (vector signed int);
28566 vector double vec_doublel (vector unsigned int);
28567 vector double vec_doublel (vector float);
28568 vector double vec_doubleo (vector signed int);
28569 vector double vec_doubleo (vector unsigned int);
28570 vector double vec_doubleo (vector float);.
28571 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28572 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28574 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28575 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28576 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28578 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28579 vec_doublel, vec_doubleh.
28580 * doc/extend.texi: Update the built-in documentation file for the
28581 new built-in functions.
28583 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28586 * ipa-inline-analysis.c (free_growth_caches): Set
28587 edge_removal_hook_holder to NULL after removing it.
28589 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28591 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28592 comparision with zero.
28594 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28595 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28596 for early expansion of vec_min and vec_max builtins.
28597 (builtin_function_type): Add min/max unsigned variants to those
28598 identified as having unsigned arguments.
28600 2017-06-02 Olivier Hainque <hainque@adacore.com>
28602 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28604 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28606 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28607 Use VALL_F16 iterator rather than VALL.
28609 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28611 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28612 Emit CBNZ inside loop when doing a strong exchange and comparing
28613 against zero. Generate the CC flags after the loop.
28615 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28617 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28618 (dl_section_ref): New.
28619 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28620 On AIX, append an expression to subtract the size of the
28621 section length to dl_section_ref.
28623 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28625 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28626 for early expansion of vector absolute builtins.
28628 2017-06-02 Richard Biener <rguenther@suse.de>
28630 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28631 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28633 2017-06-02 Richard Biener <rguenther@suse.de>
28635 PR tree-optimization/80948
28636 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28637 stmt order as well.
28639 2017-06-02 Richard Biener <rguenther@suse.de>
28641 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28643 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28644 for inductions as relevant.
28646 2017-06-02 Richard Biener <rguenther@suse.de>
28648 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28649 (vectorizable_induction): ... this. Remove dead code.
28651 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28653 * builtins. (expand_builtin_alloca): Remove second parameter and
28654 infer its value from the first parameter instead.
28655 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28657 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28659 PR rtl-optimization/80903
28660 * loop-doloop.c (add_test): Unshare sequence.
28662 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28664 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28666 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28668 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28670 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28671 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28672 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28673 (xlogue_layout::get_stub_rtx): Make static.
28674 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28675 (xlogue_layout::compute_stub_managed_regs): Rename to...
28676 (xlogue_layout::count_stub_managed_regs): ...this.
28677 (xlogue_layout::is_stub_managed_reg): New function.
28678 (xlogue_layout::m_stub_names): Rename to...
28679 (xlogue_layout::s_stub_names): ...this, make static.
28680 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28681 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28682 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28683 xlogue_layout::s_stub_names): Instantiate statics.
28684 (stub_managed_regs): Remove.
28685 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28686 (disable_call_ms2sysv_xlogues): Rename to...
28687 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28688 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28690 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28691 change after reload_completed.
28692 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28694 (ix86_expand_prologue): Likewise.
28695 (ix86_expand_epilogue): Likewise.
28696 (ix86_expand_split_stack_prologue): Likewise.
28697 (ix86_compute_frame_layout): Remove frame parameter ...
28698 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28699 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28701 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28702 (ix86_frame): Move from here ...
28703 * config/i386/i386.h (ix86_frame): ... to here.
28704 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28705 complete ix86_frame data structure instead. Remove some_ld_name.
28707 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28709 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28710 symbols that hold a DECL_VALUE_EXPR.
28712 2017-06-01 Martin Jambor <mjambor@suse.cz>
28714 PR tree-optimization/80898
28715 * tree-sra.c (process_subtree_disqualification): Removed.
28716 (disqualify_candidate): Do not acll
28717 process_subtree_disqualification.
28718 (subtree_mark_written_and_enqueue): New function.
28719 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28720 RHS has been disqualified and re-queue LHS if necessary. Apart
28721 from that, ignore disqualified RHS.
28723 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28725 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28726 address fetch for z10 or later.
28728 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28730 * config/arc/arc.md (tst_movb): Add guard when splitting.
28732 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28734 * config/arc/arc.c (arc_can_eliminate): Test against
28735 arc_frame_pointer_needed.
28737 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28739 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28740 to prevent store reordering.
28741 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28742 (type): Add block type.
28743 (stack_tie): Define special instruction to be used in
28746 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28748 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28749 constraint. It is not valid for the pattern.
28750 (noncommutative_binary_comparison): Likewise.
28752 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28754 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28757 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28759 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28760 be used by the reg-alloc.
28762 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28764 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28765 reg-alloc when having mul64 or mul32x16 instructions.
28766 (mulsidi3): Likewise.
28767 (umulsidi3): Likewise.
28768 (mulsi32x16): New pattern.
28769 (mulsi64): Likewise.
28770 (mulsidi64): Likewise.
28771 (umulsidi64): Likewise.
28772 (MUL32x16_REG): Define.
28773 (mul64_600): Use MUL32x16_REG.
28774 (mac64_600): Likewise.
28775 (umul64_600): Likewise.
28776 (umac64_600): Likewise.
28778 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28780 * config/arc/arc.md (mulsi3_700): Make it commutative.
28782 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28784 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28786 (*sign_extendsidi2_insn): Likewise for movstosw.
28788 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28790 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28791 the type of the input discriminant value. Convert the
28792 discriminant value of signedness vary.
28794 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28796 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28797 Add to -Wall section.
28799 2017-06-01 Richard Biener <rguenther@suse.de>
28801 PR middle-end/66313
28802 * fold-const.c (fold_plusminus_mult_expr): If the factored
28803 factor may be zero use a wrapping type for the inner operation.
28804 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28805 and handle moved defs.
28806 (process_assignment): Properly guard the unary op case. Return a
28807 tri-state indicating that moving the stmt before the call may allow
28808 to continue. Pass through to_move.
28809 (find_tail_calls): Handle moving unrelated defs before
28812 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28815 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28816 splitter result in the canonical way.
28818 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28820 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28821 also for 32bit target. Update insn attributes.
28822 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28824 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28826 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28827 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28828 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28829 (_mm_maskz_min_ss): New intrinsics.
28831 2017-05-31 Martin Liska <mliska@suse.cz>
28833 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28834 Change comment style to one we normally use.
28835 (vectorizable_reduction): Likewise.
28836 (vectorizable_induction): Likewise.
28837 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28838 (vectorizable_call): Likewise.
28839 (vectorizable_simd_clone_call): Likewise.
28840 (vectorizable_conversion): Likewise.
28841 (vectorizable_assignment): Likewise.
28842 (vectorizable_shift): Likewise.
28843 (vectorizable_operation): Likewise.
28844 (vectorizable_store): Likewise.
28845 (vectorizable_load): Likewise.
28846 * tree-vectorizer.h: Likewise.
28848 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28850 * passes.c (emergency_dump_function): New.
28851 * tree-pass.h (emergency_dump_function): Declare.
28852 * plugin.c (plugins_internal_error_function): Remove.
28853 * plugin.h (plugins_internal_error_function): Remove declaration.
28854 * toplev.c (internal_error_function): New static function. Use it...
28855 (general_init): ...here.
28857 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28859 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28860 (arc_rtx_costs): Add costs for new patterns.
28861 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28862 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28864 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28866 * tree-ssa-strlen.c (get_next_strinfo): New function.
28867 (get_stridx_plus_constant): Use it.
28868 (zero_length_string): Likewise.
28869 (adjust_related_strinfos): Likewise.
28870 (adjust_last_stmt): Likewise.
28872 2017-05-31 Richard Biener <rguenther@suse.de>
28875 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28876 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28878 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28880 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28881 loop_vinfo argument and use of dependence distance vectors.
28882 Check instead whether the two references differ only in their
28883 initial value and assume that they have the same alignment if the
28884 difference is a multiple of the vector alignment.
28885 (vect_analyze_data_refs_alignment): Update call accordingly.
28887 2017-05-31 Martin Liska <mliska@suse.cz>
28890 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28892 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28894 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28895 (create_intersect_range_checks): Move from ...
28896 * tree-data-ref.c (create_intersect_range_checks_index)
28897 (create_intersect_range_checks): ... to here.
28898 (create_runtime_alias_checks): New function factored from ...
28899 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28900 here. Call above function.
28901 * tree-data-ref.h (create_runtime_alias_checks): New function.
28903 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28905 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28906 segment length for dr_b and compute it in wide_int.
28908 2017-05-31 Richard Biener <rguenther@suse.de>
28910 PR tree-optimization/80906
28911 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28912 and pass through iv_map.
28913 (copy_bb_and_scalar_dependences): Adjust.
28914 (translate_pending_phi_nodes): Likewise.
28915 (copy_loop_close_phi_args): Handle code-generating IVs instead
28918 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28920 * diagnostic-color.c (color_dict): Add "type-diff".
28921 (parse_gcc_colors): Update comment.
28922 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28923 -fdiagnostics-show-template-tree and -fno-elide-type.
28924 (GCC_COLORS): Add type-diff to example.
28926 (-fdiagnostics-show-template-tree): New.
28927 (-fno-elide-type): New.
28928 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28929 the pp_format_decoder callback. Call any m_format_postprocessor's
28931 (pretty_printer::pretty_printer): Initialize
28932 m_format_postprocessor.
28933 (pretty_printer::~pretty_printer): Delete any
28934 m_format_postprocessor.
28935 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28936 (class format_postprocessor): New class.
28937 (struct pretty_printer::format_decoder): Document the new parameters.
28938 (struct pretty_printer::m_format_postprocessor): New field.
28939 * tree-diagnostic.c (default_tree_printer): Update for new
28940 bool and const char ** params.
28941 * tree-diagnostic.h (default_tree_printer): Likewise.
28943 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28945 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28946 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28947 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28948 rs6000_gen_cell_microcode code.
28949 (rs6000_final_prescan_insn): Delete.
28950 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28951 "warn-cell-microcode" entries.
28952 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28953 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28954 throughout. Change cc_reg_not_micro_cr0_operand to
28955 cc_reg_not_cr0_operand throughout.
28956 (*extendhi<mode>2_noload): Delete.
28957 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28958 (mwarn-cell-microcode): Delete.
28959 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28960 -mgen-cell-microcode and -mwarn-cell-microcode.
28962 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28965 * config/i386/constraints.md (Yd): New constraint.
28967 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28968 and (?Yd, r) alternatives. Update insn attributes.
28969 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28970 and (?*Yd, r) alternatives. Update insn attributes.
28971 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28973 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28975 * gimplify.c (gimplify_modify_expr): Don't create a
28976 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28979 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28981 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28983 2017-05-30 Richard Biener <rguenther@suse.de>
28985 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28986 and reduc_def fields.
28987 (STMT_VINFO_REDUC_TYPE): New define.
28988 (STMT_VINFO_REDUC_DEF): Likewise.
28989 (vect_force_simple_reduction): Adjust prototype.
28990 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28991 (vect_is_simple_reduction): Remove check_reduction argument.
28992 (vect_force_simple_reduction): Adjust and set
28993 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28994 (vectorizable_reduction): Do not re-do reduction analysis
28995 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28996 * tree-parloops.c (gather_scalar_reductions): Adjust.
28998 2017-05-30 Richard Biener <rguenther@suse.de>
29000 PR middle-end/80901
29001 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29004 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29006 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29007 Introduce unknown_misalignment parameter and remove vf.
29008 (vect_peeling_hash_get_lowest_cost):
29009 Pass unknown_misalignment parameter.
29010 (vect_enhance_data_refs_alignment):
29011 Fix unsupportable data ref treatment.
29013 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29015 * tree-vect-data-refs.c (vect_get_data_access_cost):
29016 Workaround for SLP handling.
29017 (vect_enhance_data_refs_alignment):
29018 Compute costs for doing no peeling at all, compare to the best
29019 peeling costs so far and avoid peeling if cheaper.
29021 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29023 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29024 Return peeling info and set costs to zero for unlimited cost
29026 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29027 with unknown misalignment. Compute and costs for unknown
29028 misalignment, compare them to the costs for known misalignment
29029 and choose the cheapest for peeling.
29031 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29033 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29034 (vect_get_peeling_costs_all_drs): Create function.
29035 (vect_peeling_hash_get_lowest_cost):
29036 Use vect_get_peeling_costs_all_drs.
29037 (vect_peeling_supportable): Create function.
29038 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29040 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29042 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29043 DR_HAS_NEGATIVE_STEP.
29044 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29045 (vect_enhance_data_refs_alignment): Use.
29046 (vect_duplicate_ssa_name_ptr_info): Use.
29047 * tree-vectorizer.h (dr_misalignment): Use.
29048 (known_alignment_for_access_p): Use.
29050 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29053 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29055 (has_section_name): New function.
29057 2017-05-30 Martin Liska <mliska@suse.cz>
29060 * auto-profile.c (get_function_decl_from_block): Fix
29063 2017-05-30 Richard Biener <rguenther@suse.de>
29065 PR middle-end/80876
29066 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29068 2017-05-30 Martin Liska <mliska@suse.cz>
29070 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29071 * dumpfile.h (struct dump_file_info): Remove ctors.
29073 2017-05-30 Martin Liska <mliska@suse.cz>
29075 * predict.def: Fix GNU coding style.
29077 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29079 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29080 Mark 'to' argument with ATTRIBUTE_UNUSED.
29082 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29084 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29085 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29086 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29089 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29091 * doc/install.texi (Options specification): Restore entry of
29092 --enable-sjlj-exceptions.
29094 2017-05-27 Michael Eager <eager@eagercon.com>
29097 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29099 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29101 * config/microblaze/microblaze.h
29102 (FIXED_REGISTERS): Update in macro.
29103 (CALL_USED_REGISTERS): Update in macro.
29105 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29107 * doc/install.texi: Add links to macOS binary distributions.
29109 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29113 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29115 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29117 2017-05-26 Martin Liska <mliska@suse.cz>
29119 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29121 2017-05-26 Martin Liska <mliska@suse.cz>
29123 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29124 always leading ';; '.
29125 (dump_bb_info): Likewise.
29126 (brief_dump_cfg): Likewise.
29127 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29128 * dumpfile.c: Remove usage of TDF_VERBOSE.
29129 * dumpfile.h (enum dump_kind): Likewise.
29130 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29131 * print-tree.c (debug_verbose): Remove.
29132 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29133 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29134 * tree-diagnostic.c (default_tree_printer): Replace
29135 TDF_DIAGNOSTIC with TDF_SLIM.
29137 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29139 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29140 in parameter loop, rather than loop_vinfo.
29141 (create_intersect_range_checks): Ditto.
29142 (vect_create_cond_for_alias_checks): Update call to above functions.
29144 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29146 PR tree-optimization/80815
29147 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29148 for merging runtime alias checks. Handle negative DR_STEPs.
29150 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29152 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29154 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29155 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29156 out code pruning runtime alias checks.
29157 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29158 factored out from above.
29159 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29161 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29163 (prune_runtime_alias_test_list): New decalaration.
29165 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29167 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29168 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29169 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29170 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29171 (operator==, comp_dr_with_seg_len_pair): Ditto.
29172 (vect_prune_runtime_alias_test_list): Ditto.
29174 2017-05-26 Martin Liska <mliska@suse.cz>
29177 * params.def: Bound partial-inlining-entry-probability param.
29179 2017-05-26 Marek Polacek <polacek@redhat.com>
29182 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29185 2017-05-26 Richard Biener <rguenther@suse.de>
29187 PR tree-optimization/80842
29188 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29191 2017-05-26 Richard Biener <rguenther@suse.de>
29193 PR tree-optimization/80844
29194 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29196 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29198 * doc/md.texi (Machine Constraints): Update x86 family
29199 machine constraints section to match 'config/i386/constraints.md'.
29201 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29203 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29205 2017-05-25 Nathan Sidwell <nathan@acm.org>
29207 * doc/invoke.texi (--enable-languages): Update documentation.
29209 2017-05-25 Martin Liska <mliska@suse.cz>
29211 * dumpfile.c: Add TDF_FOLDING.
29212 * dumpfile.h (enum dump_kind): Likewise.
29213 * genmatch.c (dt_simplify::gen_1): Use it.
29215 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29217 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29219 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29221 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29222 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29224 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29226 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29227 * match.pd (X == C): Rewrite it here.
29228 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29229 with_certain_nonzero_bits2): New predicates.
29230 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29232 2017-05-24 Nathan Sidwell <nathan@acm.org>
29234 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29237 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29240 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29242 * config/powerpcspe: New port. Files are copied from the rs6000
29243 port, with "rs6000" in filenames replaced by "powerpcspe".
29245 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29247 PR rtl-optimization/80754
29248 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29250 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29252 * config/sparc/sparc.md (length): Return the correct value for -mflat
29253 sibcalls to match output_sibcall.
29255 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29259 * config/rs6000/rs6000.c (struct machine_function): Add new field
29261 (rs6000_get_separate_components): Init that field, use it.
29262 (rs6000_components_for_bb): Use the field.
29264 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29266 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29268 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29270 PR middle-end/80823
29271 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29273 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29276 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29277 address against address_operand predicate.
29278 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29280 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29282 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29283 parameters passed indirectly.
29285 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29287 * config/i386/i386.md (*movdi_internal): Remove SSE4
29288 alternative 18 (?r, *v). Update insn attributes.
29289 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29290 Update insn attributes.
29291 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29292 Update insn attributes.
29293 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29294 alternative 1 (r, v). Remove isa attribute.
29295 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29296 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29297 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29299 2017-05-23 Tom de Vries <tom@codesourcery.com>
29301 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29304 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29306 * cgraphunit.c (symbol_table::process_new_functions): Update.
29307 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29308 (inline_generate_summary): Rename to ...
29309 (ipa_fn_summary_generate): ... this one.
29310 (inline_read_summary): Rename to ...
29311 (ipa_fn_summary_read): ... this one.
29312 (inline_write_summary): Rename to ...
29313 (ipa_fn_summary_write): ... this one.
29314 (inline_free_summary): Rename to ...
29315 (ipa_free_fn_summary): ... this one.
29316 (pass_data_local_fn_summary, pass_local_fn_summary,
29317 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29318 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29319 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29320 make_pass_ipa_fn_summary): New.
29321 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29322 inline_write_summary, inline_free_summary): Remove.
29323 (ipa_free_fn_summary) : New.
29324 * ipa-inline.c (ipa_inline): Update.
29325 (pass_ipa_inline): Do not generate summaries.
29326 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29328 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29329 and add pass_ipa_fn_summary.
29330 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29332 (make_pass_inline_parameters): Remove.
29334 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29336 * omp-low.c (struct omp_context): Remove "default_kind" member.
29339 * omp-offload.c (execute_oacc_device_lower): Remove the
29340 parallelism dimensions function attributes for unparallelized
29341 OpenACC kernels constructs.
29343 2017-05-23 Martin Liska <mliska@suse.cz>
29345 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29347 (cgraph_edge::make_speculative): Likewise.
29348 (cgraph_edge::resolve_speculation): Likewise.
29349 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29350 (cgraph_node::dump): Likewise.
29351 * cgraph.h: Likewise.
29352 * cgraphunit.c (analyze_functions): Likewise.
29353 (symbol_table::compile): Likewise.
29354 * ipa-cp.c (print_all_lattices): Likewise.
29355 (determine_versionability): Likewise.
29356 (initialize_node_lattices): Likewise.
29357 (ipcp_verify_propagated_values): Likewise.
29358 (estimate_local_effects): Likewise.
29359 (update_profiling_info): Likewise.
29360 (create_specialized_node): Likewise.
29361 (perhaps_add_new_callers): Likewise.
29362 (decide_about_value): Likewise.
29363 (decide_whether_version_node): Likewise.
29364 (identify_dead_nodes): Likewise.
29365 (ipcp_store_bits_results): Likewise.
29366 * ipa-devirt.c (dump_targets): Likewise.
29367 (ipa_devirt): Likewise.
29368 * ipa-icf.c (sem_item::dump): Likewise.
29369 (sem_function::equals): Likewise.
29370 (sem_variable::equals): Likewise.
29371 (sem_item_optimizer::read_section): Likewise.
29372 (sem_item_optimizer::execute): Likewise.
29373 (congruence_class::dump): Likewise.
29374 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29375 (dump_inline_summary): Likewise.
29376 (estimate_node_size_and_time): Likewise.
29377 (inline_analyze_function): Likewise.
29378 * ipa-inline-transform.c (inline_call): Likewise.
29379 * ipa-inline.c (report_inline_failed_reason): Likewise.
29380 (want_early_inline_function_p): Likewise.
29381 (edge_badness): Likewise.
29382 (update_edge_key): Likewise.
29383 (inline_small_functions): Likewise.
29384 * ipa-profile.c (ipa_profile): Likewise.
29385 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29386 (ipa_make_edge_direct_to_target): Likewise.
29387 (remove_described_reference): Likewise.
29388 (ipa_impossible_devirt_target): Likewise.
29389 (propagate_controlled_uses): Likewise.
29390 (ipa_print_node_params): Likewise.
29391 (ipcp_transform_function): Likewise.
29392 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29393 (propagate_pure_const): Likewise.
29394 * ipa-reference.c (generate_summary): Likewise.
29395 (read_write_all_from_decl): Likewise.
29396 (propagate): Likewise.
29397 (ipa_reference_read_optimization_summary): Likewise.
29398 * ipa-utils.c (ipa_merge_profiles): Likewise.
29399 * ipa.c (walk_polymorphic_call_targets): Likewise.
29400 (symbol_table::remove_unreachable_nodes): Likewise.
29401 (ipa_single_use): Likewise.
29402 * passes.c (execute_todo): Likewise.
29403 * predict.c (drop_profile): Likewise.
29404 * symtab.c (symtab_node::get_dump_name): New function.
29405 (symtab_node::dump_name): Likewise.
29406 (symtab_node::dump_asm_name): Likewise.
29407 (symtab_node::dump_references): Likewise.
29408 (symtab_node::dump_referring): Likewise.
29409 (symtab_node::dump_base): Likewise.
29410 (symtab_node::debug_symtab): Likewise.
29411 * tree-sra.c (convert_callers_for_node): Likewise.
29412 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29413 * value-prof.c (init_node_map): Likewise.
29415 2017-05-23 Martin Liska <mliska@suse.cz>
29417 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29418 and symtab_node::debug_symtab to symbol_table::debug.
29419 * cgraphunit.c (analyze_functions): Use the renamed function.
29420 (symbol_table::compile): Likewise.
29421 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29422 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29423 * passes.c (execute_todo): Likewise.
29424 * symtab.c (symbol_table::dump): New function.
29425 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29427 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29429 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29430 that nonconst implies exec.
29432 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29434 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29435 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29436 (inline_edge_summary_vec): Turn into ...
29437 (ipa_call_summaries): ... this one.
29438 (redirect_to_unreachable, edge_set_predicate,
29439 evaluate_properties_for_edge, inline_summary_alloc,
29440 reset_ipa_call_summary, reset_inline_summary,
29441 inline_summary_t::duplicate): Update.
29442 (inline_edge_duplication_hook): Turn to ...
29443 (ipa_call_summary_t::duplicate): ... this one.
29444 (inline_edge_removal_hook): Turn to ...
29445 (ipa_call_summary_t::remove): ... this one.
29446 (dump_inline_edge_summary): Turn to ...
29447 (dump_ipa_call_summary): ... this one.
29448 (estimate_function_body_sizes): Update.
29449 (inline_update_callee_summaries): Update.
29450 (remap_edge_change_prob): Update.
29451 (remap_edge_summaries): Update.
29452 (inline_merge_summary): Update.
29453 (do_estimate_edge_time): Update.
29454 (inline_generate_summary): Update.
29455 (inline_read_section): Update.
29456 (inline_read_summary): Update.
29457 (inline_free_summary): Update.
29458 * ipa-inline.c (can_inline_edge_p): Update.
29459 (compute_inlined_call_time): Update.
29460 (want_inline_small_function_p): Update.
29461 (edge_badness): Update.
29462 (early_inliner): Update.
29463 * ipa-inline.h (inline_edge_summary): Turn to ...
29464 (ipa_call_summary): ... this one.
29465 (ipa_call_summary_t): New class.
29466 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29467 (ipa_call_summaries): New.
29468 (inline_edge_summary): Remove.
29469 (estimate_edge_growth): Update.
29470 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29471 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29472 * ipa-split.c (execute_split_functions): Update.
29473 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29475 2017-05-23 Tom de Vries <tom@codesourcery.com>
29477 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29478 attributes): Document rdrand effective target.
29480 2017-05-23 Tom de Vries <tom@codesourcery.com>
29482 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29483 attributes): Sort alphabetically.
29485 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29487 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29489 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29492 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29493 V2DF/V2DI splat into two separate patterns, one that handles
29494 registers, and the other that only handles memory. Drop support
29495 for splatting from a GPR on ISA 2.07 and then splitting the
29496 splat into direct move and splat.
29497 (vsx_splat_<mode>_reg): Likewise.
29498 (vsx_splat_<mode>_mem): Likewise.
29500 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29502 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29504 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29506 PR middle-end/80809
29507 * omp-low.c (finish_taskreg_remap): New function.
29508 (finish_taskreg_scan): If unit size of ctx->record_type
29509 is non-constant, unshare the size expression and replace
29510 decls in it with possible outer var refs.
29512 PR middle-end/80809
29513 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29514 GOVD_SHARED rather than GOVD_PRIVATE with it.
29515 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29516 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29518 PR middle-end/80853
29519 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29520 as last argument to build_outer_var_ref for pointer bases of array
29521 section reductions.
29523 2017-05-19 Martin Sebor <msebor@redhat.com>
29525 * print-tree.c (print_node): Print DECL_READ_P flag.
29527 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29529 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29530 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29531 * cgraph.c: Likewise.
29532 * cgraphunit.c: Likewise.
29533 * gengtype.c: Likewise.
29534 * ipa-cp.c: Likewise.
29535 * ipa-devirt.c: Likewise.
29536 * ipa-icf.c: Likewise.
29537 * ipa-predicate.c: Likewise.
29538 * ipa-profile.c: Likewise.
29539 * ipa-prop.c: Likewise.
29540 * ipa-split.c: Likewise.
29542 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29543 edge_predicate_pool, dump_inline_hints,
29544 inline_summary::account_size_time, redirect_to_unreachable,
29545 edge_set_predicate, set_hint_predicate,
29546 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29547 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29548 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29549 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29550 ipa_call_summary_t::remove, initialize_growth_caches,
29551 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29552 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29553 mark_modified, unmodified_parm_1, unmodified_parm,
29554 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29555 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29556 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29557 will_be_nonconstant_predicate, record_modified_bb_info,
29558 get_minimal_bb, record_modified, param_change_prob,
29559 phi_result_unknown_predicate, predicate_for_phi_result,
29560 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29561 estimate_function_body_sizes, compute_inline_parameters,
29562 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29563 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29564 inline_update_callee_summaries, remap_edge_change_prob,
29565 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29566 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29567 inline_analyze_function, inline_summary_t::insert,
29568 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29569 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29570 inline_free_summary): Move to ipa-fnsummary.h
29571 (predicate_t): Remove.
29572 * ipa-fnsummary.c: New file.
29573 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29574 (enum inline_hints_vals, inline_hints, agg_position_info,
29575 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29576 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29577 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29578 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29579 inline_read_summary, inline_write_summary, inline_free_summary,
29580 inline_analyze_function, initialize_inline_failed,
29581 inline_merge_summary, inline_update_overall_summary,
29582 compute_inline_parameters): Move to ipa-fnsummary.h
29583 * ipa-fnsummary.h: New file.
29584 * ipa-inline-transform.h: Include ipa-inline.h.
29585 * ipa-inline.c: LIkewise.
29587 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29589 * ipa-inline.c (edge_badness): Use inlined_time instead of
29590 inline_summaries->get.
29592 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29594 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29596 2017-05-22 Nathan Sidwell <nathan@acm.org>
29598 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29599 (fdump-lang): Document 'raw' option.
29600 * dumpfile.h (TDI_tu): Delete.
29601 * dumpfile.c (dump_files): Remove translation-unit.
29602 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29604 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29606 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29607 command option from $(AWK) call.
29608 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29609 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29610 [FORMAT]: Remove handling of variable.
29611 * config/avr/t-multilib: Regenerate.
29613 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29615 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29617 (dump_inline_summary): Do not print self_time.
29618 (estimate_function_body_sizes): Do not set self_time.
29619 (compute_inline_parameters): Likewise.
29620 (inline_read_section, inline_write_summary): Do not stream self_time.
29621 * ipa-inline.h (inline_summary): Drop self_time.
29623 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29625 * ipa-inline-analysis.c (account_size_time): Rename to ...
29626 (inline_summary::account_size_time): ... this one.
29627 (reset_ipa_call_summary): Turn to ...
29628 (ipa_call_summary::reset): ... this one.
29629 (reset_inline_summary): Turn to ...
29630 (inline_summary::reset): ... this one.
29631 (inline_summary_t::remove): Update.
29632 (inline_summary_t::duplicate): Update.
29633 (ipa_call_summary_t::remove): Update.
29634 (dump_inline_summary): Update.
29635 (estimate_function_body_sizes): Update.
29636 (compute_inline_parameters): Update.
29637 (estimate_node_size_and_time): Update.
29638 (inline_merge_summary): Update.
29639 (inline_update_overall_summary): Update.
29640 (inline_read_section): Update.
29641 (inline_write_summary): Update.
29642 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29643 add account_size_time and reset member functions.
29644 (ipa_call_summary): Add reset function.
29645 * ipa-predicate.h (predicate::operator &): Constify.
29647 2017-05-22 Richard Biener <rguenther@suse.de>
29649 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29651 2017-05-19 Jason Merrill <jason@redhat.com>
29653 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29655 2017-05-19 Marek Polacek <polacek@redhat.com>
29658 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29659 TYPE_OVERFLOW_WRAPS checks.
29661 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29663 * tree-core.h (enum omp_clause_default_kind): Add
29664 "OMP_CLAUSE_DEFAULT_PRESENT".
29665 * tree-pretty-print.c (dump_omp_clause): Handle it.
29666 * gimplify.c (enum gimplify_omp_var_data): Add
29667 "GOVD_MAP_FORCE_PRESENT".
29668 (gimplify_adjust_omp_clauses_1): Map it to
29669 "GOMP_MAP_FORCE_PRESENT".
29670 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29672 * gimplify.c (oacc_default_clause): Clarify.
29674 2017-05-19 Nathan Sidwell <nathan@acm.org>
29676 LANG_HOOK_REGISTER_DUMPS
29677 * toplev.c (general_init): Call register dump lang hook.
29678 * doc/invoke.texi: Document -fdump-lang option family.
29679 * dumpfile.c (dump_files): Remove class dump here.
29680 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29681 * dumpfile.h (tree_dump_index): Remove TDI_class.
29682 * langhooks-def.h (lhd_register_dumps): Declare.
29683 (LANG_HOOKS_REGISTER_DUMPS): Define.
29684 (LANG_HOOKS_INITIALIZER): Add it.
29685 * langhooks.c (lhd_register_dumps): Define.
29686 * langhooks.h (struct lang_hooks): Add register_dumps.
29688 2017-05-19 Nathan Sidwell <nathan@acm.org>
29690 * context.h (context::set_passes): New.
29691 * context.c (context::context): Do not create pass manager.
29692 * toplev.c (general_init): Create pass manager here.
29694 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29696 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29697 use this splitter if two add or or instructions would also work for
29698 the constant we want to generate.
29700 2017-05-19 Richard Biener <rguenther@suse.de>
29703 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29704 predicate evaluation.
29706 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29708 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29710 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29713 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29715 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29716 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29717 (inline_edge_summary_vec): Turn into ...
29718 (ipa_call_summaries): ... this one.
29719 (redirect_to_unreachable, edge_set_predicate,
29720 evaluate_properties_for_edge, inline_summary_alloc,
29721 reset_ipa_call_summary, reset_inline_summary,
29722 inline_summary_t::duplicate): Update.
29723 (inline_edge_duplication_hook): Turn to ...
29724 (ipa_call_summary_t::duplicate): ... this one.
29725 (inline_edge_removal_hook): Turn to ...
29726 (ipa_call_summary_t::remove): ... this one.
29727 (dump_inline_edge_summary): Turn to ...
29728 (dump_ipa_call_summary): ... this one.
29729 (estimate_function_body_sizes): Update.
29730 (inline_update_callee_summaries): Update.
29731 (remap_edge_change_prob): Update.
29732 (remap_edge_summaries): Update.
29733 (inline_merge_summary): Update.
29734 (do_estimate_edge_time): Update.
29735 (inline_generate_summary): Update.
29736 (inline_read_section): Update.
29737 (inline_read_summary): Update.
29738 (inline_free_summary): Update.
29739 * ipa-inline.c (can_inline_edge_p): Update.
29740 (compute_inlined_call_time): Update.
29741 (want_inline_small_function_p): Update.
29742 (edge_badness): Update.
29743 (early_inliner): Update.
29744 * ipa-inline.h (inline_edge_summary): Turn to ...
29745 (ipa_call_summary): ... this one.
29746 (ipa_call_summary_t): New class.
29747 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29748 (ipa_call_summaries): New.
29749 (inline_edge_summary): Remove.
29750 (estimate_edge_growth): Update.
29751 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29752 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29753 * ipa-split.c (execute_split_functions): Update.
29754 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29756 2017-05-19 Richard Biener <rguenther@suse.de>
29758 PR middle-end/80764
29759 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29761 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29763 * config/rs6000/rs6000.c (struct machine_function): Add field
29764 fpr_is_wrapped_separately.
29765 (rs6000_get_separate_components): Use 64 components. Handle the
29766 new FPR components.
29767 (rs6000_components_for_bb): Handle the FPR components.
29768 (rs6000_emit_prologue_components): Handle the FPR components.
29769 (rs6000_emit_epilogue_components): Handle the FPR components.
29770 (rs6000_set_handled_components): Handle the FPR components.
29771 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29772 that are already separately shrink-wrapped.
29773 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29774 that are already separately shrink-wrapped.
29776 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29779 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29782 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29783 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29784 cases where the register allocator uses a move and an offsettable
29785 memory operation to/from a FPR register on ISA 2.06/2.07.
29786 (define_peephole2 for Altivec d-form store): Likewise.
29788 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29791 * config/i386/mmx.md (*mov<mode>_internal): Enable
29792 alternatives 11, 12, 13 and 14 also for 32bit targets.
29793 Remove alternatives 15, 16, 17 and 18.
29794 * config/i386/sse.md (vec_concatv2di): Change
29795 alternative (!x, *y) to (x, ?!*Yn).
29797 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29799 * dumpfile.h (enum dump_kind): Remove stray comma.
29801 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29803 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29804 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29805 predicate::num_conditions
29806 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29807 (CHANGED): turn into predicate::changed.
29808 (agg_position_info): Move to ipa-predicate.h
29809 (add_condition, predicate::add_clause, predicate::operator &=,
29810 predicate::or_with, predicate::evaluate, predicate::probability,
29811 dump_condition, dump_clause, predicate::dump,
29812 predicate::remap_after_duplication, predicate::remap_after_inlining,
29813 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29814 (evaluate_conditions_for_known_args): Update.
29815 (set_cond_stmt_execution_predicate): Update.
29816 * ipa-inline.h: Include ipa-predicate.h
29817 (condition, inline_param_summary, conditions, agg_position_info,
29818 predicate): Move to ipa-predicate.h
29819 * ipa-predicate.c: New file.
29820 * ipa-predicate.h: New file.
29822 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29824 * final.c (leaf_function_p): Check we are not in a sequence.
29826 2017-05-18 Martin Liska <mliska@suse.cz>
29828 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29829 * dumpfile.c (dump_register): Use new enum dump_kind.
29830 (get_dump_file_name): Likewise.
29831 (dump_enable_all): Likewise.
29832 (dump_switch_p_1): Likewise.
29833 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29834 * dumpfile.h (enum dump_kind): New enum type.
29835 (struct dump_file_info): Create constructor and
29836 format fields and comments.
29837 * passes.c (pass_manager::register_one_dump_file):
29839 * statistics.c (statistics_early_init): Likewise.
29840 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29841 TDF_TREE with TDF_SLIM.
29842 (gather_memory_references_ref): Likewise.
29844 2017-05-18 Martin Liska <mliska@suse.cz>
29846 * vec.h (struct vnull): Use it.
29848 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29850 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29851 (true_predicate, false_predicate, true_predicate_p,
29852 false_predicate_p): Remove.
29853 (single_cond_predicate, not_inlined_predicate): Turn to member function
29855 (add_condition): Update.
29856 (add_clause): Turn to...
29857 (predicate::add_clause): ... this one; update; allow passing NULL
29859 (and_predicates): Turn to ...
29860 (predicate::operator &=): ... this one.
29861 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29862 (or_predicates): Turn to ...
29863 (predicate::or_with): ... this one.
29864 (evaluate_predicate): Turn to ...
29865 (predicate::evaluate): ... this one.
29866 (predicate_probability): Turn to ...
29867 (predicate::probability): ... this one.
29868 (dump_condition): Update.
29869 (dump_predicate): Turn to ...
29870 (predicate::dump): ... this one.
29871 (account_size_time): Update.
29872 (edge_set_predicate): Update.
29873 (set_hint_predicate): UPdate.
29874 (evaluate_conditions_for_known_args): Update.
29875 (evaluate_properties_for_edge): Update.
29876 (remap_predicate_after_duplication): Turn to...
29877 (predicate::remap_after_duplication): ... this one.
29878 (remap_hint_predicate_after_duplication): Update.
29879 (inline_summary_t::duplicate): UPdate.
29880 (dump_inline_edge_summary): Update.
29881 (dump_inline_summary): Update.
29882 (set_cond_stmt_execution_predicate): Update.
29883 (set_switch_stmt_execution_predicate): Update.
29884 (compute_bb_predicates): Update.
29885 (will_be_nonconstant_expr_predicate): Update.
29886 (will_be_nonconstant_predicate): Update.
29887 (phi_result_unknown_predicate): Update.
29888 (predicate_for_phi_result): Update.
29889 (array_index_predicate): Update.
29890 (estimate_function_body_sizes): Update.
29891 (estimate_node_size_and_time): Update.
29892 (estimate_ipcp_clone_size_and_time): Update.
29893 (remap_predicate): Rename to ...
29894 (predicate::remap_after_inlining): ... this one.
29895 (remap_hint_predicate): Update.
29896 (inline_merge_summary): Update.
29897 (inline_update_overall_summary): Update.
29898 (estimate_size_after_inlining): Update.
29899 (read_predicate): Rename to ...
29900 (predicate::stream_in): ... this one.
29901 (read_inline_edge_summary): Update.
29902 (write_predicate): Rename to ...
29903 (predicate::stream_out): ... this one.
29904 (write_inline_edge_summary): Update.
29905 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29906 (clause_t): Turn to uint32_t
29907 (predicate): Turn to class; implement constructor and operators
29909 (size_time_entry): Update.
29910 (inline_summary): Update.
29911 (inline_edge_summary): Update.
29913 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29915 * fold-const.c (fold_binary_loc): Move transformation...
29916 * match.pd (C - X CMP X): ... here.
29918 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29920 * config/sparc/sparc.c (sparc_option_override): Set function
29921 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29922 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29924 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29926 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29928 2017-05-18 Marek Polacek <polacek@redhat.com>
29931 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29932 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29933 gimple_assign_load_p.
29935 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29937 PR middle-end/80692
29938 * real.c (do_compare): Give decimal_do_compare preference over
29939 comparing just the signs.
29941 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29943 * doc/md.texi (Canonicalization of Instructions): Describe the
29944 canonical form of instructions that inherently set a condition
29947 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29949 PR middle-end/80775
29950 * tree-cfg.c: Move deletion of unreachable case statements to after
29951 the merging of consecutive case labels.
29953 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29955 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29956 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29957 restoring of callee-saved registers.
29959 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29961 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29962 * config/visium/visium.c (single_set_and_flags): Likewise.
29963 * config/visium/visium.md (Substitutions): Likewise.
29965 2017-05-17 Martin Liska <mliska@suse.cz>
29967 * cfg.c: Introduce dump_flags_t type and
29968 use it instead of int type.
29970 * cfghooks.c: Likewise.
29971 * cfghooks.h (struct cfg_hooks): Likewise.
29972 * cfgrtl.c: Likewise.
29973 * cfgrtl.h: Likewise.
29974 * cgraph.c (cgraph_node::get_body): Likewise.
29975 * coretypes.h: Likewise.
29976 * domwalk.c: Likewise.
29977 * domwalk.h: Likewise.
29978 * dumpfile.c (struct dump_option_value_info): Likewise.
29979 (dump_enable_all): Likewise.
29980 (dump_switch_p_1): Likewise.
29981 (opt_info_switch_p): Likewise.
29982 * dumpfile.h (enum tree_dump_index): Likewise.
29983 (struct dump_file_info): Likewise.
29984 * genemit.c: Likewise.
29985 * generic-match-head.c: Likewise.
29986 * gengtype.c (open_base_files): Likewise.
29987 * gimple-pretty-print.c: Likewise.
29988 * gimple-pretty-print.h: Likewise.
29989 * graph.c (print_graph_cfg): Likewise.
29990 * graphite-scop-detection.c (dot_all_sese): Likewise.
29991 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29992 * loop-unroll.c (report_unroll): Likewise.
29993 * passes.c (pass_manager::register_one_dump_file): Likewise.
29994 * print-tree.c: Likewise.
29995 * statistics.c: Likewise.
29996 * tree-cfg.c: Likewise.
29997 * tree-cfg.h: Likewise.
29998 * tree-dfa.c: Likewise.
29999 * tree-dfa.h: Likewise.
30000 * tree-dump.c (dump_function): Likewise.
30001 * tree-dump.h (struct dump_info): Likewise.
30002 * tree-pretty-print.c: Likewise.
30003 * tree-pretty-print.h: Likewise.
30004 * tree-ssa-live.c: Likewise.
30005 * tree-ssa-live.h: Likewise.
30006 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30007 * tree-vect-loop.c: Likewise.
30008 * tree-vect-slp.c: Likewise.
30010 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30011 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30013 PR tree-optimization/80457
30014 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30015 of all arguments to a statement as scalar_to_vec operations.
30016 (vectorizable_call): Adjust call to vect_model_simple_cost for
30018 (vectorizable_conversion): Likewise.
30019 (vectorizable_assignment): Likewise.
30020 (vectorizable_shift): Likewise.
30021 (vectorizable_operation): Likewise.
30022 (vectorizable_comparison): Likewise.
30023 (vect_is_simple_cond): Record the def types for operands.
30024 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30025 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30026 for statement argument count.
30028 2017-05-16 Carl Love <cel@us.ibm.com>
30030 * config/rs6000/rs6000-c: Add support for built-in functions
30031 vector unsigned long long vec_bperm (vector unsigned long long,
30032 vector unsigned char)
30033 vector signed long long vec_mule (vector signed int,
30035 vector unsigned long long vec_mule (vector unsigned int,
30036 vector unsigned int)
30037 vector signed long long vec_mulo (vector signed int,
30039 vector unsigned long long vec_mulo (vector unsigned int,
30040 vector unsigned int)
30041 vector signed char vec_sldw (vector signed char,
30042 vector signed char,
30044 vector unsigned char vec_sldw (vector unsigned char,
30045 vector unsigned char,
30047 vector signed short vec_sldw (vector signed short,
30048 vector signed short,
30050 vector unsigned short vec_sldw (vector unsigned short,
30051 vector unsigned short,
30053 vector signed int vec_sldw (vector signed int,
30056 vector unsigned int vec_sldw (vector unsigned int,
30057 vector unsigned int,
30059 vector signed long long vec_sldw (vector signed long long,
30060 vector signed long long,
30062 vector unsigned long long vec_sldw (vector unsigned long long,
30063 vector unsigned long long,
30065 * config/rs6000/rs6000-c: Add support for built-in functions
30066 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30067 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30068 * doc/extend.texi: Update the built-in documentation for the
30069 new built-in functions.
30071 2017-05-16 Marek Polacek <polacek@redhat.com>
30075 * tree.c (save_expr): Don't fold the expression.
30077 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30079 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30080 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30081 and (?*y,m). Update insn attributes.
30083 2017-05-16 Martin Liska <mliska@suse.cz>
30085 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30086 flags argument of print_gimple_stmt, print_gimple_expr,
30087 print_generic_stmt and print_generic_expr.
30088 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30089 * coretypes.h: Likewise.
30090 * except.c (dump_eh_tree): Likewise.
30091 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30092 * gimple-pretty-print.h: Likewise.
30093 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30094 (backprop::push_to_worklist): Likewise.
30095 (backprop::pop_from_worklist): Likewise.
30096 (backprop::process_use): Likewise.
30097 (backprop::intersect_uses): Likewise.
30098 (note_replacement): Likewise.
30099 * gimple-ssa-store-merging.c
30100 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30101 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30102 (pass_store_merging::execute): Likewise.
30103 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30104 (ssa_base_cand_dump_callback): Likewise.
30105 (dump_incr_vec): Likewise.
30106 (replace_refs): Likewise.
30107 (replace_mult_candidate): Likewise.
30108 (create_add_on_incoming_edge): Likewise.
30109 (create_phi_basis): Likewise.
30110 (insert_initializers): Likewise.
30111 (all_phi_incrs_profitable): Likewise.
30112 (introduce_cast_before_cand): Likewise.
30113 (replace_one_candidate): Likewise.
30114 * gimplify.c (gimplify_expr): Likewise.
30115 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30116 (set_rename): Likewise.
30117 (rename_uses): Likewise.
30118 (copy_loop_phi_nodes): Likewise.
30119 (add_close_phis_to_merge_points): Likewise.
30120 (copy_loop_close_phi_args): Likewise.
30121 (copy_cond_phi_args): Likewise.
30122 (graphite_copy_stmts_from_block): Likewise.
30123 (translate_pending_phi_nodes): Likewise.
30124 * graphite-poly.c (print_pdr): Likewise.
30125 (dump_gbb_cases): Likewise.
30126 (dump_gbb_conditions): Likewise.
30127 (print_scop_params): Likewise.
30128 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30129 (build_cross_bb_scalars_use): Likewise.
30130 (gather_bbs::before_dom_children): Likewise.
30131 * hsa-dump.c (dump_hsa_immed): Likewise.
30132 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30133 (get_replacement_map): Likewise.
30134 * ipa-inline-analysis.c (dump_condition): Likewise.
30135 (estimate_function_body_sizes): Likewise.
30136 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30137 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30138 * ipa-prop.c (ipa_dump_param): Likewise.
30139 (ipa_print_node_jump_functions_for_edge): Likewise.
30140 (ipa_modify_call_arguments): Likewise.
30141 (ipa_modify_expr): Likewise.
30142 (ipa_dump_param_adjustments): Likewise.
30143 (ipa_dump_agg_replacement_values): Likewise.
30144 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30145 * ipa-pure-const.c (check_stmt): Likewise.
30146 (pass_nothrow::execute): Likewise.
30147 * ipa-split.c (execute_split_functions): Likewise.
30148 * omp-offload.c (dump_oacc_loop_part): Likewise.
30149 (dump_oacc_loop): Likewise.
30150 * trans-mem.c (tm_log_emit): Likewise.
30151 (tm_memopt_accumulate_memops): Likewise.
30152 (dump_tm_memopt_set): Likewise.
30153 (dump_tm_memopt_transform): Likewise.
30154 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30155 (print_loop): Likewise.
30156 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30157 (chkp_gather_checks_info): Likewise.
30158 (chkp_get_check_result): Likewise.
30159 (chkp_remove_check_if_pass): Likewise.
30160 (chkp_use_outer_bounds_if_possible): Likewise.
30161 (chkp_reduce_bounds_lifetime): Likewise.
30162 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30163 (chkp_mark_completed_bounds): Likewise.
30164 (chkp_register_incomplete_bounds): Likewise.
30165 (chkp_mark_invalid_bounds): Likewise.
30166 (chkp_maybe_copy_and_register_bounds): Likewise.
30167 (chkp_build_returned_bound): Likewise.
30168 (chkp_get_bound_for_parm): Likewise.
30169 (chkp_build_bndldx): Likewise.
30170 (chkp_get_bounds_by_definition): Likewise.
30171 (chkp_generate_extern_var_bounds): Likewise.
30172 (chkp_get_bounds_for_decl_addr): Likewise.
30173 * tree-chrec.c (chrec_apply): Likewise.
30174 * tree-data-ref.c (dump_data_reference): Likewise.
30175 (dump_subscript): Likewise.
30176 (dump_data_dependence_relation): Likewise.
30177 (analyze_overlapping_iterations): Likewise.
30178 * tree-inline.c (expand_call_inline): Likewise.
30179 (tree_function_versioning): Likewise.
30180 * tree-into-ssa.c (dump_defs_stack): Likewise.
30181 (dump_currdefs): Likewise.
30182 (dump_names_replaced_by): Likewise.
30183 (dump_update_ssa): Likewise.
30184 (update_ssa): Likewise.
30185 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30186 * tree-parloops.c (build_new_reduction): Likewise.
30187 (try_create_reduction_list): Likewise.
30188 (ref_conflicts_with_region): Likewise.
30189 (oacc_entry_exit_ok_1): Likewise.
30190 (oacc_entry_exit_single_gang): Likewise.
30191 * tree-pretty-print.h: Likewise.
30192 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30193 (get_scalar_evolution): Likewise.
30194 (add_to_evolution): Likewise.
30195 (get_loop_exit_condition): Likewise.
30196 (analyze_evolution_in_loop): Likewise.
30197 (analyze_initial_condition): Likewise.
30198 (analyze_scalar_evolution): Likewise.
30199 (instantiate_scev): Likewise.
30200 (number_of_latch_executions): Likewise.
30201 (gather_chrec_stats): Likewise.
30202 (final_value_replacement_loop): Likewise.
30203 (scev_const_prop): Likewise.
30204 * tree-sra.c (dump_access): Likewise.
30205 (disqualify_candidate): Likewise.
30206 (create_access): Likewise.
30207 (reject): Likewise.
30208 (maybe_add_sra_candidate): Likewise.
30209 (create_access_replacement): Likewise.
30210 (analyze_access_subtree): Likewise.
30211 (analyze_all_variable_accesses): Likewise.
30212 (sra_modify_assign): Likewise.
30213 (initialize_constant_pool_replacements): Likewise.
30214 (find_param_candidates): Likewise.
30215 (decide_one_param_reduction): Likewise.
30216 (replace_removed_params_ssa_names): Likewise.
30217 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30218 * tree-ssa-copy.c (dump_copy_of): Likewise.
30219 (copy_prop_visit_cond_stmt): Likewise.
30220 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30221 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30222 (record_equivalences_from_stmt): Likewise.
30223 * tree-ssa-dse.c (compute_trims): Likewise.
30224 (delete_dead_call): Likewise.
30225 (delete_dead_assignment): Likewise.
30226 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30227 (forward_propagate_into_cond): Likewise.
30228 (pass_forwprop::execute): Likewise.
30229 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30230 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30232 (move_computations_worker): Likewise.
30233 (execute_sm): Likewise.
30234 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30235 (remove_exits_and_undefined_stmts): Likewise.
30236 (remove_redundant_iv_tests): Likewise.
30237 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30238 (adjust_iv_update_pos): Likewise.
30239 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30240 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30241 (value_replacement): Likewise.
30242 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30243 * tree-ssa-pre.c (print_pre_expr): Likewise.
30244 (get_representative_for): Likewise.
30245 (create_expression_by_pieces): Likewise.
30246 (insert_into_preds_of_block): Likewise.
30247 (eliminate_insert): Likewise.
30248 (eliminate_dom_walker::before_dom_children): Likewise.
30249 (eliminate): Likewise.
30250 (remove_dead_inserted_code): Likewise.
30251 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30252 * tree-ssa-reassoc.c (get_rank): Likewise.
30253 (eliminate_duplicate_pair): Likewise.
30254 (eliminate_plus_minus_pair): Likewise.
30255 (eliminate_not_pairs): Likewise.
30256 (undistribute_ops_list): Likewise.
30257 (eliminate_redundant_comparison): Likewise.
30258 (update_range_test): Likewise.
30259 (optimize_range_tests_var_bound): Likewise.
30260 (optimize_vec_cond_expr): Likewise.
30261 (rewrite_expr_tree): Likewise.
30262 (rewrite_expr_tree_parallel): Likewise.
30263 (linearize_expr): Likewise.
30264 (break_up_subtract): Likewise.
30265 (linearize_expr_tree): Likewise.
30266 (attempt_builtin_powi): Likewise.
30267 (attempt_builtin_copysign): Likewise.
30268 (transform_stmt_to_copy): Likewise.
30269 (transform_stmt_to_multiply): Likewise.
30270 (dump_ops_vector): Likewise.
30271 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30272 (print_scc): Likewise.
30273 (set_ssa_val_to): Likewise.
30274 (visit_reference_op_store): Likewise.
30275 (visit_use): Likewise.
30276 (sccvn_dom_walker::before_dom_children): Likewise.
30277 (run_scc_vn): Likewise.
30278 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30280 (expr_hash_elt::print): Likewise.
30281 (const_and_copies::pop_to_marker): Likewise.
30282 (const_and_copies::record_const_or_copy_raw): Likewise.
30283 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30284 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30285 (dump_predicates): Likewise.
30286 (find_uninit_use): Likewise.
30287 (warn_uninitialized_phi): Likewise.
30288 (pass_late_warn_uninitialized::execute): Likewise.
30289 * tree-ssa.c (verify_vssa): Likewise.
30290 (verify_ssa): Likewise.
30291 (maybe_optimize_var): Likewise.
30292 * tree-vrp.c (dump_value_range): Likewise.
30293 (dump_all_value_ranges): Likewise.
30294 (dump_asserts_for): Likewise.
30295 (register_edge_assert_for_2): Likewise.
30296 (vrp_visit_cond_stmt): Likewise.
30297 (vrp_visit_switch_stmt): Likewise.
30298 (vrp_visit_stmt): Likewise.
30299 (vrp_visit_phi_node): Likewise.
30300 (simplify_cond_using_ranges_1): Likewise.
30301 (fold_predicate_in): Likewise.
30302 (evrp_dom_walker::before_dom_children): Likewise.
30303 (evrp_dom_walker::push_value_range): Likewise.
30304 (evrp_dom_walker::pop_value_range): Likewise.
30305 (execute_early_vrp): Likewise.
30307 2017-05-16 Richard Biener <rguenther@suse.de>
30309 * dwarf2out.c (loc_list_from_tree_1): Do not create
30310 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30312 2017-05-16 Richard Biener <rguenther@suse.de>
30314 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30316 (note_variable_value_in_expr): If we resolved the decl ref
30317 do not push to the stack.
30319 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30321 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30322 operations in fast-math mode.
30323 (vaddq_f16): Likewise.
30324 (vmul_f16): Likewise.
30325 (vmulq_f16): Likewise.
30326 (vsub_f16): Likewise.
30327 (vsubq_f16): Likewise.
30328 * config/arm/neon.md (add<mode>3): New.
30330 (fma:<VH:mode>3): New. Also remove outdated comment.
30333 2017-05-16 Martin Liska <mliska@suse.cz>
30337 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30338 (odr_types_equivalent_p): Likewise.
30340 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30342 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30344 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30347 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30348 non-interunit SSE move alternatives with '?'.
30349 (zero-extendsidi peephole2): New peephole to skip intermediate
30350 general register in SSE zero-extend sequence.
30352 2017-05-15 Jeff Law <law@redhat.com>
30354 * reorg.c (relax_delay_slots): Create a new variable to hold
30355 the temporary target rather than clobbering TARGET_LABEL.
30357 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30358 missing argument to extract_bit_field call.
30359 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30361 2017-05-15 Martin Liska <mliska@suse.cz>
30364 * gcc.c (process_command): Do not allow empty argument of -o option.
30366 2017-05-15 Renlin Li <renlin.li@arm.com>
30368 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30369 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30370 * config/aarch64/constraints.md (Usf): Add long call check.
30371 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30372 (call_value): Likewise.
30373 (sibcall): Likewise.
30374 (sibcall_value): Likewise.
30376 (call_value_insn): New.
30377 (sibcall_insn): Update rtx pattern.
30378 (sibcall_value_insn): Likewise.
30379 (call_internal): Remove.
30380 (call_value_internal): Likewise.
30381 (sibcall_internal): Likewise.
30382 (sibcall_value_internal): Likewise.
30383 (call_reg): Likewise.
30384 (call_symbol): Likewise.
30385 (call_value_reg): Likewise.
30386 (call_value_symbol): Likewise.
30388 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30391 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30393 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30395 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30396 compatible with CCGOCmode and with CCZmode.
30398 2017-05-14 Martin Sebor <msebor@redhat.com>
30400 PR middle-end/77671
30401 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30402 (gimple_fold_builtin_snprintf): Same.
30403 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30404 (gimple_fold_builtin_snprintf): Same.
30405 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30406 of character types.
30407 (is_call_safe): New function.
30408 (try_substitute_return_value): Call it.
30409 (try_simplify_call): New function.
30410 (pass_sprintf_length::handle_gimple_call): Call it.
30412 2017-05-14 Martin Sebor <msebor@redhat.com>
30414 PR middle-end/80669
30415 * builtins.c (expand_builtin_stpncpy): Simplify.
30417 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30419 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30420 * config/i386/i386.h
30421 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30422 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30423 (struct machine_function): Add new members call_ms2sysv,
30424 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30425 (struct machine_frame_state): New fields sp_realigned and
30426 sp_realigned_offset.
30427 * config/i386/i386.c
30428 (enum xlogue_stub): New enum.
30429 (enum xlogue_stub_sets): New enum.
30430 (class xlogue_layout): New class.
30431 (struct ix86_frame): New fields stack_realign_allocate_offset,
30432 stack_realign_offset and outlined_save_offset. Modify comments to
30433 detail stack layout when using out-of-line stubs.
30434 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30435 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30436 -mcall-ms2sysv-xlogues.
30437 (stub_managed_regs): New static variable.
30438 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30439 registers managed by out-of-line stub.
30440 (disable_call_ms2sysv_xlogues): New function.
30441 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30442 m->call_ms2sysv when appropriate and compute frame layout for
30444 (sp_valid_at, fp_valid_at): New inline functions.
30445 (choose_basereg): New function.
30446 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30448 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30449 Use align parameter of choose_baseaddr to generated aligned SSE movs
30451 (pro_epilogue_adjust_stack): Modify to track
30452 machine_frame_state::sp_realigned.
30453 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30454 (ix86_nsaved_sseregs): Likewise.
30455 (ix86_emit_save_regs): Likewise.
30456 (ix86_emit_save_regs_using_mov): Likewise.
30457 (ix86_emit_save_sse_regs_using_mov): Likewise.
30458 (get_scratch_register_on_entry): Likewise.
30459 (gen_frame_set): New function.
30460 (gen_frame_load): Likewise.
30461 (gen_frame_store): Likewise.
30462 (emit_outlined_ms2sysv_save): Likewise.
30463 (emit_outlined_ms2sysv_restore): Likewise.
30464 (ix86_expand_prologue): Modify stack re-alignment code and call
30465 emit_outlined_ms2sysv_save when appropriate.
30466 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30467 parameter rtx_insn *insn, which allows the function to be used to only
30468 generate the notes.
30469 (ix86_expand_epilogue): Modify validity checks of frame and stack
30470 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30471 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30472 * config/i386/predicates.md
30473 (save_multiple): New predicate.
30474 (restore_multiple): Likewise.
30475 * config/i386/sse.md
30476 (save_multiple<mode>): New pattern.
30477 (save_multiple_realign<mode>): Likewise.
30478 (restore_multiple<mode>): Likewise.
30479 (restore_multiple_and_return<mode>): Likewise.
30480 (restore_multiple_leave_return<mode>): Likewise.
30481 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30483 2017-05-14 Julia Koval <julia.koval@intel.com>
30485 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30486 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30487 (__builtin_ia32_xsetbv): New builtins.
30488 * config/i386/i386.c (ix86_expand_special_args_builtin):
30490 (ix86_expand_builtin): Special expand for new intrinsics.
30491 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30492 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30493 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30495 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30497 * cfganal.c (inverted_post_order_compute): Change argument type
30499 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30500 * df-core.c (rest_of_handle_df_initialize): Adjust.
30501 (rest_of_handle_df_finish): Likewise.
30502 (df_analyze_1): Likewise.
30503 (df_analyze): Likewise.
30504 (loop_inverted_post_order_compute): Change argument to be a vec *.
30505 (df_analyze_loop): Adjust.
30506 (df_get_n_blocks): Likewise.
30507 (df_get_postorder): Likewise.
30508 * df.h (struct df_d): Change field to be a vec.
30509 * lcm.c (compute_laterin): Adjust.
30510 (compute_available): Likewise.
30511 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30512 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30513 * tree-ssa-pre.c (compute_antic): Likewise.
30515 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30517 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30518 (depth_first_search::depth_first_search): Change structure init
30519 function to this constructor.
30520 (depth_first_search::add_bb): Rename function to this member.
30521 (depth_first_search::execute): Likewise.
30522 (flow_dfs_compute_reverse_finish): Adjust.
30524 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30526 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30527 (longest_simple_path): Likewise.
30528 * shrink-wrap.c (spread_components): Likewise.
30529 (disqualify_problematic_components): Likewise.
30530 (emit_common_heads_for_components): Likewise.
30531 (emit_common_tails_for_components): Likewise.
30532 (insert_prologue_epilogue_for_components): Likewise.
30534 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30536 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30539 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30541 * df-core.c (df_set_blocks): Start using auto_bitmap.
30542 (df_compact_blocks): Likewise.
30543 * df-problems.c (df_rd_confluence_n): Likewise.
30544 * df-scan.c (df_insn_rescan_all): Likewise.
30545 (df_process_deferred_rescans): Likewise.
30546 (df_update_entry_block_defs): Likewise.
30547 (df_update_exit_block_uses): Likewise.
30548 (df_entry_block_bitmap_verify): Likewise.
30549 (df_exit_block_bitmap_verify): Likewise.
30550 (df_scan_verify): Likewise.
30551 * lra-constraints.c (lra_constraints): Likewise.
30552 (undo_optional_reloads): Likewise.
30553 (lra_undo_inheritance): Likewise.
30554 * lra-remat.c (calculate_gen_cands): Likewise.
30555 (do_remat): Likewise.
30556 * lra-spills.c (assign_spill_hard_regs): Likewise.
30557 (spill_pseudos): Likewise.
30558 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30559 (bitmap_set_subtract_values): Likewise.
30561 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30563 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30564 management with auto_bitmap.
30565 (fix_inter_tick): Likewise.
30566 (fix_recovery_deps): Likewise.
30567 * ira.c (add_store_equivs): Likewise.
30568 (find_moveable_pseudos): Likewise.
30569 (split_live_ranges_for_shrink_wrap): Likewise.
30570 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30571 (rtx_reuse_manager::seen_def_p): Likewise.
30572 (rtx_reuse_manager::set_seen_def): Likewise.
30573 * print-rtl.h (class rtx_reuse_manager): Likewise.
30575 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30577 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30579 (migrate_btr_def): Likewise.
30580 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30581 * df-core.c (loop_post_order_compute): Likewise.
30582 (loop_inverted_post_order_compute): Likewise.
30583 * hsa-common.h: Likewise.
30584 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30585 * init-regs.c (initialize_uninitialized_regs): Likewise.
30586 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30587 (inline_small_functions): Likewise.
30588 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30589 * ira.c (combine_and_move_insns): Likewise.
30590 (build_insn_chain): Likewise.
30591 * loop-invariant.c (find_invariants): Likewise.
30592 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30593 * predict.c (tree_predict_by_opcode): Likewise.
30594 (predict_paths_leading_to): Likewise.
30595 (predict_paths_leading_to_edge): Likewise.
30596 (estimate_loops_at_level): Likewise.
30597 (estimate_loops): Likewise.
30598 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30599 (spread_components): Likewise.
30600 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30601 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30602 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30603 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30604 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30605 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30606 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30607 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30608 (mark_threaded_blocks): Likewise.
30609 (thread_through_all_blocks): Likewise.
30610 * tree-ssa.c (verify_ssa): Likewise.
30611 (execute_update_addresses_taken): Likewise.
30612 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30614 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30616 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30618 (post_order_compute): Likewise.
30619 (inverted_post_order_compute): Likewise.
30620 (pre_and_rev_post_order_compute_fn): Likewise.
30622 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30624 * genrecog.c (int_set::int_set): Explicitly construct our
30625 auto_vec base class.
30626 * vec.h (auto_vec::auto_vec): New constructor.
30628 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30630 * bitmap.h (class auto_bitmap): New constructor taking
30631 bitmap_obstack * argument.
30633 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30635 * bitmap.h (class auto_bitmap): Change type of m_bits to
30636 bitmap_head, and adjust ctor / dtor and member operators.
30638 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30640 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30641 when returned register mode doesn't match original mode.
30643 2017-05-12 Jeff Law <law@redhat.com>
30644 Jakub Jelinek <jakub@redhat.com>
30646 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30647 we look for cc setter after the compare-elim changes.
30648 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30649 within the vector to match what compare-elim now expects.
30650 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30651 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30653 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30654 after the compare-elim changes.
30655 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30656 the vector to match what compare-elim now expects.
30657 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30658 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30659 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30660 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30661 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30663 * config/visium/visium.c (single_set_and_flags): Fix where
30664 we look for cc setter after the compare-elim changes.
30665 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30666 with the vector to match what compare-elim now expects.
30667 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30668 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30669 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30670 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30671 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30672 (neg<mode>2_insn_set_overflow): Likewise.
30674 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30676 PR middle-end/79794
30677 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30678 maybe_expand_insn call, set ops[0].target. If still set after call,
30679 set alt_rtl. Add extra arg to recursive calls.
30680 (extract_bit_field): Add alt_rtl argument. Pass to
30682 * expmed.h (extract_bit_field): Fix prototype.
30683 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30684 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30685 to extract_bit_field_calls.
30686 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30687 Pass alt_rtl to extract_bit_field calls.
30688 * calls.c (store_unaligned_arguments_into_psuedos)
30689 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30690 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30692 * optabs.h (struct expand_operand): Add target bitfield.
30694 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30696 * compare-elim.c (try_eliminate_compare): Canonicalize
30697 operation with embedded compare to
30698 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30699 (set (reg) (operation)].
30701 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30703 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30706 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30707 cost of adding a carry flag for ADC instruction.
30708 [case MINUS]: Ignore the cost of subtracting a carry flag
30709 for SBB instruction.
30711 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30713 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30715 * config/rs6000/bmiintrin.h: New file.
30716 * config/rs6000/bmi2intrin.h: New file.
30717 * config/rs6000/x86intrin.h: New file.
30719 2017-05-12 Jeff Law <law@redhat.com>
30721 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30724 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30726 PR middle-end/80707
30727 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30729 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30731 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30732 early expansion of vector divide builtins.
30733 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30734 builtins identified as having unsigned arguments.
30736 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30738 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30739 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30740 expansion of vector logical operations (and, andc, or, xor,
30743 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30745 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30746 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30748 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30750 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30751 early GIMPLE expansion of vector multiplies.
30753 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30755 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30756 TARGET_HAVE_MOVT conditional.
30757 (movt splitter): Likewise.
30759 2017-05-12 Richard Biener <rguenther@suse.de>
30761 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30762 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30763 Fold all stmts not inplace.
30765 2017-05-12 Richard Biener <rguenther@suse.de>
30767 PR tree-optimization/80713
30768 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30769 inserted_exprs bit for not removed stmts.
30771 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30773 PR middle-end/69921
30774 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30775 parallelized" attribute for parallelized OpenACC kernels.
30776 * omp-offload.c (execute_oacc_device_lower): Use it.
30778 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30779 Set "oacc kernels" attribute.
30780 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30781 parameter. Adjust all users.
30782 (oacc_fn_attrib_kernels_p): Remove function.
30783 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30784 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30785 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30786 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30787 assert "oacc kernels" attribute is set.
30789 2017-05-11 Carl Love <cel@us.ibm.com>
30791 * config/rs6000/rs6000-c: Add support for built-in functions
30792 vector unsigned char vec_popcnt (vector signed char)
30793 vector unsigned char vec_popcnt (vector unsigned char)
30794 vector unsigned short vec_popcnt (vector signed short)
30795 vector unsigned short vec_popcnt (vector unsigned short)
30796 vector unsigned int vec_popcnt (vector signed int)
30797 vector unsigned int vec_popcnt (vector unsigned int)
30798 vector unsigned long long vec_popcnt (vector signed long long)
30799 vector unsigned long long vec_popcnt (vector unsigned long long)
30800 vector signed long long vec_slo (vector signed long long,
30801 vector signed char)
30802 vector signed long long vec_slo (vector signed long long,
30803 vector unsigned char)
30804 vector unsigned long long vec_slo (vector unsigned long long,
30805 vector signed char)
30806 vector unsigned long long vec_slo (vector unsigned long long,
30807 vector unsigned char)
30808 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30809 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30810 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30811 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30812 * doc/extend.texi: Update the built-in documentation file for the
30813 new built-in functions.
30815 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30817 * attribs.h (sorted_attr_string): Move machine independent
30818 functions for target clone support from the i386 port to common
30819 code. Rename ix86_function_versions to common_function_versions.
30820 Rename make_name to make_unique_name.
30821 (common_function_versions): Likewise.
30822 (make_unique_name): Likewise.
30823 (make_dispatcher_decl): Likewise.
30824 (is_function_default_version): Likewise.
30825 * attribs.c (attr_strcmp): Likewise.
30826 (sorted_attr_string): Likewise.
30827 (common_function_versions): Likewise.
30828 (make_unique_name): Likewise.
30829 (make_dispatcher_decl): Likewise.
30830 (is_function_default_version): Likewise.
30831 * config/i386/i386.c (attr_strcmp): Likewise.
30832 (sorted_attr_string): Likewise.
30833 (ix86_function_versions): Likewise.
30834 (make_name): Likewise.
30835 (make_dispatcher_decl): Likewise.
30836 (is_function_default_version): Likewise.
30837 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30839 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30842 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30843 Account for direct move costs for vec_construct of integer
30846 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30849 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30850 (UNSPEC_STX_ATOMIC): Ditto.
30851 (loaddi_via_sse): New insn.
30852 (storedi_via_sse): Ditto.
30853 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30854 Update corresponding peephole2 patterns.
30855 (atomic_storedi_fpu): Ditto.
30857 2017-05-11 Julia Koval <julia.koval@intel.com>
30859 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30860 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30862 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30863 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30864 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30866 2017-05-11 Nathan Sidwell <nathan@acm.org>
30868 * graphite-poly.c: Include dumpfile.h.
30870 * dumpfle.h (dump_function): Declare here ...
30871 * tree-dump.h (dump_function): ... not here.
30872 * dumpfile.c: #include tree-cfg.h.
30873 (dump_function): Move here from ...
30874 * tree-dump.c (dump_function): ... here.
30875 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30876 * graphite-poly.c: Don't include tree-dump.h.
30877 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30878 * print-tree.c: Likewise.
30879 * stor-layout.c: Likewise.
30880 * tree-nested.c: Likewise.
30882 * dumpfile.c (dump_start): Use TDF_FLAGS.
30883 (dump_enable_all): Fix TDF_KIND check thinko.
30885 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30887 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30888 array entries to represent two legal parameterizations of the
30889 overloaded __builtin_cmpb function, as represented by the
30890 P6_OV_BUILTIN_CMPB constant.
30891 (altivec_resolve_overloaded_builtin): Add special case handling
30892 for the __builtin_cmpb function, as represented by the
30893 P6_OV_BUILTIN_CMPB constant.
30894 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30895 (BU_P6_64BIT_2): New macro.
30896 (BU_P6_OVERLOAD_2): New macro
30897 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30898 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30899 (CMPB): Add overload support to represent both 32-bit and 64-bit
30900 compare-bytes function.
30901 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30902 support for TARGET_CMPB.
30903 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30904 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30905 documentation of the __builtin_cmpb overloaded built-in function.
30907 2017-05-11 Richard Biener <rguenther@suse.de>
30909 PR tree-optimization/80705
30910 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30911 bases are not vectorizable.
30913 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30915 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30916 when counting register pressure.
30918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30920 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30921 (iv_ca_more_deps): Renamed to ...
30922 (iv_ca_compare_deps): ... this.
30923 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30925 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30927 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30929 (determine_group_iv_costs): ... here.
30930 (find_inv_vars_cb): Record inv var if it's not recorded before.
30932 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30934 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30935 (get_shiftadd_cost): Ditto.
30937 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30939 * tree-ssa-address.c: Include header file.
30940 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30942 (add_to_parts): Refactor.
30943 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30944 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30947 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30949 PR tree-optimization/53090
30950 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30952 (extract_cond_operands): Detect condition with IV on both sides
30953 and return COMP_IV_EXPR_2.
30954 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30955 (rewrite_use_compare): Simplify by removing call to function
30956 extract_cond_operands.
30958 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30960 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30961 (extract_cond_operands): Detect condition comparing against non-
30962 invariant bound and return appropriate enum value.
30963 (find_interesting_uses_cond): Update use of extract_cond_operands.
30964 Handle its return value accordingly.
30965 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30967 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30969 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30970 nonlinear iv_use computation in loop invariant sensitive way.
30972 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30974 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30975 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30977 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30979 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30980 (dump_cand): Support iv_cand.inv_exprs.
30981 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30983 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30986 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30988 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30990 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30991 as local function. Include necessary header files.
30992 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30994 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30996 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30998 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31000 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31001 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31002 RSHIFT_EXPR and BIT_NOT_EXPR.
31004 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31006 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31007 (adjust_setup_cost): New parameter supporting round up adjustment.
31008 (struct address_cost_data): Delete.
31009 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31010 (split_address_cost, ptr_difference_cost): Delete.
31011 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31012 (struct ainc_cost_data): New struct.
31013 (get_address_cost_ainc): New function.
31014 (get_address_cost, get_computation_cost): Reimplement.
31015 (determine_group_iv_cost_address): Record inv_expr for all uses of
31017 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31018 (iv_ca_has_deps): Reimplemented to ...
31019 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31021 (iv_ca_extend): Call iv_ca_more_deps.
31023 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31025 * tree-ssa-address.c (struct mem_address): Move to header file.
31026 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31027 * tree-ssa-address.h (struct mem_address): Move from C file.
31028 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31030 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31032 * tree-affine.h (aff_combination_type): New interface.
31033 (aff_combination_zero_p): Remove static.
31034 (aff_combination_const_p): New interface.
31035 (aff_combination_singleton_var_p): New interfaces.
31037 2017-05-11 Richard Biener <rguenther@suse.de>
31039 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31040 Skip unreachable blocks and destinations.
31041 (eliminate): Move stmt removal and fixup ...
31042 (fini_eliminate): ... here. Skip inserted exprs.
31043 (pass_pre::execute): Move fini_pre after fini_eliminate.
31044 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31045 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31046 PRE to get rid of dead code that has invalid SSA form and
31047 split critical edges again.
31049 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31051 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31053 2017-05-11 Richard Biener <rguenther@suse.de>
31055 * passes.c (execute_function_todo): Verify loops if they are
31056 said to be up-to-date.
31057 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31058 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31060 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31063 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31064 handle calling assemble_external ourself.
31067 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31068 modes with zero size. Enhance comment.
31070 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31072 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31073 built-ins for vec_xl and vec_xst with short and char pointer
31076 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31078 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31079 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31080 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31081 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31082 (_mm_maskz_min_round_ss): New intrinsics.
31083 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31084 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31085 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31086 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31087 (__builtin_ia32_minss_mask_round): New builtins.
31088 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31089 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31090 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31092 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31093 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31095 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31098 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31100 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31101 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31102 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31103 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31104 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31105 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31106 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31107 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31108 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31109 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31110 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31111 (__builtin_ia32_mulss_mask_round): New builtins.
31112 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31113 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31114 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31116 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31117 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31119 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31122 2017-05-10 Julia Koval <julia.koval@intel.com>
31124 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31125 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31126 (_mm256_setr_m128i): New intrinsics.
31128 2017-05-10 Julia Koval <julia.koval@intel.com>
31130 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31131 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31132 (_mm_maskz_rcp14_ss): New intrinsics.
31133 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31134 (__builtin_ia32_rcp14ss_mask): New builtins.
31135 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31137 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31139 PR tree-optimization/51513
31140 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31141 (assert_unreachable_fallthru_edge_p): Use it.
31142 (group_case_labels_stmt): Likewise.
31143 * tree-cfg.h: Prototype it.
31144 * stmt.c: Include cfghooks.h and tree-cfg.h.
31145 (emit_case_dispatch_table) <gap_label>: New local variable.
31146 Use it to fill dispatch table gaps.
31147 Test for default_label before updating probabilities.
31148 (expand_case) <default_label>: Remove unneeded initialization.
31149 Test for unreachable default case statement and remove its edge.
31150 Set default_label accordingly.
31151 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31153 2017-05-10 Carl Love <cel@us.ibm.com>
31155 * config/rs6000/rs6000-c: Add support for built-in functions
31156 vector signed char vec_neg (vector signed char)
31157 vector signed short int vec_neg (vector short int)
31158 vector signed int vec_neg (vector signed int)
31159 vector signed long long vec_neg (vector signed long long)
31160 vector float vec_neg (vector float)
31161 vector double vec_neg (vector double)
31162 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31164 * config/rs6000/altivec.h: Add define for vec_neg
31165 * doc/extend.texi: Update the built-in documentation for the
31166 new built-in functions.
31168 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31170 PR tree-optimization/77644
31171 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31173 2017-05-10 Nathan Sidwell <nathan@acm.org>
31175 * dumpfile.h (TDI_lang_all): New.
31176 (TDF_KIND): New. Renumber others
31177 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31179 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31181 (get_dump_file_name): Adjust suffix generation.
31182 (dump_enable_all): Use TDF_KIND.
31183 * doc/invoke.texi (-fdump-lang-all): Document.
31185 * dumpfile.h: Tabify.
31187 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31190 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31191 Move member access before delete.
31193 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31195 * tree-inline.c (expand_call_inline): Split block at stmt
31198 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31201 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31202 are now unused after splitting mov{sf,sd}_hardfloat.
31203 (f32_lr2): Likewise.
31204 (f32_lm): Likewise.
31205 (f32_lm2): Likewise.
31206 (f32_li): Likewise.
31207 (f32_li2): Likewise.
31208 (f32_lv): Likewise.
31209 (f32_sr): Likewise.
31210 (f32_sr2): Likewise.
31211 (f32_sm): Likewise.
31212 (f32_sm2): Likewise.
31213 (f32_si): Likewise.
31214 (f32_si2): Likewise.
31215 (f32_sv): Likewise.
31216 (f32_dm): Likewise.
31217 (f32_vsx): Likewise.
31218 (f32_av): Likewise.
31219 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31220 For movsf, order stores so the VSX stores occur before the GPR
31221 store which encourages the register allocator to use a traditional
31222 FPR instead of a GPR. For movsd, order the stores so that the GPR
31223 store comes before the VSX stores to allow the power6 to work.
31224 This is due to the power6 not having a 32-bit integer store
31225 instruction from a FPR.
31226 (movsf_hardfloat): Likewise.
31227 (movsd_hardfloat): Likewise.
31229 2017-05-09 Martin Sebor <msebor@redhat.com>
31231 PR translation/80280
31232 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31235 PR translation/80280
31236 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31237 data member added in r247778.
31238 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31240 2017-05-09 Nathan Sidwell <nathan@acm.org>
31242 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31244 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31247 2017-05-09 Marek Polacek <polacek@redhat.com>
31249 * doc/invoke.texi: Fix typo.
31251 2017-05-09 Richard Biener <rguenther@suse.de>
31253 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31254 (vrp_val_is_min): Likewise.
31255 (set_value_range_to_value): Likewise.
31256 (set_value_range_to_nonnegative): Likewise.
31257 (gimple_assign_nonzero_p): Likewise.
31258 (gimple_stmt_nonzero_p): Likewise.
31259 (vrp_int_const_binop): Likewise. Remove unreachable case.
31260 (adjust_range_with_scev): Adjust comments.
31261 (compare_range_with_value): Likewise.
31262 (extract_range_from_phi_node): Likewise.
31263 (test_for_singularity): Likewise.
31265 2017-05-09 Richard Biener <rguenther@suse.de>
31267 * tree-vrp.c (get_single_symbol): Add assert that we don't
31268 get overflowed constants as invariant part.
31269 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31270 checks. Use wi::cmp instead of recursing for integer constants.
31271 (compare_values): Just ignore whether we assumed undefined
31272 overflow instead of failing the compare.
31273 (extract_range_for_var_from_comparison_expr): Add comment before the
31274 TREE_NO_WARNING sets.
31275 (test_for_singularity): Likewise.
31276 (extract_range_from_comparison): Do not disable optimization
31277 when we assumed undefined overflow.
31278 (extract_range_basic): Remove init of unused var.
31280 2017-05-09 Richard Biener <rguenther@suse.de>
31282 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31283 (extract_range_from_multiplicative_op_1): Adjust.
31284 (extract_range_from_binary_expr_1): Use int_const_binop.
31286 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31289 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31290 rs6000_store_data_bypass_p in seven define_bypass directives and
31291 in several comments.
31292 * config/rs6000/rs6000-protos.h: Add prototype for
31293 rs6000_store_data_bypass_p function.
31294 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31295 function implements slightly different (rs6000-specific) semantics
31296 than store_data_bypass_p, returning false rather than aborting
31297 with assertion error when arguments do not satisfy the
31298 requirements of store data bypass.
31299 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31300 rs6000_store_data_bypass_p.
31302 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31304 * config/xtensa/xtensa-protos.h
31305 (xtensa_initial_elimination_offset): New declaration.
31306 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31307 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31308 macro definition, add case for FRAME_POINTER_REGNUM when
31309 FRAME_GROWS_DOWNWARD.
31310 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31311 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31312 xtensa_initial_elimination_offset.
31314 2017-05-08 Nathan Sidwell <nathan@acm.org>
31316 * doc/invoke.texi: Alphabetize -fdump options.
31318 2017-05-08 Martin Sebor <msebor@redhat.com>
31320 PR translation/80280
31321 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31323 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31325 * target.def (compute_frame_layout): New optional target hook.
31326 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31327 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31328 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31330 * reload1.c (verify_initial_elim_offsets): Likewise.
31331 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31332 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31333 (arm_get_frame_offsets): Split up into this ...
31334 (arm_compute_frame_layout): ... and this function.
31336 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31338 * config/aarch64/constraints.md (Usa): New constraint.
31339 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31341 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31343 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31344 with_multilib_list after it has been checked.
31346 2017-05-08 Richard Biener <rguenther@suse.de>
31348 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31349 (bitmap_set_subtract_values): Likewise.
31351 2017-05-08 Richard Biener <rguenther@suse.de>
31353 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31354 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31356 (gimple_stmt_nonzero_warnv_p): Rename to ...
31357 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31359 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31360 (extract_range_basic): Adjust, do not disable propagation on
31361 strict overflow sensitive simplification.
31362 (vrp_visit_cond_stmt): Likewise.
31364 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31366 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31367 body size unconditionally.
31369 2017-05-07 Jeff Law <law@redhat.com>
31372 2017-05-06 Jeff Law <law@redhat.com>
31373 PR tree-optimization/78496
31374 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31377 PR tree-optimization/78496
31378 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31379 (simplify_stmt_using_ranges): Call it.
31380 (vrp_dom_walker::before_dom_children): Extract equivalences
31381 from an ASSERT_EXPR with an equality comparison against a
31384 2017-05-06 Jeff Law <law@redhat.com>
31386 PR tree-optimization/78496
31387 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31390 PR tree-optimization/78496
31391 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31392 (simplify_stmt_using_ranges): Call it.
31393 (vrp_dom_walker::before_dom_children): Extract equivalences
31394 from an ASSERT_EXPR with an equality comparison against a
31397 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31399 * lra-constraints.c (lra_copy_reg_equiv): New function.
31400 (split_reg): Use it to copy equivalence information from the
31401 original register to the spill register.
31403 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31405 PR rtl-optimization/75964
31406 * simplify-rtx.c (simplify_const_relational_operation): Remove
31407 invalid handling of comparisons of integer ABS.
31409 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31411 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31412 initialize to zero.
31413 (init_regs): Remove declaration.
31414 (function_arg_advance_32): Initialize error_p as boolean variable.
31416 2017-05-05 Nathan Sidwell <nathan@acm.org>
31418 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31419 lines. Use for (;;).
31421 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31423 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31424 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31425 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31426 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31427 VF=2 that require versioning.
31429 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31431 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31434 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31436 * diagnostic.h (diagnostic_override_option_index): Convert from
31437 macro to inline function.
31439 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31441 * diagnostic.c (last_module_changed_p): New function.
31442 (set_last_module): New function.
31443 (diagnostic_report_current_module): Convert macro usage to
31444 the above functions.
31445 * diagnostic.h (diagnostic_context::last_module): Strengthen
31446 from const line_map * to const line_map_ordinary *.
31447 (diagnostic_last_module_changed): Delete macro.
31448 (diagnostic_set_last_module): Delete macro.
31450 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31452 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31453 with diagnostic_report_diagnostic.
31454 (diagnostic_n_impl_richloc): Likewise.
31455 * diagnostic.h (report_diagnostic): Delete macro.
31456 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31457 with diagnostic_report_diagnostic.
31458 * substring-locations.c (format_warning_va): Likewise.
31460 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31462 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31463 save/restor of format_spec. Move option-printing code to...
31464 (print_option_information): ...this new function, and
31465 reimplement by simply printing to the pretty_printer,
31466 rather than appending to the format string.
31468 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31470 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31471 handling logic into...
31472 (update_effective_level_from_pragmas): ...this new function.
31474 2017-05-04 Andrew Waterman <andrew@sifive.com>
31476 * config/riscv/riscv.opt (mstrict-align): New option.
31477 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31478 (SLOW_UNALIGNED_ACCESS): Define.
31479 (riscv_slow_unaligned_access): Declare.
31480 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31482 (riscv_slow_unaligned_access): New variable.
31483 (rocket_tune_info): Set slow_unaligned_access to true.
31484 (optimize_size_tune_info): Set slow_unaligned_access to false.
31485 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31486 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31487 (riscv_option_override): Set riscv_slow_unaligned_access.
31488 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31490 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31492 * config/riscv/riscv.md: Unify indentation.
31494 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31499 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31501 (extendsi<mode>2): Add support for doing sign extension via
31502 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31503 don't have ISA 3.0 instructions.
31504 (extendsi<mode>2 splitter): Likewise.
31505 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31506 generate the normal insns since SImode can now go in vector
31507 registers. Disallow the special UNSPECs needed for previous
31508 machines to hide SImode being used. Add new insns
31509 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31510 (fix_trunc<mode>si2_stfiwx): Likewise.
31511 (fix_trunc<mode>si2_internal): Likewise.
31512 (fixuns_trunc<mode>si2): Likewise.
31513 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31514 (fctiw<u>z_<mode>_smallint): Likewise.
31515 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31516 of floating point to 32-bit integer from doing a direct move to
31517 the GPR registers to do a store.
31518 (fctiwz_<mode>): Break long line.
31520 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31522 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31523 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31524 (addr_list, addr_offset_valid_p): New.
31525 (split_address_groups): Check offset validity with above function.
31526 (gt-tree-ssa-loop-ivopts.h): Include header file.
31528 2017-05-05 Nathan Sidwell <nathan@acm.org>
31530 * config.gcc (arm*-*-*): Add missing 'fi'.
31532 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31534 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31535 included in -fopt-info does not matter.
31536 * doc/optinfo.texi (-fopt-info): Fix description of default
31537 behavour. Explicitly say order of options included in -fopt-info
31540 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31542 * config.gcc: Allow combinations of aprofile and rmprofile values for
31543 --with-multilib-list.
31544 * config/arm/t-multilib: New file.
31545 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31546 variables. Remove setting of ISA and floating-point ABI in
31547 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31548 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31549 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31550 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31552 * config/arm/t-rmprofile: Likewise except for the matches changes.
31553 * doc/install.texi (--with-multilib-list): Document the combination of
31554 aprofile and rmprofile values and warn about pitfalls in doing that.
31556 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31558 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31559 (movdi_aarch64): Likewise.
31561 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31563 PR tree-optimization/80632
31564 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31566 (build_arrays): Initialize it for virtual phis.
31567 (fix_phi_nodes): Use it for virtual phis.
31569 PR tree-optimization/80558
31570 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31571 [x, y] op z into [x op, y op z] for op & or | if conditions
31574 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31575 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31578 * config/arm/arm.md (use_literal_pool): Remove.
31579 (64-bit immediate split): No longer takes cost into consideration
31580 if arm_disable_literal_pool is enabled.
31581 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31582 used when arm_disable_literal_pool is enabled.
31583 (arm_max_const_double_inline_cost): Remove use of
31584 arm_disable_literal_pool.
31585 (push_minipool_fix): Add assert.
31586 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31587 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31588 (no_literal_pool_sf_immediate): New.
31590 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31592 PR tree-optimization/80613
31593 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31594 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31596 2017-05-05 Richard Biener <rguenther@suse.de>
31598 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31600 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31602 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31603 of this flag from insn conditions due to removal from r247495.
31605 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31607 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31609 (arm_early_store_addr_dep_ptr): Likewise.
31610 * config/arm/aarch-common-protos.h
31611 (arm_early_load_addr_dep_ptr): Add prototype.
31612 (arm_early_store_addr_dep_ptr): Likewise.
31613 * config/arm/cortex-a53.md: Add new bypasses.
31615 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31617 * tree.c (next_type_uid): Change type to unsigned.
31618 (type_hash_canon): Decrement back next_type_uid if
31619 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31620 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31623 2017-05-04 Martin Sebor <msebor@redhat.com>
31625 * builtins.c: Fix a trivial typo in a comment.
31627 PR middle-end/79234
31628 * builtins.c (check_sizes): Adjust to handle reading past the end.
31629 Avoid printing excessive upper bound of ranges. Use %E to print
31630 tree nodes instead of converting them to %wu.
31631 (expand_builtin_memchr): New function.
31632 (compute_dest_size): Rename...
31633 (compute_objsize): ...to this.
31634 (expand_builtin_memcpy): Adjust.
31635 (expand_builtin_mempcpy): Adjust.
31636 (expand_builtin_strcat): Adjust.
31637 (expand_builtin_strcpy): Adjust.
31638 (check_strncat_sizes): Adjust.
31639 (expand_builtin_strncat): Adjust.
31640 (expand_builtin_strncpy): Adjust and simplify.
31641 (expand_builtin_memset): Adjust.
31642 (expand_builtin_bzero): Adjust.
31643 (expand_builtin_memcmp): Adjust.
31644 (expand_builtin): Handle memcmp.
31645 (maybe_emit_chk_warning): Check strncat just once.
31647 2017-05-04 Martin Sebor <msebor@redhat.com>
31649 PR preprocessor/79214
31650 PR middle-end/79222
31651 PR middle-end/79223
31652 * builtins.c (check_sizes): Add inlining context and issue
31653 warnings even when -Wno-system-headers is set.
31654 (check_strncat_sizes): Same.
31655 (expand_builtin_strncat): Same.
31656 (expand_builtin_memmove): New function.
31657 (expand_builtin_stpncpy): Same.
31658 (expand_builtin): Handle memmove and stpncpy.
31660 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31662 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31663 which is not used any more.
31665 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31667 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31669 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31671 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31672 (cortexa53_tunings): Likewise.
31673 (cortexa57_tunings): Likewise.
31674 (cortexa72_tunings): Likewise.
31675 (cortexa73_tunings): Likewise.
31677 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31679 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31680 Set loop alignment to 8.
31682 2017-05-04 Martin Sebor <msebor@redhat.com>
31684 PR translation/80280
31685 * builtins.c (expand_builtin_object_size): Add missing quoting to
31686 %D and like directives.
31687 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31688 (hsa_type_for_tree_type): Same.
31689 (verify_function_arguments): Same.
31690 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31691 * varasm.c (get_section): Same.
31694 2017-05-04 Martin Sebor <msebor@redhat.com>
31696 PR translation/80280
31697 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31699 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31701 * config/aarch64/aarch64.c (generic_addrcost_table):
31702 Change HI/TI mode setting.
31704 2017-05-04 Martin Jambor <mjambor@suse.cz>
31706 PR tree-optimization/80622
31707 * tree-sra.c (comes_initialized_p): New function.
31708 (build_accesses_from_assign): Only set write lazily when
31709 comes_initialized_p is false.
31710 (analyze_access_subtree): Use comes_initialized_p.
31711 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31712 instead of testing for PARM_DECL.
31714 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31716 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31717 constraint on operand 0 to allow more general addressing modes.
31718 Adjust output template.
31719 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31721 * config/aarch64/aarch64-protos.h
31722 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31723 * config/aarch64/constraints.md (Dp): New address constraint.
31724 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31727 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31729 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31730 update use of estimate_ipcp_clone_size_and_time.
31731 (estimate_local_effects): Update use of
31732 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31733 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31734 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31735 Return nonspecialized time.
31737 2017-05-04 Richard Biener <rguenther@suse.de>
31739 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31740 for the last VUSE which def dominates the PHI. Directly call
31742 (get_continuation_for_phi_1): Remove.
31744 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31746 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31747 to explain the use of truncating division. Cap the number of
31748 iterations to the maximum given by nb_iterations_upper_bound,
31751 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31753 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31754 * configure: Regenerate.
31755 * config.in: Regenerate.
31756 * config/i386/driver-mingw32.c: new file.
31757 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31758 * config.host: Link driver-mingw32.o on MinGW host.
31759 * doc/install.texi: Document new --enable-mingw-wildcard configure
31762 2017-05-04 Marek Polacek <polacek@redhat.com>
31764 PR tree-optimization/80612
31765 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31767 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31768 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31770 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31771 (movt splitter): Likewise.
31772 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31773 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31774 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31775 block for Thumb-1 with MOVT.
31776 (thumb2_legitimate_address_p): Move code block ...
31777 (can_avoid_literal_pool_for_label_p): ... into this new function.
31778 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31780 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31781 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31782 "M-profile targets with the MOVT instruction".
31784 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31786 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31787 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31788 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31790 2017-05-04 Martin Liska <mliska@suse.cz>
31792 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31793 variable cond_code.
31795 2017-05-04 Richard Biener <rguenther@suse.de>
31797 * tree.c (array_at_struct_end_p): Handle arrays at struct
31798 end with flexarrays more conservatively. Refactor and treat
31799 arrays of arrays or aggregates more strict. Fix
31800 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31801 * tree.c (array_at_struct_end_p): Adjust prototype.
31802 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31803 * gimple-fold.c (get_range_strlen): Likewise.
31804 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31806 2017-05-04 Richard Biener <rguenther@suse.de>
31808 PR tree-optimization/31130
31809 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31811 (supports_overflow_infinity): Likewise.
31812 (is_negative_overflow_infinity): Likewise.
31813 (is_positive_overflow_infinity): Likewise.
31814 (is_overflow_infinity): Likewise.
31815 (stmt_overflow_infinity): Likewise.
31816 (overflow_infinity_range_p): Likewise.
31817 (usable_range_p): Remove as always returning true.
31818 (make_overflow_infinity): Remove.
31819 (negative_overflow_infinity): Likewise.
31820 (positive_overflow_infinity): Likewise.
31821 (avoid_overflow_infinity): Likewise.
31822 (set_value_range): Adjust accordingly.
31823 (set_value_range_to_nonnegative): Likewise, remove now unused
31824 overflow_infinity arg.
31825 (vrp_operand_equal_p): Adjust.
31826 (update_value_range): Likewise.
31827 (range_int_cst_singleton_p): Likewise.
31828 (operand_less_p): Likewise.
31829 (compare_values_warnv): Likewise.
31830 (extract_range_for_var_from_comparison_expr): Likewise.
31831 (vrp_int_const_binop): Likewise.
31832 (zero_nonzero_bits_from_vr): Likewise.
31833 (extract_range_from_multiplicative_op_1): Likewise.
31834 (extract_range_from_binary_expr_1): Likewise.
31835 (extract_range_from_unary_expr): Likewise.
31836 (extract_range_from_comparison): Likewise.
31837 (extract_range_basic): Likewise.
31838 (adjust_range_with_scev): Likewise.
31839 (compare_ranges): Likewise.
31840 (compare_range_with_value): Likewise.
31841 (dump_value_range): Likewise.
31842 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31844 (simplify_cond_using_ranges): Adjust.
31846 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31848 * brig-builtins.def: Added a builtin for class_f64.
31849 * builtin-types.def: Added a builtin type needed by class_f64.
31851 2017-05-03 Jason Merrill <jason@redhat.com>
31853 * timevar.def: Add TV_CONSTEXPR.
31855 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31857 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31859 2017-05-03 Martin Jambor <mjambor@suse.cz>
31861 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31862 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31863 * ipa-cp.c (ipcp_propagate_stage): Do not call
31864 ipa_update_after_lto_read.
31865 * ipa-inline.c (ipa_inline): Likewise.
31867 2017-05-03 Martin Jambor <mjambor@suse.cz>
31869 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31870 tag. Added a default constructor and a destructor.
31871 (ipa_edge_args_sum_t): New class;
31872 (ipa_edge_args_sum): Declare.
31873 (ipa_edge_args_vector): Remove declaration.
31874 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31875 (ipa_free_edge_args_substructures): Remove declaration.
31876 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31877 (ipa_edge_args_info_available_for_edge_p): Likewise.
31878 * ipa-prop.c (ipa_edge_args_vector): Removed.
31879 (edge_removal_hook_holder): Likewise.
31880 (edge_duplication_hook_holder): Likewise.
31881 (ipa_edge_args_sum): New variable.
31882 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31883 ipa_edge_args_vector.
31884 (ipa_free_edge_args_substructures): Likewise.
31885 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31886 ipa_edge_args_vector.
31887 (ipa_edge_removal_hook): Turned into method
31888 ipa_edge_args_sum_t::remove.
31889 (ipa_edge_duplication_hook): Turned into method
31890 ipa_edge_args_sum_t::duplicate.
31891 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31892 registering edge hooks.
31893 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31894 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31895 ipa_edge_args_sum instead of ipa_edge_args_vector.
31896 * ipa-profile.c (ipa_profile): Likewise.
31898 2017-05-03 Martin Jambor <mjambor@suse.cz>
31900 * symbol-summary.h (function_summary): New method exists.
31901 (function_summary::symtab_removal): Deallocate through release.
31902 (call_summary): New class.
31903 (gt_ggc_mx): New overload.
31904 (gt_pch_nx): Likewise.
31905 (gt_pch_nx): Likewise.
31907 2017-05-03 Jeff Law <law@redhat.com>
31909 PR tree-optimization/78496
31910 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31911 from simplify_cond_using_ranges. Split off code to walk
31912 backwards through casts into ...
31913 (simplify_cond_using_ranges_2): New function.
31914 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31915 (execute_vrp): After identifying jump threads, call
31916 simplify_cond_using_ranges_2.
31918 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31921 * ipa-inline.h (inline_summary): Add ctor.
31922 (create_ggc): Do not use ggc_cleared_alloc.
31924 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31925 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31927 * gcc.c (handle_braces): Support escaping in switch matching
31929 * doc/invoke.texi (Spec Files): Document it.
31930 Remove superfluous @code markup in items.
31932 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31934 * diagnostic-show-locus.c (struct column_range): New struct.
31935 (get_affected_columns): New function.
31936 (get_printed_columns): New function.
31937 (struct correction): New struct.
31938 (correction::ensure_capacity): New function.
31939 (correction::ensure_terminated): New function.
31940 (struct line_corrections): New struct.
31941 (line_corrections::~line_corrections): New dtor.
31942 (line_corrections::add_hint): New function.
31943 (layout::print_trailing_fixits): Reimplement in terms of the new
31945 (selftest::test_overlapped_fixit_printing): New function.
31946 (selftest::diagnostic_show_locus_c_tests): Call it.
31948 2017-05-03 Nathan Sidwell <nathan@acm.org>
31950 Canonicalize canonical type hashing
31951 * tree.h (type_hash_canon_hash): Declare.
31952 * tree.c (type_hash_list, attribute_hash_list): Move into
31953 type_hash_canon_hash.
31954 (build_type_attribute_qual_variant): Break out hash code calc into
31955 type_hash_canon_hash.
31956 (type_hash_canon_hash): New. Generic type hash computation.
31957 (build_range_type_1, build_array_type_1, build_function_type,
31958 build_method_type_directly, build_offset_type, build_complex_type,
31959 make_vector_type): Call it.
31961 2017-05-03 Richard Biener <rguenther@suse.de>
31963 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31964 When all DRs have unknown misaligned do not always peel
31965 when there is a store but apply the same costing model as if
31966 there were only loads.
31968 2017-05-03 Richard Biener <rguenther@suse.de>
31971 PR tree-optimization/80492
31972 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31973 compare_base_decls returning dont-know properly.
31975 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31977 * config/arm/iterators.md (CCSI): New mode iterator.
31978 (arch): New mode attribute.
31979 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31980 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31981 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31982 code iterator for success result mode.
31983 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31984 the corresponding new insn generators.
31986 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31989 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31990 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31992 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31994 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31995 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31996 (DDR_A): Wrap DDR argument in brackets.
31997 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31998 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31999 (DDR_REVERSED_P): Likewise.
32001 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32003 PR tree-optimization/79472
32004 * tree-switch-conversion.c (struct switch_conv_info): Add
32005 contiguous_range and default_case_nonstandard fields.
32006 (collect_switch_conv_info): Compute contiguous_range and
32007 default_case_nonstandard fields, don't clear final_bb if
32008 contiguous_range and only the default case doesn't have the required
32010 (check_all_empty_except_final): Set default_case_nonstandard instead
32011 of failing if contiguous_range and the default case doesn't have empty
32013 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32014 and only the default case doesn't have the required constants. Skip
32016 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32017 if default_case_nonstandard.
32018 (build_constructors): Build constant 1 just once. Assert that default
32019 values aren't inserted in between cases if contiguous_range. Skip
32021 (build_arrays): Skip virtual phis.
32022 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32023 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32024 Handle virtual phis.
32025 (gen_inbound_check): Handle default_case_nonstandard case.
32026 (process_switch): Adjust check_final_bb caller. Call
32027 gather_default_values with the first non-default case instead of
32028 default case if default_case_nonstandard.
32030 2017-05-02 Nathan Sidwell <nathan@acm.org>
32032 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32033 check. Fix formatting.
32035 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32037 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32038 errors when comparing specialized and unspecialized times.
32040 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32042 * diagnostic-show-locus.c
32043 (layout::should_print_annotation_line_p): Make private.
32044 (layout::print_annotation_line): Make private.
32045 (layout::annotation_line_showed_range_p): Make private.
32046 (layout::show_ruler): Make private.
32047 (layout::print_source_line): Make private. Pass in line and
32048 line_width, rather than calling location_get_source_line. Drop
32050 (layout::print_leading_fixits): New method.
32051 (layout::print_any_fixits): Rename to...
32052 (layout::print_trailing_fixits): ...this, and make private.
32053 Don't print newline fixits.
32054 (diagnostic_show_locus): Move logic for printing one row into...
32055 (layout::print_line): ...this new function. Move the
32056 location_get_source_line call and error-handling from
32057 print_source_line to here. Call print_leading_fixits, and rename
32058 print_any_fixits to print_trailing_fixits.
32059 (selftest::test_fixit_insert_containing_newline): Update now that
32060 newlines are partially supported.
32061 (selftest::test_fixit_insert_containing_newline_2): New test.
32062 (selftest::test_fixit_replace_containing_newline): Update comments.
32063 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32064 * edit-context.c (class added_line): New class.
32065 (class edited_line): Describe newline handling in comment.
32066 (edited_line::actually_edited_p): New method.
32067 (edited_line::print_content): Delete redundant decl.
32068 (edited_line::m_predecessors): New field.
32069 (edited_file::print_content): Call edited_line::print_content.
32070 (edited_file::print_diff): Update to support newlines.
32071 (edited_file::print_diff_hunk): Likewise.
32072 (edited_file::print_run_of_changed_lines): New function.
32073 (edited_file::print_diff_line): Convert to...
32074 (print_diff_line): ...this.
32075 (edited_file::get_effective_line_count): New function.
32076 (edited_line::edited_line): Initialize new field m_predecessors.
32077 (edited_line::~edited_line): Clean up m_predecessors.
32078 (edited_line::apply_fixit): Handle newlines.
32079 (edited_line::get_effective_line_count): New function.
32080 (edited_line::print_content): New function.
32081 (edited_line::print_diff_lines): New function.
32082 (selftest::test_applying_fixits_insert_containing_newline): New
32084 (selftest::test_applying_fixits_replace_containing_newline): New
32086 (selftest::insert_line): New function.
32087 (selftest::test_applying_fixits_multiple_lines): Add example of
32089 (selftest::edit_context_c_tests): Call the new tests.
32091 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32093 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32094 parameter cand. Update dump information.
32095 (get_computation_cost): Update uses.
32097 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32099 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32100 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32101 (get_computation_at, rewrite_use_address): Update use of
32102 get_computation_aff.
32104 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32106 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32107 (get_computation): Delete.
32108 (get_computation_cost): Implement like get_computation_cost_at.
32109 Use get_computation_at.
32110 (get_computation_cost_at): Delete.
32111 (rewrite_use_nonlinear_expr): Use get_computation_at.
32112 (rewrite_use_compare, remove_unused_ivs): Ditto.
32114 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32116 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32118 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32120 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32121 (ivopts_global_cost_for_size): Rename parameter and update uses.
32122 (iv_ca_recount_cost): Update uses.
32123 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32124 candidates seperately in n_invs and n_cands.
32125 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32127 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32129 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32130 (find_inv_vars_cb): New.
32131 (find_depends): Renamed to ...
32132 (find_inv_vars): ... this.
32133 (add_candidate_1, force_var_cost): Call find_inv_vars.
32134 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32136 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32138 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32139 inv_vars. Add inv_exprs.
32140 (struct iv_cand): Rename depends_on to inv_vars.
32141 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32142 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32143 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32144 (dump_cand): Dump inv_vars.
32145 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32146 (record_invariant, find_depends, add_candidate_1): Ditto.
32147 (set_group_iv_cost, force_var_cost): Ditto.
32148 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32149 (get_computation_cost_at, get_computation_cost): Ditto.
32150 (determine_group_iv_cost_generic): Ditto.
32151 (determine_group_iv_cost_address): Ditto.
32152 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32153 (determine_group_iv_costs): Ditto.
32154 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32155 (iv_ca_set_remove_invariants): Renamed to ...
32156 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32157 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32158 (iv_ca_set_add_invariants): Renamed to ...
32159 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32160 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32161 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32162 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32163 (create_new_ivs): Remove useless dump.
32165 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32167 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32169 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32170 (iv_ca_set_no_cp, create_new_iv): Ditto.
32172 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32174 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32176 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32178 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32179 function tree_check2.
32181 2017-05-02 Martin Liska <mliska@suse.cz>
32183 * doc/gcov.texi: Add missing preposition.
32184 * gcov.c (function_info::function_info): Properly fill up
32185 all member variables.
32187 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32189 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32191 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32193 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32195 2017-05-02 Martin Liska <mliska@suse.cz>
32198 * lto-streamer-in.c (lto_read_tree_1): Remove
32199 LTO_STREAMER_DEBUG.
32200 * lto-streamer.c (struct tree_hash_entry): Likewise.
32201 (struct tree_entry_hasher): Likewise.
32202 (tree_entry_hasher::hash): Likewise.
32203 (tree_entry_hasher::equal): Likewise.
32204 (lto_streamer_init): Likewise.
32205 (lto_orig_address_map): Likewise.
32206 (lto_orig_address_get): Likewise.
32207 (lto_orig_address_remove): Likewise.
32208 * lto-streamer.h: Likewise.
32209 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32210 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32212 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32214 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32215 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32216 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32217 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32218 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32219 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32220 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32221 (mm_maskz_sub_ss): New intrinsics.
32222 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32223 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32224 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32225 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32226 (__builtin_ia32_subss_mask_round): New builtins.
32227 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32228 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32229 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32231 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32232 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32234 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32237 2017-05-02 Martin Jambor <mjambor@suse.cz>
32239 PR tree-optimization/78687
32240 * tree-sra.c (access): New field parent.
32241 (process_subtree_disqualification): New function.
32242 (disqualify_candidate): Call it.
32243 (build_accesses_from_assign): Reset write flag if creating an
32245 (build_access_subtree): Fill in parent field and also prpagate
32246 down grp_write flag.
32247 (create_artificial_child_access): New parameter set_grp_write, set
32248 grp_write to its value.
32249 (propagate_subaccesses_across_link): Also propagate grp_write flag
32251 (propagate_all_subaccesses): Push the closest parent back to work
32252 queue if add_access_to_work_queue returned true.
32254 2017-05-02 Richard Biener <rguenther@suse.de>
32256 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32257 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32258 -fstrict-overflow documentation.
32259 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32260 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32261 flag_strict_overflow.
32262 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32263 * lto-opts.c (lto_write_options): Do not stream it.
32264 * lto-wrapper.c (merge_and_complain): Do not handle it.
32265 * opts.c (default_options_table): Do not set -fstrict-overflow.
32266 (finish_options): Likewise do not clear it when sanitizing.
32267 * simplify-rtx.c (simplify_const_relational_operation): Do not
32268 test flag_strict_overflow.
32270 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32272 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32273 using enabled attribute.
32274 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32275 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32276 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32277 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32278 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32279 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32280 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32281 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32282 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32283 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32285 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32287 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32289 2017-05-02 Richard Biener <rguenther@suse.de>
32291 PR tree-optimization/80591
32293 2017-04-10 Richard Biener <rguenther@suse.de>
32295 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32298 2017-05-02 Richard Biener <rguenther@suse.de>
32300 PR tree-optimization/80549
32301 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32302 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32303 headers if they do not have a preheader.
32305 2017-05-02 Martin Liska <mliska@suse.cz>
32308 * common.opt: Fix typo.
32309 * doc/invoke.texi: Likewise.
32311 2017-05-01 Jan Beulich <jbeulich@suse.com>
32313 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32314 swapping, add (x,x,m,x,n) alternative.
32316 2017-05-01 Nathan Sidwell <nathan@acm.org>
32318 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32319 unnecessary unadjusted_alignment check.
32321 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32324 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32326 * gimplify.c (gimplify_cilk_detach): New function.
32327 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32328 * tree-core.h: Document EXPR_CILK_SPAWN.
32329 * tree.h (EXPR_CILK_SPAWN): Define.
32331 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32333 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32334 to use new fixit_hint representation, using the "replace" logic.
32335 (get_line_span_for_fixit_hint): Likewise.
32336 (layout::print_any_fixits): Likewise.
32337 (selftest::test_one_liner_many_fixits): Rename to...
32338 (selftest::test_one_liner_many_fixits_1): ...this, and update
32339 comment and expected output to reflect that the multiple fix-it
32340 hints are now consolidated into one insertion.
32341 (selftest::test_one_liner_many_fixits_2): New test.
32342 (selftest::test_diagnostic_show_locus_one_liner): Update for
32344 (selftest::test_fixit_consolidation): Update for fix-it API
32346 * diagnostic.c (print_parseable_fixits): Likewise.
32347 * edit-context.c (edited_line::m_line_events): Convert from
32348 auto_vec <line_event *> to auto_vec <line_event>.
32349 (class line_event): Convert from abstract base class to a concrete
32350 class, taking over the role of replace_event.
32351 (class insert_event): Delete.
32352 (class replace_event): Rename to class line_event. Convert to
32354 (edit_context::add_fixits): Reimplement.
32355 (edit_context::apply_insert): Delete.
32356 (edit_context::apply_replace): Rename to...
32357 (edit_context::apply_fixit): ...this. Convert to half-open range.
32358 (edited_file::apply_insert): Delete.
32359 (edited_file::apply_replace): Rename to...
32360 (edited_file::apply_fixit): ...this.
32361 (edited_line::~edited_line): Drop deletion of events.
32362 (edited_line::apply_insert): Delete.
32363 (edited_line::apply_replace): Rename to...
32364 (edited_line::apply_fixit): ...this. Convert to half-open range.
32365 Update for change to type of m_line_events.
32366 * edit-context.h (edit_context::apply_insert): Delete.
32367 (edit_context::apply_replace): Rename to...
32368 (edit_context::apply_fixit): ...this.
32370 2017-05-01 Martin Sebor <msebor@redhat.com>
32372 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32375 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32378 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32379 __get_cpuid_max returns 0.
32380 (__get_cpuid_count): Ditto.
32382 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32384 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32385 replacement expression is another instance of one of its arguments.
32387 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32390 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32391 check for stack push/pop autoinc.
32392 * config/i386/i386.c (ix86_agi_dependent): Return false
32393 if the only reason why modified_in_p returned true is that
32394 addr is SP based and set_insn is a push or pop.
32396 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32398 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32401 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32404 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32405 (account_size_time): Use two predicates - exec_pred and
32407 (evaluate_conditions_for_known_args): Compute both clause and
32409 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32410 (inline_summary_t::duplicate): Update.
32411 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32413 (compute_inline_parameters): Likewise.
32414 (estimate_edge_size_and_time): Update caluclation of time.
32415 (estimate_node_size_and_time): Compute both time and nonspecialized
32417 (estimate_ipcp_clone_size_and_time): Update.
32418 (inline_merge_summary): Update.
32419 (do_estimate_edge_time): Update.
32420 (do_estimate_edge_size): Update.
32421 (do_estimate_edge_hints): Update.
32422 (inline_read_section, inline_write_summary): Stream both new predicates.
32423 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32425 (compute_inlined_call_time): Cleanup.
32426 (big_speedup_p): Update.
32427 (edge_badness): Update.
32428 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32429 (size_time_entry): Replace predicate by exec_predicate and
32430 nonconst_predicate.
32431 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32432 (estimate_edge_time): Return also nonspec_time.
32433 (reset_edge_growth_cache): Update.
32435 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32437 PR rtl-optimization/80491
32438 * ifcvt.c (noce_process_if_block): When looking for x setter
32439 with missing else_bb, don't check only the insn right before
32440 cond_earliest, but look for the last insn that x is modified in
32441 within the same bb.
32443 PR rtl-optimization/80491
32444 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32446 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32448 PR tree-optimization/80487
32449 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32451 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32453 PR tree-optimization/79697
32454 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32455 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32456 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32458 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32459 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32461 2017-04-28 Martin Sebor <msebor@redhat.com>
32463 PR tree-optimization/80523
32464 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32465 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32467 (maybe_warn, format_directive, parse_directive): Use new functions.
32468 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32470 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32472 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32474 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32476 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32477 target_header_dir): Set correctly.
32478 * configure: Regenerated.
32479 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32480 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32481 instead of SYSTEM_HEADER_DIR.
32483 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32485 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32486 (estimate_local_effects): Likewise.
32487 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32488 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32489 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32490 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32491 do_estimate_edge_time, estimate_edge_time): Likewise.
32492 * ipa-inline-analysis.c (estimate_node_size_and_time,
32493 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32494 (estimate_time_after_inlining): Remove.
32496 2017-04-28 Martin Liska <mliska@suse.cz>
32498 * doc/gcov.texi: Enhance documentation of gcov.
32500 2017-04-28 Martin Liska <mliska@suse.cz>
32502 * doc/gcov.texi: Sort options in alphabetic order.
32503 * doc/gcov-dump.texi: Likewise.
32504 * doc/gcov-tool.texi: Likewise.
32505 * gcov.c (print_usage): Likewise.
32506 * gcov-dump.c (print_usage): Likewise.
32507 * gcov-tool.c (print_merge_usage_message): Likewise.
32508 (print_rewrite_usage_message): Likewise.
32509 (print_overlap_usage_message): Likewise.
32511 2017-04-28 Martin Liska <mliska@suse.cz>
32513 PR gcov-profile/53915
32514 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32516 2017-04-28 Martin Liska <mliska@suse.cz>
32518 PR gcov-profile/79891
32519 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32520 is marked by compiler as living on a line.
32521 (get_cycles_count): Remove usage of the union.
32522 (output_intermediate_file): Likewise.
32523 (find_source): Fix GNU coding style.
32524 (accumulate_line_counts): Remove old non-all block mode.
32525 (output_lines): Remove usage of the union.
32526 * profile.c (output_location): Include all BBs, even if
32527 belonging to a same line (and file) as a previous BB.
32529 2017-04-28 Martin Liska <mliska@suse.cz>
32531 * gcov.c (process_args): Handle new argument 'w'.
32532 (read_graph_file): Assign ID to BBs.
32533 (output_branch_count): Display BB # if verbose flag is set.
32534 (output_lines): Likewise for arcs.
32535 (print_usage): Add '--verbose' option help.
32536 * doc/gcov.texi: Document --verbose (-w) option.
32538 2017-04-28 Martin Liska <mliska@suse.cz>
32540 * gcov.c (struct block_location_info): New struct.
32541 (process_file): Fill up the new structure.
32542 (read_graph_file): Replace usage of encoding by the newly added
32544 (add_line_counts): Likewise.
32545 (accumulate_line_counts): Remove usage of the union.
32546 (function_info::function_info): New function.
32547 (function_info::~function_info): Likewise.
32548 (process_file): Call delete instead of release_function.
32549 (release_function): Release the function.
32550 (release_structures): Call delete instead of release_function.
32551 (solve_flow_graph): Replace usage of num_blocks.
32552 (find_exception_blocks): Likewise.
32553 (output_lines): Fix GNU coding style.
32555 2017-04-28 Martin Liska <mliska@suse.cz>
32558 * coverage.c (coverage_remove_note_file): New function.
32559 * coverage.h: Declare the function.
32560 * toplev.c (finalize): Clean if an error has been seen.
32562 2017-04-28 Martin Liska <mliska@suse.cz>
32564 PR gcov-profile/80031
32565 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32566 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32567 * gcov.c (read_graph_file): Read just number of blocks.
32568 * profile.c (branch_prob): Do not stream 0 flags per a basic
32571 2017-04-28 Martin Liska <mliska@suse.cz>
32573 * gcov-dump.c (tag_*): Add new argument to declarations.
32574 (dump_gcov_file): Likewise.
32575 (tag_blocks): Add and use new argument depth.
32576 (tag_arcs): Likewise.
32577 (tag_lines): Likewise.
32578 (tag_counters): Likewise.
32579 (tag_summary): Likewise.
32580 (dump_working_sets): Use depth to do a proper indentation.
32582 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32585 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32586 * symtab.c (symtab_node::debug_symtab): Move definition here.
32588 2017-04-28 Richard Biener <rguenther@suse.de>
32590 * lto-streamer.h (LTO_major_version): Bump to 7.
32592 2017-04-28 Richard Biener <rguenther@suse.de>
32594 * tree-vrp.c (assert_info): New struct.
32595 (add_assert_info): New helper.
32596 (register_edge_assert_for_2): Refactor to add asserts to a vector
32598 (register_edge_assert_for_1): Likewise.
32599 (register_edge_assert_for): Likewise.
32600 (finish_register_edge_assert_for): New helper actually registering
32601 asserts where live on edge.
32602 (find_conditional_asserts): Adjust.
32603 (find_switch_asserts): Likewise.
32604 (evrp_dom_walker::try_find_new_range): Generalize.
32605 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32607 2017-04-27 Marek Polacek <polacek@redhat.com>
32610 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32611 arg10 and arg11 to itype.
32613 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32615 * doc/extend.texi (Object Size Checking): Improve grammar.
32617 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32620 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32621 that the logic for permitting reciprocal estimates matches that
32624 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32627 * tree.c (type_cache_hasher::equal): Only compare
32628 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32629 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32630 non-aggregate element types.
32631 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32632 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32634 2017-04-27 Richard Biener <rguenther@suse.de>
32636 PR middle-end/80533
32637 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32638 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32639 keeping a reference to a trailing array.
32641 2017-04-27 Richard Biener <rguenther@suse.de>
32643 PR middle-end/80539
32644 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32645 being in loop-closed SSA form conservatively.
32646 (chrec_fold_multiply_poly_poly): Likewise.
32648 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32650 PR middle-end/79665
32651 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32652 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32654 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32657 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32658 (aarch64_function_arg_alignment): Return unsigned int again, but still
32659 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32660 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32661 Don't emit -Wpsabi note.
32662 (aarch64_function_arg_boundary): Likewise.
32663 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32666 2017-04-26 Nathan Sidwell <nathan@acm.org>
32668 * tree.h (crc32_unsigned_n): Declare.
32669 (crc32_unsigned, crc32_unsigned): Make inline.
32670 * tree.c (crc32_unsigned_bits): Replace with ...
32671 (crc32_unsigned_n): ... this.
32672 (crc32_unsigned, crc32_byte): Remove.
32673 (crc32_string): Remove unnecessary braces.
32675 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32677 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32678 * ipa-inline-analysis.c (MAX_TIME): Remove.
32679 (account_size_time): Use sreal for time.
32680 (dump_inline_summary): Update.
32681 (estimate_function_body_sizes): Update.
32682 (estimate_edge_size_and_time): Update.
32683 (estimate_calls_size_and_time): Update.
32684 (estimate_node_size_and_time): Update.
32685 (inline_merge_summary): Update.
32686 (inline_update_overall_summary): Update.
32687 (estimate_time_after_inlining): Update.
32688 (inline_read_section): Update.
32689 (inline_write_summary): Update.
32690 * ipa-inline.c (compute_uninlined_call_time): Update.
32691 (compute_inlined_call_time): Update.
32692 (recursive_inlining): Update.
32693 (inline_small_functions): Update.
32694 (dump_overall_stats): Update.
32695 * ipa-inline.h: Include sreal.h.
32696 (size_time_entry): Turn time to sreal.
32697 (inline_summary): Turn self_time nad time to sreal.
32699 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32701 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32703 (sreal::stream_out, sreal::stream_in): New.
32704 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32706 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32708 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32711 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32714 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32716 (dimode_scalar_chain::compute_convert_gain): Ditto.
32717 (dimode_scalar_chain::make_vector_copies): Ditto.
32718 (dimode_scalar_chain::convert_reg): Ditto.
32719 (dimode_scalar_chain::convert_insn): Ditto.
32720 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32721 (VI248_AVX512BW_1): New mode iterator.
32722 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32723 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32726 2017-04-25 Martin Sebor <msebor@redhat.com>
32728 PR tree-optimization/80497
32729 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32730 constants are representable in HOST_WIDE_INT.
32731 (parse_directive): Ditto.
32733 2017-04-25 Martin Sebor <msebor@redhat.com>
32736 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32737 (new_zero_array): Adjust signature.
32738 (dom_info::dom_init): Used unsigned rather that size_t.
32739 (dom_info::dom_info): Same.
32741 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32742 Jakub Jelinek <jakub@redhat.com>
32745 * config/arm/arm.c: Include gimple.h.
32746 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32747 returns negative, increment ncrn only if it returned positive.
32748 (arm_needs_doubleword_align): Return int instead of bool,
32749 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32750 members, but if there is any such non-FIELD_DECL
32751 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32752 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32753 returns negative, increment nregs only if it returned positive.
32754 (arm_setup_incoming_varargs): Likewise.
32755 (arm_function_arg_boundary): Emit -Wpsabi note if
32756 arm_needs_doubleword_align returns negative, return
32757 DOUBLEWORD_ALIGNMENT only if it returned positive.
32759 2017-04-25 Marek Polacek <polacek@redhat.com>
32762 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32763 first argument to type.
32765 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32768 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32769 type checks to test for compatibility instead of equality.
32771 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32772 Jakub Jelinek <jakub@redhat.com>
32775 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32777 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32778 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32779 the alignment computation, but return their maximum in warn_alignment.
32780 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32781 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32783 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32784 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32787 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32789 * config/arc/simdext.md (dmpyh): Fix typo.
32791 2017-04-25 Richard Biener <rguenther@suse.de>
32793 PR tree-optimization/80492
32794 * alias.c (compare_base_decls): Handle registers with asm
32795 specification conservatively.
32796 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32797 compare_base_decls returning dont-know properly.
32799 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32801 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32802 (legitimate_offset_address_p): New function.
32803 (arc_legitimate_address_p): Use above function.
32805 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32807 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32809 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32811 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32812 ACCH registers whenever they are available.
32814 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32816 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32817 double regs fix when not used.
32819 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32821 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32823 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32824 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32826 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32828 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32829 when emitting short ADD instructions.
32831 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32833 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32835 (cmpsi_cc_c_insn): Likewise.
32836 (cbranchsi4_scratch): Compute proper instruction length using
32837 compact_hreg_operand.
32838 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32840 2017-04-25 Richard Biener <rguenther@suse.de>
32842 PR middle-end/80509
32843 * passes.c (pass_manager::pass_manager): Initialize
32844 m_name_to_pass_map.
32846 2017-04-25 Richard Biener <rguenther@suse.de>
32848 PR tree-optimization/79201
32849 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32851 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32854 * config/s390/vector.md: Split MEM->GPR vector moves for
32855 non-s_operand addresses.
32857 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32860 * config/s390/predicates.md (reload_const_wide_int_operand): New
32862 * config/s390/s390.md ("movti"): Remove d/P alternative.
32863 ("movti_bigconst"): New pattern definition.
32865 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32868 * s390-protos.h (s390_expand_cs_hqi): Removed.
32869 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32870 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32871 modes as well as CCZ1mode and CCZmode.
32872 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32873 signature of s390_emit_compare_and_swap.
32874 (s390_expand_cs_hqi): Likewise, make static.
32875 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32876 compare-and-swap, in some cases.
32877 (s390_expand_cs): Wrapper function.
32878 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32880 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32881 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32882 patterns for small and large integers. Forbid symref memory operands.
32883 Move expander to s390.c. Require cc register.
32884 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32885 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32886 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32887 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32888 symref memory operands. Remove CC mode and call s390_match_ccmode
32890 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32892 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32894 * config/s390/s390.md (define_peephole2): New peephole to help
32895 combining the load-and-test pattern with volatile memory.
32897 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32899 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32900 with CCZmode for TARGET_Z196.
32902 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32904 PR rtl-optimization/80501
32905 * combine.c (make_compound_operation_int): Set subreg_code to SET
32906 even for AND with mask of the sign bit of mode.
32908 PR rtl-optimization/80500
32909 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32910 sum's initial value.
32912 2017-04-25 Julian Brown <julian@codesourcery.com>
32913 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32915 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32917 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32919 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32921 2017-04-25 Julian Brown <julian@codesourcery.com>
32922 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32924 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32925 (thunderx2t99_sha): New Reservation.
32927 2017-04-25 Julian Brown <julian@codesourcery.com>
32928 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32930 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32931 type for 1-element load.
32933 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32935 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32937 2017-04-24 Martin Jambor <mjambor@suse.cz>
32939 PR tree-optimization/80293
32940 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32941 char arrays not totally scalarizable if it is false.
32942 (analyze_all_variable_accesses): Pass correct value in the new
32943 parameter. Add a statistics counter.
32945 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32947 PR middle-end/79931
32948 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32950 2017-04-24 Richard Biener <rguenther@suse.de>
32952 PR tree-optimization/80494
32953 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32954 out for complex types.
32956 2017-04-24 Richard Biener <rguenther@suse.de>
32958 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32959 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32960 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32961 (DFS): Adjust and never fail.
32962 (sccvn_dom_walker::fail): Remove.
32963 (sccvn_dom_walker::before_dom_children): Adjust.
32964 (run_scc_vn): Likewise and never fail.
32965 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32966 (pass_fre::execute): Likewise.
32968 2017-04-24 Richard Biener <rguenther@suse.de>
32970 PR tree-optimization/79725
32971 * tree-ssa-sink.c (statement_sink_location): Return whether
32972 failure reason was zero uses. Move that check later.
32973 (sink_code_in_bb): Deal with zero uses by removing the stmt
32976 2017-04-24 Richard Biener <rguenther@suse.de>
32979 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32980 pointer-based references.
32982 2017-04-24 Richard Biener <rguenther@suse.de>
32985 * pass_manager.h (pass_manager::operator new): Remove.
32986 (pass_manager::operator delete): Likewise.
32987 * passes.c (pass_manager::operator new): Remove.
32988 (pass_manager::operator delete): Likewise.
32989 (pass_manager::pass_manager): Zero individual pass members.
32991 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32994 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32995 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32996 Check "XEXP (src, 1)" operand here.
32997 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32998 Check "XEXP (src, 1)" operand here.
32999 (dimode_scalar_chain::make_vector_copies): Detect count register
33000 of a shift instruction. Zero extend count register from QImode
33001 to DImode to satisfy vector shift pattern count operand predicate.
33002 Substitute vector shift count operand with a DImode copy.
33003 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33006 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33008 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33009 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33010 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33011 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33012 (UNSPEC_NOREX_MEM): Remove definition.
33014 2017-04-21 Richard Biener <rguenther@suse.de>
33016 PR tree-optimization/79547
33017 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33018 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33019 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33020 without any constraints.
33022 2017-04-21 Richard Biener <rguenther@suse.de>
33024 PR tree-optimization/78847
33025 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33027 2017-04-21 Richard Biener <rguenther@suse.de>
33029 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33030 (build_distinct_type_copy): Likewise.
33031 (build_variant_type_copy): Likewise.
33032 * tree.c (build_qualified_type): Pass down mem-stat info.
33033 (build_distinct_type_copy): Likewise.
33034 (build_variant_type_copy): Likewise.
33036 2017-04-21 Richard Biener <rguenther@suse.de>
33038 PR tree-optimization/80237
33039 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33041 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33042 for a simplified result.
33044 2016-04-21 Richard Biener <rguenther@suse.de>
33046 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33047 sth as strict as a simple_iv but a chrec without symbols and an
33048 operand defined in the loop we are peeling (and not some subloop).
33049 (propagate_constants_for_unrolling): Propagate all constants.
33051 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33054 * config/i386/i386.c (print_reg): Remove assert for disalowed
33055 regno values, call output_operand_lossage instead.
33057 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33060 * config/i386/constraints.md (Yc): New register constraint.
33061 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33062 Use Yc constraint for alternative 2 of operand 0. Remove
33063 preferred_for_speed attribute.
33065 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33067 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33068 lastprivate clauses in SIMT case.
33070 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33072 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33074 2017-04-20 Richard Biener <rguenther@suse.de>
33076 PR tree-optimization/57796
33077 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33078 as N scalar stores.
33079 (vect_model_load_cost): Cost gathers as N scalar loads.
33081 2017-04-20 Richard Biener <rguenther@suse.de>
33083 * ggc-page.c (ggc_allocated_p): Rename to ...
33084 (safe_lookup_page_table_entry): ... this and return the lookup
33086 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33088 2017-04-20 Richard Biener <rguenther@suse.de>
33090 PR tree-optimization/80453
33091 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33092 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33093 from the conditions.
33094 (vn_phi_eq): Pass them down.
33095 (vn_phi_lookup): Record them.
33096 (vn_phi_insert): Likewise.
33098 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33100 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33101 uninitialized variable warning to avoid buffer overrun.
33103 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33106 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33107 is suppressed for '{ 0 }' in C.
33109 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33111 * BASE-VER: Set to 8.0.0.
33113 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33115 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33116 priority .init_array and .fini_array section with SECTION_NOTYPE
33119 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33121 PR middle-end/80423
33122 * tree.h (build_array_type): Add typeless_storage default argument.
33123 * tree.c (type_cache_hasher::equal): Also compare
33124 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33125 (build_array_type): Add typeless_storage argument, set
33126 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33128 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33129 (build_array_type): Likewise. Add typeless_storage argument.
33131 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33132 Jakub Jelinek <jakub@redhat.com>
33134 PR tree-optimization/80426
33135 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33136 operation on symbolic operands, also compute the overflow for the
33137 invariant part when the operation degenerates into a negation.
33139 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33142 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33143 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33146 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33148 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33151 * config/avr/avr.c (tree.h): Include it.
33152 (cgraph.h): Include it.
33153 (avr_encode_section_info): Don't warn for uninitialized progmem
33154 variable if it's just an alias.
33156 2017-04-19 Richard Biener <rguenther@suse.de>
33159 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33160 when needed by AutoPGO.
33162 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33165 * doc/lto.texi: Remove an extra 'that'.
33167 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33169 PR rtl-optimization/80429
33170 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33171 are only used in debug insns.
33173 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33174 Vladimir Makarov <vmakarov@redhat.com>
33176 * config/sparc/predicates.md (input_operand): Add comment. Return
33177 true for any memory operand when LRA is in progress.
33178 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33180 2017-04-18 Jeff Law <law@redhat.com>
33183 * mips.md ({return,simple_return}_internal): Do not overwrite
33186 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33188 PR tree-optimization/80443
33189 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33190 instead of adding 1, subtract -1 and similarly instead of subtracting
33193 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33195 PR rtl-optimization/80357
33196 * haifa-sched.c (tmp_bitmap): New variable.
33197 (model_recompute): Handle duplicate use records.
33198 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33199 (free_global_sched_pressure_data): Free it.
33201 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33204 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33205 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33206 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33207 instead of SYSTEM_HEADER_DIR.
33209 2017-04-18 Jeff Law <law@redhat.com>
33211 PR middle-end/80422
33212 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33213 predecessors after walking up the insn chain.
33215 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33218 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33219 sizetype type into debug info.
33221 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33224 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33225 unneeded test for TARGET_UPPER_REGS_SF.
33226 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33228 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33231 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33232 instead of gsi_after_labels.
33234 2017-04-18 Jeff Law <law@redhat.com>
33236 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33240 2017-04-13 Jeff Law <law@redhat.com>
33241 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33242 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33244 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33247 * config/avr/avr.c (intl.h): Include it.
33248 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33250 2017-04-18 Martin Liska <mliska@suse.cz>
33252 PR gcov-profile/78783
33253 * gcov-tool.c (gcov_output_files): Validate that destination
33254 file is either removed by the tool or by a user.
33256 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33257 Guy Benyei <guybe@mellanox.com>
33259 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33260 block, and do not negate it, the stored id is already negative.
33262 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33264 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33266 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33269 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33270 masks of options that should be turned off if the VSX vector
33271 options are turned off.
33272 (OTHER_P8_VECTOR_MASKS): Likewise.
33273 (OTHER_VSX_VECTOR_MASKS): Likewise.
33274 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33275 rs6000_disable_incompatible_switches to validate no type switches
33277 (rs6000_incompatible_switch): New function to disallow turning on
33278 other vector options if -mno-vsx, -mno-power8-vector, or
33279 -mno-power9-vector are specified.
33281 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33283 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33285 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33287 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33288 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33289 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33290 (ARG_POINTER_CFA_OFFSET): Likewise.
33292 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33294 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33295 conditions to take advantage of various optimizations.
33297 2017-04-13 Jeff Law <law@redhat.com>
33299 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33300 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33301 (zero_extendsidi2_dext): Likewise.
33303 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33306 * fold-const.c (fold_ternary_loc): Revert
33307 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33310 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33312 PR rtl-optimization/80343
33313 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33314 new scratch pseudo.
33316 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33319 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33320 to ubsan_encode_value.
33322 2017-04-13 Jeff Law <law@redhat.com>
33324 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33325 appearing in DEBUG_INSNs.
33327 2017-04-13 Martin Liska <mliska@suse.cz>
33329 PR gcov-profile/80413
33330 * gcov-io.c (gcov_write_string): Copy to buffer just when
33331 allocated size is greater than zero.
33333 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33336 * dwarf2out.c (decls_for_scope): Ignore declarations of
33337 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33339 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33342 * ipa-visibility.c (non_local_p): Fix typos.
33343 (localize_node): When localizing symbol in same comdat group,
33344 dissolve the group only when we know external symbols are going
33346 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33348 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33350 PR tree-optimization/79390
33351 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33352 order does not result in usable sequence, retry with reversed operand
33358 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33359 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33360 op0 instead of fold_convert_loc (loc, type, arg0).
33362 2017-04-12 Jeff Law <law@redhat.com>
33364 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33365 has a delay slot in the generated code.
33367 * config/cris/cris.md (cris_preferred_reload_class): Return
33368 GENNONACR_REGS rather than GENERAL_REGS.
33370 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33373 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33374 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33375 signedness of the result type.
33377 2017-04-12 Richard Biener <rguenther@suse.de>
33378 Jeff Law <law@redhat.com>
33380 PR tree-optimization/80359
33381 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33382 trim stores to TARGET_MEM_REFs.
33384 2017-04-12 Richard Biener <rguenther@suse.de>
33386 PR tree-optimization/79390
33387 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33388 threading case even more.
33390 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33393 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33394 for quad_address_p for TImode, instead of just not indexed_address.
33396 2017-04-12 Richard Biener <rguenther@suse.de>
33397 Bernd Edlinger <bernd.edlinger@hotmail.de>
33399 PR middle-end/79671
33400 * alias.c (component_uses_parent_alias_set_from): Handle
33401 TYPE_TYPELESS_STORAGE.
33402 (get_alias_set): Likewise.
33403 * tree-core.h (tree_type_common): Add typeless_storage flag.
33404 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33405 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33406 for types containing members with TYPE_TYPELESS_STORAGE.
33407 (place_field): Likewise.
33408 (layout_type): Likewise for ARRAY_TYPE.
33409 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33410 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33411 TYPE_TYPELESS_STORAGE.
33412 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33414 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33417 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33418 first argument to type.
33420 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33424 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33425 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33426 (rs6000_expand_binop_builtin): Likewise.
33427 (rs6000_expand_ternop_builtin): Likewise; also add missing
33428 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33429 vshasigma built-ins.
33430 * doc/extend.texi: Document that vec_xxpermdi's third argument
33431 must be a constant.
33433 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33435 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33436 Use shift_const cost parameter when calculating gain of STV shifts.
33438 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33440 PR rtl-optimization/70478
33441 * lra-constraints.c (process_alt_operands): Check memory for
33442 disfavoring memory insn operand.
33444 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33446 PR middle-end/80100
33447 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33448 left shift in unsigned HOST_WIDE_INT type.
33450 PR rtl-optimization/80385
33451 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33452 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33455 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33456 if they have any depend clauses.
33458 2017-04-11 Martin Liska <mliska@suse.cz>
33461 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33462 * ipa-split.c (split_function): Create a local comdat symbol
33463 if caller is in a comdat group.
33465 2017-04-11 Martin Liska <mliska@suse.cz>
33468 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33471 2017-04-11 Martin Sebor <msebor@redhat.com>
33473 PR middle-end/80364
33474 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33475 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33477 (directive::set_width, directive::set_precision, format_character):
33479 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33482 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33485 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33486 conflict, set target->arch_name instead of target->cpu_name.
33488 2017-04-11 Richard Biener <rguenther@suse.de>
33490 PR tree-optimization/80374
33491 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33492 build_zero_cst, remove fold_convertible_p check again.
33494 2017-04-11 Martin Liska <mliska@suse.cz>
33497 * ubsan.c (instrument_object_size): Do not instrument register
33500 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33503 * config/i386/i386-builtin-types.def
33504 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33505 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33506 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33507 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33508 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33509 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33510 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33511 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33512 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33513 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33514 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33515 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33516 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33517 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33518 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33519 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33520 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33521 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33522 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33523 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33524 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33525 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33526 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33527 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33528 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33529 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33530 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33531 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33532 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33533 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33534 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33535 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33536 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33537 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33538 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33539 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33540 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33541 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33542 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33543 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33544 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33545 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33546 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33547 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33548 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33549 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33551 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33552 flag to second_arg_count, handle 4 argument function type _COUNT
33553 aliases, handle second_arg_count on second argument rather than last.
33555 2017-04-10 Jeff Law <law@redhat.com>
33557 PR tree-optimization/80374
33558 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33559 record anything if we can not convert integer_zero_node to the
33562 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33565 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33566 Enhance special handling given to the TARGET_P9_MINMAX option in
33567 relation to certain other options.
33569 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33571 PR tree-optimization/80153
33572 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33573 remove POINTER_PLUS_EXPR's base part directly, rather than through
33576 2017-04-10 Richard Biener <rguenther@suse.de>
33577 Bin Cheng <bin.cheng@arm.com>
33579 PR tree-optimization/80153
33580 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33581 the first element of pointer type aff_tree. Build result expr in
33583 (add_elt_to_tree): Convert to type unconditionally. Remove other
33584 fold_convert calls.
33585 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33586 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33588 2017-04-10 Richard Biener <rguenther@suse.de>
33590 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33593 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33595 PR rtl-optimization/70478
33596 * lra-constraints.c (curr_small_class_check): New.
33597 (update_and_check_small_class_inputs): New.
33598 (process_alt_operands): Update curr_small_class_check. Disfavor
33599 alternative insn memory operands. Check available regs for small
33602 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33605 * config/mips/mips.opt (-mvirt): Update description.
33606 * doc/invoke.texi (-mvirt): Likewise.
33608 2017-04-10 Richard Biener <rguenther@suse.de>
33610 PR middle-end/80362
33611 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33612 looking for NEGATE_EXPR in -A / -B to A / B folding.
33614 2017-04-10 Martin Liska <mliska@suse.cz>
33616 PR gcov-profile/80224
33617 * gcov.c (print_usage): Fix usage string.
33618 (get_gcov_intermediate_filename): Remove.
33619 (output_gcov_file): Use both for normal and intermediate format.
33620 (generate_results): Do not initialize special file for
33621 intermediate format.
33623 2017-04-10 Richard Biener <rguenther@suse.de>
33625 PR tree-optimization/80304
33626 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33629 2017-04-10 Nathan Sidwell <nathan@acm.org>
33632 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33633 (rs6000_init_builtins): Use it.
33635 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33637 * config/arm/arm.md (<mrc>): Add mode to SET source.
33638 (<mrrc>): Likewise.
33640 2017-04-10 Richard Biener <rguenther@suse.de>
33642 PR middle-end/80344
33643 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33645 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33648 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33649 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33650 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33651 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33652 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33653 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33654 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33655 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33656 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33657 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33658 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33659 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33660 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33661 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33662 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33663 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33664 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33665 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33666 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33667 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33668 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33669 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33670 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33672 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33674 PR rtl-optimization/70478
33675 * lra-constraints.c: Reverse the last patch.
33677 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33679 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33680 Add comment for WCHAR_T.
33682 2017-04-08 Martin Liska <mliska@suse.cz>
33685 2017-04-07 Martin Liska <mliska@suse.cz>
33688 * ipa-split.c (split_function): Add function part to a same comdat
33691 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33694 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33696 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33698 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33699 (vec_pairing): ...to this.
33700 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33701 (rs6000_sched_init): Adjust for name change.
33702 (struct rs6000_sched_context): Likewise.
33703 (rs6000_init_sched_context): Likewise.
33704 (rs6000_set_sched_context): Likewise.
33706 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33712 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33714 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33715 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33716 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33718 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33720 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33722 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33724 PR rtl-optimization/70703
33725 * ira-color.c (update_conflict_hard_regno_costs): Use
33726 int64_t instead of HOST_WIDE_INT.
33728 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33730 PR rtl-optimization/70478
33731 * lra-constraints.c (process_alt_operands): Disfavor alternative
33732 insn memory operands.
33734 2017-04-07 Jeff Law <law@redhat.com>
33736 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33737 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33739 2017-04-07 Martin Liska <mliska@suse.cz>
33742 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33743 Show error message instead of an ICE.
33745 2017-04-07 Martin Liska <mliska@suse.cz>
33748 * ipa-split.c (split_function): Add function part to a same comdat
33751 2017-04-07 Richard Biener <rguenther@suse.de>
33753 PR middle-end/80341
33754 * tree.c (get_unwidened): Also handle ! for_type case for
33756 * convert.c (do_narrow): Split out from ...
33757 (convert_to_integer_1): ... here. Do not pass final truncation
33758 type to get_unwidened for TRUNC_DIV_EXPR.
33760 2017-04-07 Richard Biener <rguenther@suse.de>
33762 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33764 (aff_combination_const): Adjust.
33765 (aff_combination_scale): Likewise.
33766 (aff_combination_add_elt): Likewise.
33767 (aff_combination_add_cst): Likewise.
33768 (aff_combination_convert): Likewise.
33769 (add_elt_to_tree): Likewise. Remove unused argument.
33770 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33772 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33774 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33776 * config/arm/arm.c (arm_default_short_enums): Use
33777 ARM_DEFAULT_SHORT_ENUMS.
33778 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33780 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33783 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33784 members with redundant out-of-class redeclaration.
33786 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33789 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33790 * config/i386/i386.md (*zero_extendsidi2):
33791 Add (?*x,*x) and (?*v,*v) alternatives.
33793 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33796 * config/i386/i386.c (ix86_expand_builtin)
33797 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33798 mode from insn data. Convert operands to insn operand mode.
33799 Copy operands that don't satisfy insn predicate to a register.
33801 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33803 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33806 2017-04-06 Richard Biener <rguenther@suse.de>
33808 PR tree-optimization/80334
33809 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33810 preserve alignment of accesses.
33812 2017-04-06 Richard Biener <rguenther@suse.de>
33814 PR tree-optimization/80262
33815 * tree-sra.c (build_ref_for_offset): Preserve address-space
33817 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33818 Drop useless address-space information on MEM_REF offsets.
33820 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33822 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33824 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33826 PR rtl-optimization/70703
33827 * ira-color.c (update_conflict_hard_regno_costs): Use
33828 HOST_WIDE_INT instead of long.
33830 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33833 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33834 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33836 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33837 for x86_64 target. Handle -m3dnowa option.
33839 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33841 PR rtl-optimization/70703
33842 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33843 (update_conflict_hard_regno_costs): Use long instead of unsigned
33844 arithmetic for cost calculation.
33846 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33847 Bernd Edlinger <bernd.edlinger@hotmail.de>
33850 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33853 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33856 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33857 ptr_mode with Pmode throughout.
33858 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33859 into probe_stack_range and use DImode.
33861 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33864 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33865 call_eh_return is true.
33867 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33869 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33870 Initialize last_match_fntype_index.
33872 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33875 * tree-nvr.c: Include internal-fn.h.
33876 (pass_return_slot::execute): Ignore internal calls without
33879 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33880 Richard Biener <rguenther@suse.de>
33883 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33884 captures used multiple times, except for the last use.
33885 * generic-match-head.c: Include gimplify.h.
33887 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33889 PR tree-optimization/79390
33890 * target.h (struct noce_if_info): Declare.
33891 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33892 * target.def (noce_conversion_profitable_p): New target hook.
33893 * ifcvt.h (struct noce_if_info): New type, moved from ...
33894 * ifcvt.c (struct noce_if_info): ... here.
33895 (noce_conversion_profitable_p): Renamed to ...
33896 (default_noce_conversion_profitable_p): ... this. No longer
33898 (noce_try_store_flag_constants, noce_try_addcc,
33899 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33900 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33901 instead of noce_conversion_profitable_p.
33902 * config/i386/i386.c: Include ifcvt.h.
33903 (ix86_option_override_internal): Don't override
33904 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33905 (ix86_noce_conversion_profitable_p): New function.
33906 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33907 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33908 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33909 * doc/tm.texi: Regenerated.
33911 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33913 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33916 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33919 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33920 instructions for small multiply cores.
33922 2017-04-04 Jeff Law <law@redhat.com>
33924 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33926 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33927 that are not set by the loop over the elements.
33929 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33932 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33933 int mode, convert_modes it to mode as unsigned, otherwise use
33934 lowpart_subreg to mode rather than SImode.
33935 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33936 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33937 Use DImode instead of SImode for the shift count operand.
33938 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33941 2017-04-04 Richard Biener <rguenther@suse.de>
33943 PR middle-end/80281
33944 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33945 arithmetic done for the negate or the plus. Simplify.
33946 (A - (-B) -> A + B): Likewise.
33947 * fold-const.c (split_tree): Make sure to not negate pointers.
33949 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33951 PR rtl-optimization/60818
33952 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33953 a compare of comparisons with the thing compared if this results
33954 in a different machine mode.
33956 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33958 * alias.c (base_alias_check): Fix typo in comment.
33959 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33960 * cgraphunit.c (symbol_table::compile): Likewise.
33961 * collect2.c (maybe_run_lto_and_relink): Likewise.
33962 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33963 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33964 * config/avr/avr.c (avr_map_op_t): Likewise.
33965 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33966 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33967 * config/epiphany/epiphany.md (movcc): Likewise.
33968 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33969 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33971 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33972 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33973 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33974 * config/sh/sh.c (sh_rtx_costs): Likewise.
33975 * fold-const.c (fold_truth_andor): Likewise.
33976 * genautomata.c (collapse_flag): Likewise.
33977 * gengtype.h (struct type::u::s): Likewise.
33978 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33979 * input.c (FORMAT_AMOUNT): Likewise.
33980 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33981 (known_aggs_to_agg_replacement_list): Likewise.
33982 * ipa-inline-analysis.c: Likewise.
33983 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33984 * ipa-polymorphic-call.c
33985 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33986 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33987 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33989 * modulo-sched.c (apply_reg_moves): Likewise.
33990 * omp-expand.c (build_omp_regions_1): Likewise.
33991 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33992 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33993 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33994 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33995 * value-prof.c: Likewise.
33996 * var-tracking.c (val_reset): Likewise.
33998 2017-04-03 Richard Biener <rguenther@suse.de>
34000 PR tree-optimization/80275
34001 * fold-const.c (split_address_to_core_and_offset): Handle
34004 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34006 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34007 descriptors is at least equal to that of functions.
34009 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34011 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34013 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34016 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34017 (mov<IMOD4:mode>): New expander.
34018 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34020 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34022 PR rtl-optimization/79405
34023 * fwprop.c (propagations_left): New variable.
34024 (forward_propagate_into): Decrement it.
34025 (fwprop_init): Initialize it.
34026 (fw_prop): If the variable has reached zero, stop propagating.
34027 (fwprop_addr): Ditto.
34029 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34032 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34033 a FUNCTION_DECL, pass it as decl instead of origin to
34036 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34038 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34041 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34044 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34045 TARGET_VSX_SMALL_INTEGER.
34047 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34049 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34050 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34052 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34054 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34055 extraction from odd-numbered MSA register.
34057 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34059 PR middle-end/80173
34060 * expmed.c (store_bit_field_1): Don't attempt to create
34061 a word subreg out of hard registers wider than word if they
34062 have HARD_REGNO_NREGS of 1 for their mode.
34064 PR middle-end/80163
34065 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34066 conversions to integer types wider than word and pointer.
34069 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34070 (rtx_equal_for_cselib_p): Pass 0 to it.
34071 * cselib.c (cselib_hasher::equal): Likewise.
34072 (rtx_equal_for_cselib_1): Add depth argument. If depth
34073 is 128, don't look up VALUE locs and punt. Increment
34074 depth in recursive calls when walking VALUE locs.
34076 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34078 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34079 (make_gcov_file_name): Use the canonical path name for generating
34081 (read_line): Fix handling of files with ascii null bytes.
34083 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34085 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34086 to initialise a vector register instead
34087 of using a const_int.
34089 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34091 PR translation/80189
34092 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34093 diagnostic messages.
34095 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34098 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34099 (dfp_diex_<mode>): Update mode of operand 1.
34100 * doc/extend.texi (dxex, dxexq): Document change to return type.
34101 (diex, diexq): Document change to argument type.
34103 2017-03-30 Martin Jambor <mjambor@suse.cz>
34106 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34107 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34108 it reflects the signature changes performed at the callee side.
34109 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34110 to cgraph_build_function_type_skip_args.
34111 (build_function_decl_skip_args): Adjust call to the above function.
34113 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34116 * config/i386/sse.md
34117 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34118 register as dest whenever it is a MEM not rtx_equal_p to the
34119 corresponding dup operand, and when forcing into reg move the
34120 reg into the memory afterwards.
34121 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34122 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34123 for the force_reg mode.
34124 (avx512vl_vextractf128<mode>): Use register as dest either
34125 always when a MEM, or when it is a MEM not rtx_equal_p to the
34126 corresponding dup operand, or even not when it is a CONST_VECTOR
34127 depending on the mode and lo vs. hi.
34128 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34130 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34131 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34132 Likewise. Require that operands[2] is even.
34133 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34134 Remove extraneous parens. Require that operands[2] is a multiple
34136 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34137 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34138 disallow memory then.
34140 2017-03-30 Richard Biener <rguenther@suse.de>
34142 PR tree-optimization/77498
34143 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34144 to non-constants over backedges.
34146 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34148 PR rtl-optimization/80233
34149 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34150 as last_combined_insn. Do not test for BARRIER_P separately.
34152 2017-03-29 Andreas Schwab <schwab@suse.de>
34155 * calls.c (prepare_call_address): Convert funexp to Pmode before
34156 copying to temp reg.
34158 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34160 PR tree-optimization/80158
34161 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34162 Handle possible future case of more than one alternate
34164 (replace_rhs_if_not_dup): Likewise.
34165 (replace_one_candidate): Likewise.
34167 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34169 PR rtl-optimization/80193
34170 * ira.c (ira): Do not check allocation for LRA.
34172 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34174 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34175 (nvptx_output_simt_exit): Declare.
34176 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34177 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34178 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34179 (nvptx_declare_function_name): ...here. Emit declaration of local
34180 memory space buffer for omp_simt_enter insn.
34181 (nvptx_output_unisimt_switch): New.
34182 (nvptx_output_softstack_switch): New.
34183 (nvptx_output_simt_enter): New.
34184 (nvptx_output_simt_exit): New.
34185 * config/nvptx/nvptx.h (struct machine_function): New fields
34186 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34187 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34188 (UNSPECV_SIMT_EXIT): Ditto.
34189 (omp_simt_enter_insn): New insn.
34190 (omp_simt_enter): New expansion.
34191 (omp_simt_exit): New insn.
34192 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34194 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34195 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34196 (expand_GOMP_SIMT_EXIT): New.
34197 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34198 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34199 (GOMP_SIMT_EXIT): Ditto.
34200 * target-insns.def (omp_simt_enter): New insn.
34201 (omp_simt_exit): Ditto.
34202 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34204 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34205 (lower_rec_input_clauses): Likewise.
34206 (lower_lastprivate_clauses): Handle SIMT privatization.
34208 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34209 (ompdevlow_adjust_simt_enter): New.
34210 (find_simtpriv_var_op): New.
34211 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34212 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34214 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34215 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34216 (copy_decl_for_dup_finish): Ditto.
34218 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34220 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34223 * config/i386/i386.c (ix86_option_override_internal): Always
34224 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34226 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34228 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34230 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34232 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34233 mark new edge's irreducible flag accordign to it.
34234 (vect_do_peeling): Check loop preheader edge's irreducible flag
34235 and pass it to function slpeel_add_loop_guard.
34237 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34239 PR tree-optimization/80218
34240 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34241 Update block frequencies and counts.
34243 2017-03-28 Richard Biener <rguenther@suse.de>
34245 PR tree-optimization/78644
34246 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34247 of a simplification result we may not use it at all.
34249 2017-03-28 Richard Biener <rguenther@suse.de>
34252 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34253 without arguments, generate default definition of a SSA name.
34255 2017-03-28 Richard Biener <rguenther@suse.de>
34257 PR middle-end/80222
34258 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34259 TYPE_REF_CAN_ALIAS_ALL references.
34260 * fold-const.c (fold_indirect_ref_1): Likewise.
34262 2017-03-28 Martin Liska <mliska@suse.cz>
34265 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34266 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34268 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34269 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34271 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34272 (EXTRA_SPECS): Define.
34273 (SUBTARGET_EXTRA_SPECS): Likewise.
34274 (SUBTARGET_CPP_SPEC): Likewise.
34275 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34276 SUBTARGET_EXTRA_SPECS.
34277 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34279 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34281 * config/arc/simdext.md (vst64_insn): Update pattern.
34282 (vld32wh_insn): Likewise.
34283 (vld32wl_insn): Likewise.
34284 (vld64_insn): Likewise.
34285 (vld32_insn): Likewise.
34287 2017-03-28 Marek Polacek <polacek@redhat.com>
34290 * fold-const.c (fold_comparison): Use protected_set_expr_location
34291 instead of SET_EXPR_LOCATION.
34293 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34295 * tree.c (add_expr): Avoid name lookup warning.
34297 2017-03-27 Jeff Law <law@redhat.com>
34299 PR tree-optimization/80216
34300 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34301 function name. Limit recursion depth.
34302 (record_temporary_equivalences): Corresponding changes.
34304 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34306 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34309 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34312 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34314 * cfgcleanup.c (reg_note_cfa_p): New array.
34315 (insns_have_identical_cfa_notes): New function.
34316 (old_insns_match_p): Don't cross-jump in between /f
34317 and non-/f instructions. If both i1 and i2 are frame related,
34318 verify all CFA notes, their order and content.
34320 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34323 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34324 HImode and SImode with zero extend to DImode to one insn.
34325 (bswap<mode>2_extenddi): Likewise.
34326 (bswapsi2_extenddi): Likewise.
34327 (bswaphi2_extendsi): Likewise.
34328 (bswaphi2): Combine bswap HImode and SImode into one insn.
34329 Separate memory insns from swapping register.
34330 (bswapsi2): Likewise.
34331 (bswap<mode>2): Likewise.
34332 (bswaphi2_internal): Delete, no longer used.
34333 (bswapsi2_internal): Likewise.
34334 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34335 store, and gpr<-gpr swap insns.
34336 (bswap<mode>2_store): Likewise.
34337 (bswaphi2_reg): Register only splitter, combine with the splitter.
34338 (bswaphi2 splitter): Likewise.
34339 (bswapsi2_reg): Likewise.
34340 (bswapsi2 splitter): Likewise.
34341 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34342 the insns into load, store, and register/register insns.
34343 (bswapdi2_ldbrx): Likewise.
34344 (bswapdi2_load): Likewise.
34345 (bswapdi2_store): Likewise.
34346 (bswapdi2_reg): Likewise.
34348 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34350 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34351 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34353 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34356 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34358 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34359 special handling for target option conflicts between dform
34360 options (-mpower9-dform, -mpower9-dform-vector,
34361 -mpower9-dform-scalar) and -mno-direct-move.
34363 2017-03-27 Richard Biener <rguenther@suse.de>
34365 PR tree-optimization/80181
34366 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34368 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34370 * config/arc/predicates.md (move_double_src_operand): Replace the
34371 call to move_double_src_operand with a call to address_operand.
34373 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34375 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34376 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34377 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34379 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34381 * config/arc/predicates.md (long_immediate_loadstore_operand):
34382 Consider scaled addresses cases.
34384 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34386 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34387 restored when in interrupt.
34388 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34389 doesn't have delay slot.
34391 2017-03-27 Richard Biener <rguenther@suse.de>
34394 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34395 inlined thunk clones.
34397 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34400 * asan.c (instrument_derefs): Copy over last operand from
34401 original COMPONENT_REF to the new COMPONENT_REF with
34402 DECL_BIT_FIELD_REPRESENTATIVE.
34403 * ubsan.c (instrument_object_size): Likewise.
34405 2017-03-27 Richard Biener <rguenther@suse.de>
34407 PR tree-optimization/80170
34408 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34409 sure DR/SCEV didnt fold in constants we do not see when looking
34410 at the reference base alignment.
34412 2017-03-27 Richard Biener <rguenther@suse.de>
34414 PR middle-end/80171
34415 * gimple-fold.c (fold_ctor_reference): Properly guard against
34416 NULL return value from canonicalize_constructor_val.
34418 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34421 * config/i386/i386.c (ix86_expand_builtin)
34422 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34423 flags reg setting and flags reg using instructions.
34424 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34425 clobbering instructions to zero extend op2.
34427 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34429 * doc/install.texi (Configuration) <--with-aix-soname>:
34430 Update link to AIX ld.
34432 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34434 PR rtl-optimization/80160
34435 PR rtl-optimization/80159
34436 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34437 reg_alternate_class into account.
34439 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34442 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34443 to consider in curr_insn_transform.
34445 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34447 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34448 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34449 and emit_mode_inner.
34451 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34453 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34454 argument to the overloaded builtin variants. Use the new flag to
34455 deprecate certain builtin variants.
34456 * config/s390/s390-builtin-types.def: Add new builtin types.
34457 * config/s390/s390-builtins.h: Support new flags field for
34458 overloaded builtins.
34459 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34460 (s390_macro_to_expand): Enable vector float data type.
34461 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34462 builtins by incrementing the __VEC__ version number.
34463 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34465 (s390_resolve_overloaded_builtin): Emit error messages depending
34466 on the builtin flags.
34467 * config/s390/s390.c (s390_expand_builtin): Support additional
34468 flags argument. Change error message to match the messages
34469 emitted in s390-c.c.
34470 * config/s390/s390.md: New UNSPEC_* constants.
34471 (op_type): Add new instruction types.
34472 * config/s390/vecintrin.h: Add new builtins and test data class
34474 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34475 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34476 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34477 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34478 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34479 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34481 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34482 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34483 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34484 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34486 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34487 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34488 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34489 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34491 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34492 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34493 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34494 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34495 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34496 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34497 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34499 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34500 ("vec_scatter_element<V_HW_4:mode>_DI")
34501 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34502 ("vec_fpint<mode>", "vflls")
34503 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34504 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34505 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34506 ("*vec_cmphe<mode>_cc"): ... these.
34508 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34509 mode constant instead of magic value.
34511 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34513 * config/s390/s390.c (s390_expand_vec_compare): Support other
34514 vector floating point modes than just V2DF.
34515 (s390_expand_vcond): Likewise.
34516 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34517 (s390_cannot_change_mode_class): Prevent mode changes between TF
34518 and V1TF in vector registers.
34519 * config/s390/s390.md (DF, SF): New mode attributes.
34520 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34521 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34522 SFmode support for VRs.
34523 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34525 (VFT, VF_HW): New mode iterators.
34526 (vw, sdx): New mode attributes.
34527 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34528 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34529 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34530 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34531 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34532 also the new vector floating point modes. Renaming to ...
34534 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34535 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34536 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34537 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34538 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34539 ("vec_unordered<mode>"): ... these.
34541 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34542 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34543 ("*vec_extendv2df"): New insn definitions.
34545 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34547 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34548 ("mulditi3_2", "*muldi3_sign"): New patterns.
34549 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34550 rename the pattern definition.
34552 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34554 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34556 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34558 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34560 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34561 instruction if possible.
34562 * config/s390/vector.md (vec_halfnumelts): New mode
34564 ("*vec_vllezlf<mode>"): New pattern.
34566 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34568 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34569 ("popcountv4si2", "popcountv2di2"): Rename to ...
34570 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34571 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34573 ("popcount<mode>2_vxe"): New pattern.
34575 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34577 * common/config/s390/s390-common.c (processor_flags_table): Add
34579 * config.gcc: Add arch12.
34580 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34581 Default to arch12 for unknown CPU model numbers.
34582 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34583 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34584 PROCESSOR_max sanity check.
34585 * config/s390/s390-opts.h (enum processor_type): Add
34587 * config/s390/s390.c (processor_table): Add arch12.
34588 (s390_expand_builtin): Add check for B_VXE flag.
34589 (s390_issue_rate): Add PROCESSOR_ARCH12.
34590 (s390_get_sched_attrmask): Likewise.
34591 (s390_get_unit_mask): Likewise.
34592 (s390_sched_score): Enable z13 scheduling for arch12.
34593 (s390_sched_reorder): Likewise.
34594 (s390_sched_variable_issue): Likewise.
34595 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34597 (s390_tune_attr): Use z13 scheduling also for arch12.
34598 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34599 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34600 (TARGET_VXE_P): New macros.
34601 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34602 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34603 * config/s390/s390.opt: Add arch12 as processor_type.
34605 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34607 * config/s390/s390.md
34608 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34609 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34610 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34612 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34613 Rename expanders to ...
34615 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34616 ("fixuns_truncdddi2_emu"): ... these.
34618 ("fixuns_trunc<mode>si2_emu"): New expander.
34620 ("*fixuns_truncdfdi2_z13"): Rename to ...
34621 ("*fixuns_truncdfdi2_vx"): ... this.
34623 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34625 * config/s390/2964.md: Remove the single element vector compare
34626 instructions which are no longer used.
34627 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34629 (s390_canonicalize_comparison): Remove handling of DFmode
34631 (s390_expand_vec_compare_scalar): Remove function.
34632 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34633 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34635 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34637 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34639 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34641 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34642 will anyway by matched by mov<mode>_64dfp.
34644 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34646 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34647 vlef/vstef. Add missing operand to vleif.
34649 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34651 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34652 pair for all vector types with 64 bit elements.
34653 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34654 * config/s390/vector.md (V_HW_64): ... here.
34655 (V_128_NOSINGLE): New mode iterator.
34656 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34657 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34658 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34659 ("*vec_load_pairv2di"): Change to ...
34660 ("*vec_load_pair<mode>"): ... this one.
34662 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34664 * config/s390/constraints.md: Add comments.
34665 (jKK): Reject element sizes > 8 bytes.
34666 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34668 * config/s390/s390.md: Add the s_operand checks formerly in
34669 s390_split_ok_p to various splitters where they are still
34671 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34672 for 128 bit vectors. Plus two splitters.
34674 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34676 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34679 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34682 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34683 error if the boundary argument is not constant.
34685 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34687 PR rtl-optimization/80112
34688 * loop-doloop.c (doloop_condition_get): Don't check condition
34689 if cmp isn't SET with IF_THEN_ELSE src.
34691 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34693 PR tree-optimization/80158
34694 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34695 replacing a candidate statement, also replace it for the
34696 candidate's alternate interpretation.
34697 (replace_rhs_if_not_dup): Likewise.
34698 (replace_one_candidate): Likewise.
34700 2017-03-24 Richard Biener <rguenther@suse.de>
34702 PR tree-optimization/80167
34703 * graphite-isl-ast-to-gimple.c
34704 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34706 (translate_isl_ast_to_gimple::get_rename): Likewise.
34708 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34710 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34711 handling of certain combinations of target options, including the
34712 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34713 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34715 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34718 * config/arm/arm.md (*load_multiple): Add reload_completed to
34719 matching condition.
34721 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34722 Richard Biener <rguenth@suse.de>
34724 PR tree-optimization/79908
34725 PR tree-optimization/80136
34726 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34727 been cast away, gimplify_and_add suffices.
34729 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34731 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34733 2017-03-23 Richard Biener <rguenther@suse.de>
34735 PR tree-optimization/80032
34736 * gimplify.c (gimple_push_cleanup): Forced unconditional
34737 cleanups still have to go to the conditional_cleanups
34740 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34742 PR tree-optimization/80072
34743 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34745 (next_operand_entry_id): Change type to unsigned int.
34746 (sort_by_operand_rank): Make sure to return the right return value
34747 even if unsigned fields are bigger than INT_MAX.
34748 (struct oecount): Change cnt and id type to unsigned int.
34749 (oecount_hasher::equal): Formatting fix.
34750 (oecount_cmp): Make sure to return the right return value
34751 even if unsigned fields are bigger than INT_MAX.
34752 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34755 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34756 TREE_READONLY on result if writing it more than once.
34759 * doc/invoke.texi (-fsanitize=thread): Document that with
34760 -fnon-call-exceptions atomics are not able to throw
34764 * tsan.c: Include tree-eh.h.
34765 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34766 maybe_clean_or_replace_eh_stmt where needed.
34767 (instrument_memory_accesses): Add cfg_changed argument.
34768 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34769 if it returned true.
34770 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34772 PR rtl-optimization/63191
34773 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34774 wrapper function, moved the whole old content into ...
34775 (ix86_delegitimize_address_1): ... this. New inline function.
34776 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34777 true as last argument instead of ix86_delegitimize_address.
34779 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34781 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34782 cortexa57_branch_cost.
34784 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34786 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34788 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34791 * doc/md.texi (Constraints): Document wA constraint.
34792 * config/rs6000/constraints.md (wA): New.
34793 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34794 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34795 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34796 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34798 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34801 * gimplify.c (is_oacc_declared): New function.
34802 (oacc_default_clause): Use it to set default flags for acc declared
34803 variables inside parallel regions.
34804 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34805 declared variables.
34806 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34807 declare attribute to any decl as necessary.
34809 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34812 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34813 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34814 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34815 (arm_arch_lpae): This.
34816 * config/arm/arm.c (arm_arch7ve): Rename into ...
34817 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34818 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34821 2017-03-22 Martin Liska <mliska@suse.cz>
34824 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34825 error message instead of an ICE.
34827 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34829 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34831 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34833 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34835 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34838 2017-03-21 Martin Sebor <msebor@redhat.com>
34840 * doc/extend.texi: Use "cannot" instead of "can't."
34841 * doc/hostconfig.texi: Same.
34842 * doc/install.texi: Same.
34843 * doc/invoke.texi: Same.
34844 * doc/loop.texi: Same.
34845 * doc/md.texi: Same.
34846 * doc/objc.texi: Same.
34847 * doc/rtl.texi: Same.
34848 * doc/tm.texi: Same.
34849 * doc/tm.texi.in: Same.
34850 * doc/trouble.texi: Same.
34852 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34855 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34856 (collect_checksum_attributes): Set it.
34857 (die_checksum_ordered): Use it.
34859 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34861 PR tree-optimization/79908
34862 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34863 change: For a VA_ARG whose LHS has been cast away, use
34864 force_gimple_operand to construct the side effects.
34866 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34868 PR translation/80001
34869 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34870 more amenable to translation.
34871 (oacc_loop_auto_partitions): Likewise.
34873 2017-03-21 Marek Polacek <polacek@redhat.com>
34874 Martin Sebor <msebor@redhat.com>
34876 PR tree-optimization/80109
34877 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34878 on INTEGRAL_TYPE_P.
34880 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34881 Segher Boessenkool <segher@kernel.crashing.org>
34884 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34885 check reg_used_between_p between insn and one of succ or succ2
34886 depending on if succ is artificial insn not inserted into insn
34889 2017-03-21 Martin Liska <mliska@suse.cz>
34891 PR gcov-profile/80081
34892 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34893 * doc/gcc.texi: Include gcov-dump stuff.
34894 * doc/gcov-dump.texi: New file.
34896 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34898 PR rtl-optimization/79150
34899 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34900 conditional jump, if the jump is the last insn of the loop.
34902 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34903 Richard Biener <rguenth@suse.de>
34905 PR tree-optimization/79908
34906 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34907 been cast away, use force_gimple_operand to construct the side
34910 2017-03-21 Martin Liska <mliska@suse.cz>
34912 PR libfortran/79956
34913 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34916 2017-03-21 Brad Spengler <spender@grsecurity.net>
34919 * plugin.c (htab_hash_plugin): New function.
34920 (add_new_plugin): Use it and adjust.
34921 (parse_plugin_arg_opt): Adjust.
34922 (init_one_plugin): Likewise.
34924 2017-03-21 Richard Biener <rguenther@suse.de>
34926 PR tree-optimization/80032
34927 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34928 if set force the cleanup to happen unconditionally.
34929 (gimplify_target_expr): Push inserted clobbers with force_uncond
34930 to avoid them being removed by control-dependent DCE.
34932 2017-03-21 Richard Biener <rguenther@suse.de>
34934 PR tree-optimization/80122
34935 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34936 va_arg_pack_len when the inlined call stmt requires pack
34938 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34940 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34943 * tsan.c (instrument_builtin_call): If the memory model argument
34944 is not a constant, assume it is valid.
34947 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34950 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34952 PR rtl-optimization/79910
34953 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34954 if its dest is used by an insn before I2 (other than the combined
34955 insns themselves, which are properly handled already).
34957 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34960 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34962 * combine.c (record_used_regs): New static function.
34963 (try_combine): Handle situations where there is an additional
34964 instruction between I2 and I3 which needs to have a LOG_LINK
34968 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34970 * combine.c (try_combine): Delete redundant i1 test. Call
34971 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34973 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34976 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34977 alternatives 13/14.
34979 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34981 PR tree-optimization/80054
34982 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34983 the optimization if a PHI or any of its arguments is not dominated
34984 by the candidate's basis. Use gphi* rather than gimple* as
34986 (replace_profitable_candidates): Clean up a gimple* variable that
34987 should be a gphi* variable.
34989 2017-03-20 Martin Sebor <msebor@redhat.com>
34992 * doc/extend.texi (attribute constructor): Document present limitation.
34994 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34997 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34998 __POWER9_VECTOR__ #ifdef control, change template definition to
34999 use Power9-specific built-in function.
35000 (vec_any_eq): Likewise.
35001 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35002 to control outcomes from this test.
35003 (vector_ae_<mode>p): For VEC_F modes, likewise.
35005 2017-03-20 Ian Lance Taylor <iant@google.com>
35007 * config/i386/i386.c (ix86_function_regparm): Save an extra
35008 register for -fsplit-stack with DECL_STATIC_CHAIN.
35010 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35013 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35014 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35016 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35018 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35020 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35023 2017-03-20 Marek Polacek <polacek@redhat.com>
35026 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35028 2017-03-20 Richard Biener <rguenther@suse.de>
35030 PR tree-optimization/80113
35031 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35032 allocate extra SSA name for PHI def.
35033 (add_close_phis_to_outer_loops): Likewise.
35034 (add_close_phis_to_merge_points): Likewise.
35035 (copy_loop_close_phi_args): Likewise.
35036 (copy_cond_phi_nodes): Likewise.
35038 2017-03-20 Martin Liska <mliska@suse.cz>
35040 PR middle-end/79753
35041 * tree-chkp.c (chkp_build_returned_bound): Do not build
35042 returned bounds for a LHS that's not a BOUNDED_P type.
35044 2017-03-20 Martin Liska <mliska@suse.cz>
35048 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35049 COMPLEX_CST and VECTOR_CST.
35051 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35054 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35055 target operand. A new splitter adds the clobber statement in case
35056 the target operand is dead anyway.
35058 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35060 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35061 to age-old versions of binutils and glibc.
35063 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35065 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35067 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35069 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35071 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35073 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35074 requirement for binutils 2.13.
35076 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35078 * combine.c (try_combine): Delete redundant i1 test. Call
35079 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35081 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35083 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35084 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35086 <riscv64-*-elf>: Re-arrange section
35087 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35088 <riscv32-*-linux>: Likewise.
35089 <riscv64-*-elf>: Likewise
35090 <riscv64-*-linux>: Likewise.
35092 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35095 * aarch64.opt(verbose-cost-dump): Fix typo.
35097 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35100 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35101 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35103 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35105 * reload.c (find_reloads): When reloading a nonoffsettable address,
35106 use RELOAD_OTHER for it and its address reloads.
35108 PR rtl-optimization/79910
35109 * combine.c (record_used_regs): New static function.
35110 (try_combine): Handle situations where there is an additional
35111 instruction between I2 and I3 which needs to have a LOG_LINK
35114 2017-03-17 Jeff Law <law@redhat.com>
35116 PR tree-optimization/71437
35117 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35118 conditional in the hash table first.
35119 (vrp_dom_walker::before_dom_children): Extract condition from
35120 ASSERT_EXPR. Record condition, its inverion and any implied
35121 conditions as well.
35123 2017-03-17 Marek Polacek <polacek@redhat.com>
35124 Markus Trippelsdorf <markus@trippelsdorf.de>
35126 PR tree-optimization/80079
35127 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35130 2017-03-17 Richard Biener <rguenther@suse.de>
35132 PR middle-end/80075
35133 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35134 Properly verify the LHS before the RHS possibly claims to be
35136 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35139 2017-03-17 Martin Jambor <mjambor@suse.cz>
35141 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35142 (List of -O2 options): Likewise.
35143 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35146 2017-03-17 Tom de Vries <tom@codesourcery.com>
35148 * gcov-dump.c (print_usage): Print bug_report_url.
35150 2017-03-17 Richard Biener <rguenther@suse.de>
35152 PR middle-end/80050
35153 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35154 (parser::peek): Likewise.
35156 2017-03-17 Richard Biener <rguenther@suse.de>
35158 PR tree-optimization/80048
35159 * sese.c (free_sese_info): Properly release rename_map and
35160 copied_bb_map elements.
35162 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35164 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35165 Add linked-list forward and backlinks. Insert on
35166 construction, remove on destruction.
35167 (class pass_store_merging): Add m_stores_head field.
35168 (pass_store_merging::terminate_and_process_all_chains):
35169 Iterate over m_stores_head list.
35170 (pass_store_merging::terminate_all_aliasing_chains):
35172 (pass_store_merging::execute): Check for debug stmts first.
35173 Push new chains onto the m_stores_head stack.
35175 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35178 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35179 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35180 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35182 2017-03-16 Jeff Law <law@redhat.com>
35184 PR tree-optimization/71437
35185 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35186 member function. Implementation moved into after_dom_children
35187 member function and into the threader's thread_outgoing_edges
35189 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35190 some code into new thread_outgoing_edges.
35191 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35192 definition. Simplify marker handling (do it here). Assume we always
35193 have the available expression and the const/copies tables.
35194 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35196 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35197 * tree-vrp.c (equiv_stack): No longer file scoped.
35198 (vrp_dom_walker): New class.
35199 (vrp_dom_walker::before_dom_children): New member function.
35200 (vrp_dom_walker::after_dom_children): Likewise.
35201 (identify_jump_threads): Setup domwalker. Use it rather than
35202 walking edges in a random order by hand. Simplify setup/finalization.
35203 (finalize_jump_threads): Remove.
35204 (vrp_finalize): Do not call identify_jump_threads here.
35205 (execute_vrp): Do it here instead and call thread_through_all_blocks
35208 PR tree-optimization/71437
35209 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35211 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35213 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35214 (dom_opt_dom_walker::thread_across_edge): Remove
35215 handle_dominating_asserts argument. All callers changed.
35216 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35217 changes. Remove calls to lhs_of_dominating_assert. Other
35218 uses of handle_dominating_asserts turn into unconditional code
35219 (simplify_control_stmt_condition_1): Likewise.
35220 (simplify_control_stmt_condition): Likewise.
35221 (thread_through_normal_block, thread_across_edge): Likewise.
35222 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35223 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35224 object if it is not an SSA_NAME.
35225 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35226 before calling into the VRP specific simplifiers.
35227 (identify_jump_threads): Remove handle_dominating_asserts
35230 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35233 * tree-diagnostic.c (default_tree_printer): No longer static.
35234 * tree-diagnostic.h (default_tree_printer): New prototype.
35236 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35238 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35239 Change ins into fmov.
35241 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35243 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35244 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35245 Use h_con constraint for operand 1.
35246 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35247 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35249 2017-03-15 Jeff Law <law@redhat.com>
35251 PR tree-optimization/71437
35252 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35253 (record_temporary_equivalences): Use it.
35255 PR tree-optimization/71437
35256 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35257 tree-ssa-scopedtables.
35258 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35259 (record_conditions, record_cond, vuse_eq): Likewise.
35260 (record_edge_info): Adjust to API tweak of record_conditions.
35261 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35262 (record_temporary_equivalences, optimize_stmt): Likewise.
35263 (eliminate_redundant_computations): Likewise.
35264 (record_equivalences_from_stmt): Likewise.
35265 * tree-ssa-scopedtables.c: Include options.h and params.h.
35266 (vuse_eq): New function, moved from tree-ssa-dom.c
35267 (build_and_record_new_cond): Likewise.
35268 (record_conditions): Likewise. Accept vector of conditions rather
35269 than edge_equivalence structure for first argument.
35270 for the first argument.
35271 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35272 from tree-ssa-dom.c.
35273 (avail_exprs_stack::record_cond): Likewise.
35274 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35275 from tree-ssa-dom.c.
35276 (avail_exprs_stack): Add new member functions lookup_avail_expr
35278 (record_conditions): Declare.
35280 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35283 * lra-constraints.c (process_alt_operands): Increase reject for
35284 reloading an input/output operand.
35286 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35289 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35290 insns to convert from signed/unsigned char/short to IEEE 128-bit
35292 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35294 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35297 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35298 subreg of inner mode for values already in registers.
35300 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35302 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35303 iteration reg is used after the loop.
35305 2017-03-14 Martin Sebor <msebor@redhat.com>
35307 PR tree-optimization/79800
35308 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35309 precision in negative-positive range.
35310 (format_floating): Call non-const overload with adjusted precision.
35312 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35315 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35318 2017-03-14 Martin Sebor <msebor@redhat.com>
35320 PR middle-end/80020
35321 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35322 * builtins.def (aligned_alloc): Use it.
35325 * Makefile.in (GTFILES): Add calls.c.
35326 * calls.c: Include "gt-calls.h".
35328 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35330 PR rtl-optimization/79728
35331 * regs.h (struct target_regs): New field
35332 x_contains_allocatable_regs_of_mode.
35333 (contains_allocatable_regs_of_mode): New macro.
35334 * reginfo.c (init_reg_sets_1): Initialize it, and change
35335 contains_reg_of_mode so it includes global regs as well.
35336 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35337 rather than contains_regs_of_mode.
35339 2017-03-14 Martin Liska <mliska@suse.cz>
35341 * doc/invoke.texi: Document options that can't be combined with
35342 -fcheck-pointer-bounds.
35344 2017-03-14 Martin Liska <mliska@suse.cz>
35346 PR middle-end/79831
35347 * doc/invoke.texi (-Wchkp): Document the option.
35349 2017-03-14 Martin Liska <mliska@suse.cz>
35351 * Makefile.in: Install gcov-dump.
35353 2017-03-14 Martin Liska <mliska@suse.cz>
35355 * multiple_target.c (expand_target_clones): Bail out for
35356 an invalid attribute.
35358 2017-03-14 Richard Biener <rguenther@suse.de>
35360 * alias.c (struct alias_set_entry): Pack properly.
35361 * cfgloop.h (struct loop): Likewise.
35362 * cse.c (struct set): Likewise.
35363 * ipa-utils.c (struct searchc_env): Likewise.
35364 * loop-invariant.c (struct invariant): Likewise.
35365 * lra-remat.c (struct cand): Likewise.
35366 * recog.c (struct change_t): Likewise.
35367 * rtl.h (struct address_info): Likewise.
35368 * symbol-summary.h (function_summary): Likewise.
35369 * tree-loop-distribution.c (struct partition): Likewise.
35370 * tree-object-size.c (struct object_size_info): Likewise.
35371 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35372 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35373 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35374 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35375 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35376 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35377 (struct _stmt_vec_info): Likewise.
35379 2017-03-14 Martin Liska <mliska@suse.cz>
35382 * multiple_target.c (create_dispatcher_calls): Check that
35383 a target can create a function dispatcher.
35385 2017-03-14 Martin Liska <mliska@suse.cz>
35388 * multiple_target.c (expand_target_clones): Drop local.local
35389 flag for default implementation.
35391 2017-03-14 Richard Biener <rguenther@suse.de>
35393 PR tree-optimization/80030
35394 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35396 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35398 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35399 gcc_fallthrough() instead of __attribute__((fallthrough));
35401 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35403 * doc/gcc.texi: Remove "up" link to (DIR).
35404 * doc/gccint.texi: Ditto.
35406 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35408 * doc/install.texi (Specific) <avr>: Remove reference to
35411 2017-03-13 Jeff Law <law@redhat.com>
35413 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35414 attribute rather than comments.
35416 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35417 match_scratch operand is highest.
35419 2017-03-13 Martin Liska <mliska@suse.cz>
35421 PR middle-end/78339
35422 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35423 is a CHKP clone, use original declaration.
35425 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35427 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35428 (arc_conditional_register_usage): Use a different allocation order
35429 when optimizing for size.
35430 * common/config/arc/arc-common.c (arc_option_optimization_table):
35431 Section anchors default on when optimizing for size.
35433 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35435 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35437 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35439 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35440 * config/arc/arc.md (cpu_facility): Add cd variant.
35441 (*movqi_insn): Add code density variant.
35442 (*movhi_insn): Likewise.
35443 (*movqi_insn): Likewise.
35444 (*addsi3_mixed): Likewise.
35445 (subsi3_insn): Likewise.
35447 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35449 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35451 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35453 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35454 expressions with MINUS and UNARY ops.
35456 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35459 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35461 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35462 between vec_select and vector argument.
35463 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35464 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35465 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35466 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35467 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35468 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35470 2017-03-13 Richard Biener <rguenther@suse.de>
35473 * params.def (vect-max-peeling-for-alignment): Fix typo.
35475 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35477 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35478 issue that only occurred with binutils below 2.18.
35480 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35482 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35483 refer to binutils 2.11/2.12 minimum.
35485 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35487 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35488 ftp.kernel.org and simplify binutils requirement.
35490 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35492 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35494 (Optimize Options): Ditto. Also remove redundancy.
35496 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35498 PR translation/79848
35499 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35501 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35502 to G_ to avoid double translation.
35504 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35506 PR translation/79923
35507 * auto-profile.c (get_combined_location): Convert leading
35508 character of diagnostics to lower case and remove trailing period.
35509 (read_profile): Likewise for various diagnostics.
35510 * config/arm/arm.c (arm_option_override): Remove trailing period
35511 from various diagnostics.
35512 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35513 (msp430_expand_delay_cycles): Likewise.
35515 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35518 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35519 full command-line argument, rather than just "str".
35520 (aarch64_validate_march): Likewise.
35521 (aarch64_validate_mtune): Likewise.
35523 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35525 PR rtl-optimization/78911
35526 * lra-assigns.c (must_not_spill_p): New function.
35527 (spill_for): Use it.
35529 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35531 PR tree-optimization/79981
35532 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35533 ATOMIC_COMPARE_EXCHANGE ifn result.
35534 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35535 IFN_ATOMIC_COMPARE_EXCHANGE.
35537 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35540 * opts.c (parse_sanitizer_options): Add missing question mark to
35541 "did you mean" message.
35543 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35545 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35547 (VMULEUH_UNS): Likewise.
35548 (VMULOUB_UNS): Likewise.
35549 (VMULOUH_UNS): Likewise.
35550 * config/rs6000/rs6000.c (builtin_function_type): Remove
35551 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35553 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35556 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35557 x with result of extra_parsing_for_operand_code_0.
35558 (function_reader::extra_parsing_for_operand_code_0): Convert
35559 return type from void to rtx, returning x. When reading
35560 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35561 larger size containing struct block_symbol.
35563 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35565 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35566 -mfloat128-hardware without -m64.
35568 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35571 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35572 entries to the case statement that marks unsigned arguments to
35573 overloaded functions.
35575 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35577 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35578 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35580 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35583 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35584 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35586 2017-03-10 Martin Liska <mliska@suse.cz>
35590 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35591 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35594 2017-03-10 Olivier Hainque <hainque@adacore.com>
35596 * tree-switch-conversion (array_value_type): Start by resetting
35597 candidate type to it's main variant.
35599 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35601 PR rtl-optimization/79909
35602 * combine.c (try_combine): Use simplify_replace_rtx on individual
35603 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35604 of the whole CALL_INSN_FUNCTION_USAGE.
35606 PR tree-optimization/79972
35607 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35608 get_range_info on SSA_NAMEs. Formatting fixes.
35610 2017-03-10 Richard Biener <rguenther@suse.de>
35611 Jakub Jelinek <jakub@redhat.com>
35613 PR tree-optimization/77975
35614 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35615 edge to be constant.
35616 (get_val_for): For constant x return it. Formatting fix.
35617 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35618 would use the same bases as the current one.
35620 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35622 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35623 instead of vec_select for V1TImode.
35624 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35626 (VSX_LE_128): Add V1TI to this mode iterator.
35627 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35628 (*vsx_le_perm_store_<mode>): Likewise.
35629 (pre-reload splitter for VSX stores): Likewise.
35630 (post-reload splitter for VSX stores): Likewise.
35631 (*vsx_xxpermdi2_le_<mode>): Likewise.
35632 (*vsx_lxvd2x2_le_<mode>): Likewise.
35633 (*vsx_stxvd2x2_le_<mode>): Likewise.
35635 2017-03-09 Michael Eager <eager@eagercon.com>
35637 Correct failures with --enable-checking=yes,rtl.
35639 * config/microblaze/microblaze.c (microblaze_expand_shift):
35640 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35641 test for const0_rtx.
35642 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35643 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35645 2017-03-09 Richard Biener <rguenther@suse.de>
35647 PR tree-optimization/79977
35648 * graphite-scop-detection.c (scop_detection::merge_sese):
35649 Handle the case of extra exits to blocks dominating the entry.
35651 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35653 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35656 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35658 PR rtl-optimization/79949
35659 * lra-constraints.c (process_alt_operands): Check memory when
35660 trying to predict a cycle. Print about the overall increase.
35662 2017-03-09 Richard Biener <rguenther@suse.de>
35664 PR middle-end/79971
35665 * gimple-expr.c (useless_type_conversion_p): Preserve
35666 TYPE_SATURATING for fixed-point types.
35668 2017-03-09 Richard Biener <rguenther@suse.de>
35671 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35672 alignment of BLKmode params.
35674 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35677 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35678 (VALL_NO_V2Q): Likewise.
35680 * config/aarch64/aarch64-simd.md
35681 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35683 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35684 VALL_NO_V2Q mode iterator.
35685 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35687 2017-03-09 Martin Liska <mliska@suse.cz>
35689 PR tree-optimization/79631
35690 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35691 tree_int_cst_sign_bit just for INTEGER constants.
35693 2017-03-09 Martin Liska <mliska@suse.cz>
35697 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35700 2017-03-09 Marek Polacek <polacek@redhat.com>
35703 * tree.c (inchash::add_expr): Handle TREE_VEC.
35705 2017-03-09 Martin Liska <mliska@suse.cz>
35708 (chkp_narrow_size_and_offset): New function.
35709 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35710 (void chkp_parse_bit_field_ref): New function.
35711 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35712 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35714 2017-03-09 Martin Liska <mliska@suse.cz>
35717 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35718 (chkp_find_bounds_1): Remove gcc_unreachable.
35720 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35723 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35724 BUILT_IN_SYNC*, determine the access type from the size suffix and
35725 always build a MEM_REF with that type. Handle forgotten
35726 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35729 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35730 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35731 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35732 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35733 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35734 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35735 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35736 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35737 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35738 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35739 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35740 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35741 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35742 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35743 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35744 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35745 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35746 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35747 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35748 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35749 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35750 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35751 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35752 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35753 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35754 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35755 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35756 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35757 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35758 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35759 definitions outside of __OPTIMIZE__ guarded section.
35762 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35763 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35764 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35765 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35768 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35770 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35771 ("vfenez<mode>"): Add missing constraints.
35773 2017-03-08 Martin Sebor <msebor@redhat.com>
35776 * config/nds32/nds32.c (nds32_option_override):
35777 Fix misspelled diagnostic.
35779 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35782 * gimplify.c (gimplify_omp_for): Replace index var in outer
35783 taskloop statement with an artificial variable and add
35784 OMP_CLAUSE_PRIVATE clause for it.
35786 2017-03-08 Richard Biener <rguenther@suse.de>
35788 PR tree-optimization/79955
35789 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35790 for accesses that are completely outside of the variable.
35792 2017-03-08 Andrew Haley <aph@redhat.com>
35794 PR tree-optimization/79943
35795 * tree-ssa-loop-split.c (compute_new_first_bound): When
35796 calculating the new upper bound, (END-BEG) should be added, not
35799 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35801 * config/avr/avr.md (setmemhi): Make sure match_dup
35802 operand number comes before match_scratch.
35804 2017-03-08 Richard Biener <rguenther@suse.de>
35806 PR tree-optimization/79920
35807 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35808 with ncopies == 1 to ...
35809 (vect_transform_slp_perm_load): ... here. Properly compute
35810 all element loads by iterating VF times over the group. Do
35811 not handle ncopies (computed in a broken way) in
35812 vect_create_mask_and_perm.
35814 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35817 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35818 is a uniform vector, use uniform_vector_p return value instead of
35819 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35821 2017-03-07 Marek Polacek <polacek@redhat.com>
35823 PR middle-end/79809
35824 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35825 (alloca_call_type): Likewise.
35827 2017-03-07 Martin Liska <mliska@suse.cz>
35829 * gcov.c (process_args): Put comment to correct location.
35831 2017-03-07 Martin Liska <mliska@suse.cz>
35833 PR middle-end/68270
35834 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35835 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35836 (chkp_narrow_bounds_for_field): Likewise.
35837 (chkp_parse_array_and_component_ref): Pass one more argument to
35840 2017-03-07 Richard Biener <rguenther@suse.de>
35842 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35845 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35847 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35848 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35850 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35853 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35854 to end of description.
35855 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35857 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35859 PR rtl-optimization/79901
35860 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35862 (*avx512f_<code><mode>3<mask_name>): ... this.
35863 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35864 iterator instead of VI8_AVX2_AVX512BW.
35866 PR rtl-optimization/79901
35867 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35868 min/max expander, expand it using expand_vec_cond_expr.
35871 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35874 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35877 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35878 to void * for PCH reasons.
35879 * dwarf2out.c (output_loc_operands, output_die): Cast
35880 v.val_vec.array to unsigned char *.
35882 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35885 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35888 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35890 PR rtl-optimization/79571
35891 * lra-constraints.c (process_alt_operands): Calculate static
35892 reject and subtract it from overall when only addresses will be
35895 2017-03-06 Julia Koval <julia.koval@intel.com>
35898 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35899 incoming stack boundary to 128 for 64-bit targets.
35901 2017-03-06 Richard Biener <rguenther@suse.de>
35903 PR tree-optimization/79894
35904 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35905 to NULL after folding it.
35907 2017-03-06 Richard Biener <rguenther@suse.de>
35909 PR tree-optimization/79824
35910 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35911 check disabling peeling for gaps.
35913 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35915 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35916 attributes): Document gettimeofday.
35918 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35920 * config/s390/s390.c (s390_option_override_internal): Set
35921 PARAM_MIN_VECT_LOOP_BOUND
35923 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35925 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35926 * config/s390/s390.md: Likewise.
35928 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35931 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35932 (<avx2_avx512>_perm<mode>): Rename to ...
35933 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35935 (<avx512>_perm<mode>_mask): Rename to ...
35936 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35938 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35939 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35940 instead of VI8F_256_512.
35941 (avx512f_perm<mode>): New define_expand.
35942 (avx512f_perm<mode>_mask): Likewise.
35943 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35944 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35946 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35948 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35949 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35951 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35953 2017-03-06 Martin Liska <mliska@suse.cz>
35956 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35957 when having a SSA NAME w/o VAR_DECL assigned to it.
35959 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35961 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35962 msa_dpsub_<su>_d): Fix MODE for vec_select.
35964 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35966 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35968 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35970 2017-03-06 Richard Biener <rguenther@suse.de>
35972 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35973 * plugin.c (register_plugin_info): Likewise.
35974 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35976 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35978 * config/i386/sse.md (sse_storehps, sse_storelps,
35979 avx_<castmode><avxsizesuffix>_<castmode>,
35980 avx512f_<castmode><avxsizesuffix>_<castmode>,
35981 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35982 in condition that at least one operand is not a MEM.
35984 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35986 PR middle-end/79805
35987 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35988 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35990 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35991 gimple_call_nothrow_p flag based on whether original builtin can throw.
35992 If it can, emit following stmts on the fallthrough edge.
35993 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35994 don't create new bb if inserting just debug stmts on the edge, try to
35995 insert them on the fallthru bb or just reset debug stmts.
35997 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36000 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36001 restore recog_data (including the operand rtxes inside it) around
36002 the call to get_insn_template.
36004 2017-03-03 Martin Sebor <msebor@redhat.com>
36006 PR tree-optimization/79699
36007 * context.c (context::~context): Free MPFR caches to avoid
36008 a memory leak on program exit.
36010 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36012 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36013 Use wide_int::ulow () instead of .elt (0).
36015 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36017 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36018 (*pushxf): Limit oF constraint to 32bit targets and add oC
36019 constraint for 64bit targets.
36020 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36021 (*pushdf): Change rmF constraint to rmC.
36023 2017-03-03 Martin Liska <mliska@suse.cz>
36025 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36026 Remove unused variable.
36028 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36031 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36032 is a memory operand, increase num_memory.
36033 (ix86_expand_args_builtin): Likewise.
36035 2017-03-03 Jan Hubicka <jh@suse.cz>
36038 * ipa-devirt.c (maybe_record_node): Properly handle
36039 __cxa_pure_virtual visibility.
36041 2017-03-03 Martin Liska <mliska@suse.cz>
36043 PR tree-optimization/79803
36044 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36046 (pass_loop_prefetch::execute): Disabled optimization if an
36047 assumption about L1 cache size is not met.
36049 2017-03-03 Martin Liska <mliska@suse.cz>
36051 PR rtl-optimization/79574
36052 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36053 (hash_scan_set): Likewise.
36054 (dump_hash_table): Likewise.
36055 (hoist_code): Likewise.
36057 2017-03-03 Richard Biener <rguenther@suse.de>
36059 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36060 in place of uhigh ().
36061 (fixed_convert_from_real): Likewise.
36063 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36066 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36068 2017-03-03 Richard Biener <rguenther@suse.de>
36070 PR middle-end/79818
36071 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36072 TYPE_OVERFLOW_UNDEFINED check.
36074 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36076 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36078 (vector_ae_<mode>_p): Likewise.
36079 (vector_nez_<mode>_p): Likewise.
36080 (vector_ne_v2di_p): Likewise.
36081 (vector_ae_v2di_p): Likewise.
36082 (vector_ne_<mode>_p): Likewise.
36083 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36085 (vsx_tsqrt<mode>2_fe): Likewise.
36087 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36090 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36092 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36094 PR rtl-optimization/79780
36095 * cprop.c (one_cprop_pass): When second and further conditional trap
36096 in a single basic block is turned into an unconditional trap, turn it
36097 into a deleted note to avoid RTL verification failures.
36099 2017-03-02 Richard Biener <rguenther@suse.de>
36101 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36103 2017-03-02 Richard Biener <rguenther@suse.de>
36105 PR tree-optimization/79345
36107 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36108 param and abort the walk, returning -1 if it is hit.
36109 (walk_aliased_vdefs): Take a limit param and pass it on.
36110 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36111 defaulting to 0 and return a signed int.
36112 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36113 (check_defs): New helper.
36114 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36115 about uninitialized memory.
36116 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36117 bogus uninitialized warning.
36118 (fixed_convert_from_real): Likewise.
36120 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36122 PR tree-optimization/66768
36123 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36124 iv_use if base object can't be determined.
36126 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36128 PR tree-optimization/79345
36129 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36130 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36131 (get_pattern_stats): Initialize it.
36132 * genemit.c (gen_expand): Verify match_scratch numbers come after
36133 match_operand/match_dup numbers.
36134 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36135 match_scratch numbers.
36136 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36138 * config/s390/s390.md (trunctdsd2): Likewise.
36140 2017-03-02 Richard Biener <rguenther@suse.de>
36142 * wide-int.h (wide_int_storage::operator=): Implement in terms
36145 2017-03-02 Richard Biener <rguenther@suse.de>
36147 PR tree-optimization/79777
36148 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36149 the to insert expression to sth existing.
36151 2017-03-01 Martin Sebor <msebor@redhat.com>
36153 PR middle-end/79692
36154 * gimple-ssa-sprintf.c
36155 (directive::known_width_and_precision): New function.
36156 (format_integer): Use it.
36157 (get_mpfr_format_length): Consider the full range of precision
36158 when computing %g output with the # flag. Set the likely byte
36159 count to 3 rather than 1 when precision is indeterminate.
36160 (format_floating): Correct the lower bound of precision.
36162 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36164 * doc/invoke.texi: Document default code model for 64-bit Linux.
36166 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36169 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36170 udiv rather than div since input pattern is unsigned.
36172 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36174 * config/i386/i386.c (print_reg): Warn for values of
36175 unsupported size in integer register.
36177 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36180 * config/rs6000/predicates.md (current_file_function_operand): Do
36181 not allow self calls to be local if the function is replaceable.
36183 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36186 * config/rs6000/altivec.h (vec_ctz and others): Change the
36187 preprocessor macro that controls conditional compilation from
36188 _ARCH_PWR9 to __POWER9_VECTOR__.
36189 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36190 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36191 control (instead of _ARCH_PWR9 control) so that template
36192 definition uses power9-specific function.
36193 (vec_any_eq): Likewise.
36194 (vec_all_ne): Change macro definition to use a power9-specific
36195 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36196 _ARCH_PWR9 control).
36197 (vec_any_eq) Likewise.
36198 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36199 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36200 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36201 support for xvcmpnedp instruction.
36202 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36203 macro expansion so that Power9 implementation of vec_all_ne does
36204 not use the AltiVec predicate framework.
36205 (VCMPNEH_P): Likewise.
36206 (VCMPNEW_P): Likewise.
36207 (VCMPNED_P): Likewise.
36208 (VCMPNEFP_P): Likewise.
36209 (VCMPNEDP_P): Likewise.
36210 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36211 implementation of vec_any_eq to not use AltiVec predicate
36213 (VCMPAEH_P): Likewise.
36214 (VCMPAEW_P): Likewise.
36215 (VCMPAED_P): Likewise.
36216 (VCMPAEFP_P): Likewise.
36217 (VCMPAEDP_P): Likewise.
36218 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36219 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36220 not use the AltiVec predicate framework.
36221 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36222 of vec_any_eq to not use AltiVec predicate framework.
36223 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36224 support for predefined __POWER9_VECTOR__ macro to indicate that
36225 Power9 instruction selection is enabled.
36226 (altivec_overloaded_builtins): Remove extraneous
36227 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36228 function argument types RS6000_BTI_bool_V16QI and
36229 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36230 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36231 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36232 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36233 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36234 Power9 for implementations of vec_cmpne. Change the signature for
36235 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36236 (representing vec_all_ne) to remove the previously described first
36237 argument of type RS6000_BTI_INTSI, as this was an artifact of
36238 reliance on the AltiVec predicate framework, which is no longer
36239 used in the implementation of these functions. Add
36240 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36241 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36242 since, unlike the AltiVec predicate framework implementation, we
36243 do not share function descriptors between vec_alle and vec_anyeq.
36244 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36245 set of modes that receive special treatment even when
36246 TARGET_P9_VECTOR is true. The special treatment emits code that
36247 does not depend on Power9 instructions.
36248 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36249 define_expand to not rely on AltiVec predicate framework.
36250 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36252 (vector_ne_v2di_p): Change this define_expand to not rely on
36253 AltiVec predicate framework.
36254 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36256 (vector_ne_<mode>_p): Change this define_expand to not rely on
36257 AltiVec predicate framework.
36258 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36260 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36261 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36262 define_insn pattern.
36263 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36264 define_insn pattern because the xvcmpne<VSs>. instruction is not
36266 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36267 instruction is not supported.
36269 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36271 * config/nvptx/nvptx.c: Include intl.h.
36273 2017-03-01 Martin Jambor <mjambor@suse.cz>
36276 * ipa-prop.h (ipa_bits): Removed field known.
36277 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36278 to pointers. Adjusted their comments to warn about their sharing.
36279 (ipcp_transformation_summary): Change bits to a vector of pointers.
36280 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36281 (ipa_get_ipa_bits_for_value): Declare.
36282 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36283 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36284 (ipa_bits_hash_table): Likewise.
36285 (ipa_vr_ggc_hash_traits): Likewise.
36286 (ipa_vr_hash_table): Likewise.
36287 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36288 being pointers and vr_known being removed.
36289 (ipa_set_jf_unknown): Likewise.
36290 (ipa_get_ipa_bits_for_value): New function.
36291 (ipa_set_jfunc_bits): Likewise.
36292 (ipa_get_value_range): New overloaded functions.
36293 (ipa_set_jfunc_vr): Likewise.
36294 (ipa_compute_jump_functions_for_edge): Use the above functions to
36295 construct bits and vr parts of jump functions.
36296 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36297 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36299 (ipcp_grow_transformations_if_necessary): Also allocate
36300 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36302 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36303 them. Fix too long lines.
36304 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36305 vr_known being removed.
36306 (ipa_read_jump_function): Use new setter functions to construct bits
36307 and vr parts of jump functions or set them to NULL.
36308 (write_ipcp_transformation_info): Adjust for bits being pointers.
36309 (read_ipcp_transformation_info): Likewise.
36310 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36312 Include gt-ipa-prop.h.
36313 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36315 (ipcp_store_bits_results): Likewise.
36316 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36317 Do not write to existing jump functions but use a temporary instead.
36319 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36322 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36323 attempt to use its first operand as BIT_FIELD_REF base.
36325 2017-03-01 Richard Biener <rguenther@suse.de>
36327 PR middle-end/79721
36328 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36329 interpolating formula in wrapping arithmetic.
36330 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36332 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36334 PR tree-optimization/79734
36335 * tree-vect-generic.c (expand_vector_condition): Optimize
36336 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36337 Handle VEC_COND_EXPR where comparison has different inner width from
36338 type's inner width.
36340 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36342 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36343 markup, and similar issues. Remove @opindex entries for things
36344 that aren't options. Add missing -mmpy-option entries.
36346 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36348 PR tree-optimization/79737
36349 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36350 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36351 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36352 instead of byte_size. Formatting fix.
36353 (shift_bytes_in_array_right): Formatting fix.
36355 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36358 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36359 condition on optimize for the leaf function test.
36361 2017-02-28 Martin Liska <mliska@suse.cz>
36364 * read-rtl-function.c (function_reader::handle_unknown_directive):
36365 Bail out when one uses -flto.
36367 2017-02-28 Martin Liska <mliska@suse.cz>
36369 * common.opt: Replace space with tabular for options of <number>
36371 * config/i386/i386.opt: Show <number> value for
36372 -mlarge-data-threshold.
36373 * opts.c (print_filtered_help): Do not display number in hexadecimal
36376 2017-02-28 Martin Liska <mliska@suse.cz>
36378 * common.opt: Fix --help=option -Q for options which are of
36381 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36383 * config/i386/i386.c (print_reg): Error out for values
36384 of 8-bit size in invalid integer register.
36386 2017-02-28 Martin Sebor <msebor@redhat.com>
36388 PR tree-optimization/79691
36389 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36391 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36394 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36395 gcc_unreachable with output_operand_lossage.
36397 2017-02-28 Richard Biener <rguenther@suse.de>
36399 PR tree-optimization/79740
36400 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36402 (visit_nary_op): Insert the nary into the hashtable if we
36403 pattern-matched sth.
36404 * tree-ssa-pre.c (eliminate_insert): Robustify.
36406 2017-02-28 Richard Biener <rguenther@suse.de>
36408 PR middle-end/79731
36409 * fold-const.c (decode_field_reference): Reject out-of-bound
36412 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36414 * config/i386/i386.c: Include intl.h.
36415 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36416 instead of just cond ? "..." : "...".
36417 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36418 * coverage.c (read_counts_file): Likewise.
36419 * omp-offload.c: Include intl.h.
36420 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36421 of just cond ? "..." : "...".
36422 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36423 of just cond ? "..." : "...".
36425 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36428 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36430 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36431 'tune for' CPU name.
36432 * config/arm/arm-cpu-data.h: Regenerated.
36434 2017-02-28 Richard Biener <rguenther@suse.de>
36436 PR tree-optimization/79732
36437 * tree-inline.c (expand_call_inline): Do not shadow var.
36439 2017-02-28 Richard Biener <rguenther@suse.de>
36441 PR tree-optimization/79723
36442 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36443 address-space properly.
36445 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36447 * doc/optinfo.texi (Optimization groups): Fix option used for
36449 * doc/invoke.texi (-fopt-info): Document "omp".
36450 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36451 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36452 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36453 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36454 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36456 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36458 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36459 OPTGROUP_OMP with "omp".
36461 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36464 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36465 for arithmetic shift of unsigned V2DI.
36467 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36469 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36470 arc/linux.h headers.
36471 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36472 (LINK_SPEC): Likewise.
36473 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36474 (EXTRA_SPECS): Likewise.
36475 (STARTFILE_SPEC): Likewise.
36476 (ENDFILE_SPEC): Likewise.
36477 (LIB_SPEC): Likewise.
36478 (TARGET_SDATA_DEFAULT): Likewise.
36479 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36480 (MULTILIB_DEFAULTS): Likewise.
36481 (DWARF2_UNWIND_INFO): Likewise.
36482 * config/arc/big.h: New file.
36483 * config/arc/elf.h: Likewise.
36484 * config/arc/linux.h: Likewise.
36485 * config/arc/t-uClibc: Remove.
36487 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36489 PR tree-optimization/77536
36490 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36491 (tree_transform_and_unroll_loop): Use above function to compute the
36492 estimated niter of unrolled loop and use it when scaling profile.
36493 Also use count info rather than frequency if it's non-zero.
36494 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36495 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36496 (vect_transform_loop): Call above function.
36498 2017-02-27 Richard Biener <rguenther@suse.de>
36500 PR tree-optimization/45397
36501 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36502 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36503 (visit_nary_op): Add pattern matching for CSEing sign-changed
36504 or truncated operations with wider ones.
36506 2017-02-27 Richard Biener <rguenther@suse.de>
36508 PR tree-optimization/79690
36509 * tree-vect-stmts.c (vectorizable_store): Use vector type
36510 built from the DR with address-space.
36512 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36514 * doc/invoke.texi (Optimize Options): Refine the description
36515 of asan-use-after-return.
36517 2017-02-25 Alan Modra <amodra@gmail.com>
36519 PR rtl-optimization/79584
36520 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36521 base, not ad->base_term, the reg within base. Remove assertion
36522 that ad->base == ad->base_term. Replace gen_int_mode using
36523 bogus mode with const0_rtx.
36525 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36527 PR middle-end/79396
36528 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36529 FMA_EXPR like tcc_binary or tcc_unary.
36531 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36534 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36536 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36537 (output_loc_operands): Handle DW_OP_call_ref and
36538 DW_OP_GNU_variable_value.
36539 (struct variable_value_struct): New type.
36540 (struct variable_value_hasher): Likewise.
36541 (variable_value_hash): New variable.
36542 (string_types): Remove.
36543 (copy_loc_descr): New function.
36544 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36545 (prepend_loc_descr_to_each): New function.
36546 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36547 instead of add_loc_descr_to_each if the first argument is single
36548 location list and the second has multiple.
36549 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36550 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36551 when looking for variable value which doesn't have other location info.
36552 (loc_list_from_tree): Formatting fix.
36553 (gen_array_type_die): Simplify DW_AT_string_length handling.
36554 (adjust_string_types): Remove.
36555 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36556 string_types. Call resolve_variable_values.
36557 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36558 (resolve_addr_in_expr): Likewise. Add A argument.
36559 (copy_deref_exprloc): Remove deref argument. Adjust for the
36560 original expression being DW_OP_GNU_variable_value with optionally
36561 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36562 optionally after it.
36563 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36564 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36565 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36566 (variable_value_hasher::hash, variable_value_hasher::equal): New
36568 (resolve_variable_value_in_expr, resolve_variable_value,
36569 resolve_variable_values, note_variable_value_in_expr,
36570 note_variable_value): New functions.
36571 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36574 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36577 * opts.h (handle_generated_option): Add GENERATED_P argument.
36578 * opts-common.c (handle_option): Adjust function comment.
36579 (handle_generated_option): Add GENERATED_P argument, pass it to
36581 (control_warning_option): Pass false to handle_generated_option
36583 * opts.c (maybe_default_option): Pass true to handle_generated_option
36585 * optc-gen.awk: Likewise.
36587 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36589 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36590 a REG, look at the REG it is a SUBREG of.
36591 (splitter for cmpeqsi_t): Ditto.
36593 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36595 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36596 the special USEs with the pattern of the insn, not the insn itself.
36598 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36601 * doc/invoke.texi: Document -mload-store-pairs.
36603 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36604 Sandra Loosemore <sandra@codesourcery.com>
36606 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36607 argument isn't a CONST_INT.
36608 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36609 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36610 (nios2_validate_compare): Bypass alternate compare logic if *op2
36611 is not a CONST_INT.
36612 (ldstwm_operation_p): Return false if first_base is not a REG or
36613 if first_offset is not a CONST_INT.
36615 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36617 * config/cris/cris.md: Use correct operand in a define_peephole2.
36619 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36621 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36623 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36625 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36626 this_insn if it is an INSN or JUMP_INSN.
36627 (force_offsettable): Look at base, not at addr.
36628 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36629 on things that aren't necessarily CONST_INTs.
36631 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36633 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36634 -mfpmath=sse is the default also for x86-32 targets with SSE2
36635 instruction set when @option{-ffast-math} is enabled
36637 2017-02-24 Jeff Law <law@redhat.com>
36639 PR rtl-optimizatoin/79286
36640 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36643 2017-02-24 Richard Biener <rguenther@suse.de>
36645 PR tree-optimization/79389
36646 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36649 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36651 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36652 function comment to reflect reality.
36653 (loop_exits_before_overflow): Fix typo in function description.
36655 2017-02-24 Richard Biener <rguenther@suse.de>
36657 PR tree-optimization/79389
36658 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36659 properly that a threading opportunity exists. Detect conditional
36660 copy/constant propagation opportunities.
36662 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36664 * config/visium/visium.md (type): Add trap.
36665 (b): New mode attribute.
36666 (*btst): Rename into...
36667 (*btst<mode>): ...this and adjust.
36668 (*cbranchsi4_btst_insn): Rename into...
36669 (*cbranch<mode>4_btst_insn): ...this and adjust.
36670 (trap): New define_insn.
36672 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36674 PR tree-optimization/79389
36675 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36676 (noce_reversed_cond_code): New function.
36677 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36678 reversed_comparison_code. Formatting fix.
36679 (noce_try_store_flag): Test rev_cond != NULL in addition to
36680 reversed_comparison_code.
36681 (noce_try_store_flag_constants): Likewise.
36682 (noce_try_store_flag_mask): Likewise.
36683 (noce_try_addcc): Use rev_cond if non-NULL instead of
36684 reversed_comparison_code.
36685 (noce_try_cmove_arith): Likewise. Formatting fixes.
36686 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36687 (noce_find_if_block): Initialize rev_cond.
36688 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36689 instead of false as last argument never attempt to reverse it
36692 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36694 PR tree-optimization/79663
36695 * tree-predcom.c (combine_chains): Process refs in reverse order
36696 only for ZERO length chains, and add explaining comment.
36698 2017-02-23 Jeff Law <law@redhat.com>
36700 PR tree-optimization/79578
36701 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36702 in call to operand_equal_p.
36704 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36707 * config/i386/cpuid.h: Fix another undefined behavior.
36709 2017-02-23 Richard Biener <rguenther@suse.de>
36711 PR tree-optimization/79683
36712 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36713 vector types for data-refs.
36715 2017-02-23 Martin Liska <mliska@suse.cz>
36717 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36719 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36721 PR middle-end/79665
36722 * internal-fn.c (get_range_pos_neg): Moved to ...
36723 * tree.c (get_range_pos_neg): ... here. No longer static.
36724 * tree.h (get_range_pos_neg): New prototype.
36725 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36726 are known to be in between 0 and signed maximum inclusive, try to
36727 expand both unsigned and signed divmod and use the cheaper one from
36730 2017-02-22 Jeff Law <law@redhat.com>
36732 PR tree-optimization/79578
36733 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36734 to compare base operands.
36736 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36739 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36740 gpc_reg_operand instead of fpr_reg_operand.
36742 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36744 * config/mips/mips.c (mips_return_in_memory): Force FP
36745 vector types to be returned in memory for o32 ABI.
36747 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36749 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36750 instead of DW_TAG_member for static data member declarations and don't
36751 set no_linkage_name for static inline data members.
36752 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36755 2017-02-22 Martin Liska <mliska@suse.cz>
36757 * doc/invoke.texi: Replace inequality signs with square brackets
36760 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36763 * lra-constraints.c (simplify_operand_subreg): Handle
36764 WORD_REGISTER_OPERATIONS targets.
36766 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36769 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36770 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36771 elimination by swapping fld*.
36773 2017-02-22 Richard Biener <rguenther@suse.de>
36775 PR tree-optimization/79673
36776 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36777 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36778 irrelevant address-space qualifiers and avoiding a
36779 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36781 2017-02-22 Richard Biener <rguenther@suse.de>
36783 PR tree-optimization/79666
36784 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36785 to not symbolically negate if that may introduce undefined
36788 2017-02-22 Martin Liska <mliska@suse.cz>
36791 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36792 * data-streamer-out.c (streamer_write_gcov_count_stream):
36794 * value-prof.c (stream_out_histogram_value): Make assert more
36795 precise based on type of counter.
36797 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36800 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36801 Use nonimmediate_operand instead of memory_operand for operand 1.
36802 (float-extend standard_x87sse_constant_load splitter): Ditto.
36804 2017-02-21 Jeff Law <law@redhat.com>
36806 PR tree-optimization/79621
36807 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36808 blocks with edges to themselves.
36810 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36813 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36814 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36815 Use gimple_call_builtin_p.
36818 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36819 on temporarily removed DEBUG_INSNs.
36821 PR tree-optimization/79649
36822 * tree-loop-distribution.c (classify_partition): Give up on
36823 non-generic address space loads/stores.
36825 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36827 * doc/loop.texi (Loop manipulation): Remove nonexistent
36828 tree_ssa_loop_version from the documentation.
36829 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36831 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36834 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36835 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36836 * config/rs6000/rs6000.c: Include except.h.
36837 (rs6000_expand_split_stack_prologue): Call
36838 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36840 2017-02-21 Martin Jambor <mjambor@suse.cz>
36843 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36844 have been analyzed.
36846 2017-02-21 Martin Jambor <mjambor@suse.cz>
36848 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36849 for backward compatibility only.
36850 * doc/invoke.texi (Option Summary): Remove all references to
36851 -fipa-cp-alignment.
36853 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36857 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36859 * lra-constraints.c (curr_insn_transform): Handle
36860 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36862 2017-02-21 Martin Liska <mliska@suse.cz>
36864 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36866 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36868 PR translation/79638
36869 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36871 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36874 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36875 (arm_function_ok_for_sibcall): Return false for an indirect call by
36876 descriptor if all the argument registers are used.
36877 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36878 alignment of the function.
36880 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36882 PR tree-optimization/61441
36883 * simplify-rtx.c (simplify_const_unary_operation): For
36884 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36885 the sNaN unmodified.
36887 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36889 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36890 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36891 instead of SYSTEM_HEADER_DIR.
36893 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36894 Martin Liška <mliska@suse.cz>
36896 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36897 Fix typos and grammar, use active voice, and clarify.
36899 2017-02-20 Marek Polacek <polacek@redhat.com>
36901 PR middle-end/79537
36902 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36905 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36907 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36910 * config/i386/i386.c (ix86_expand_builtin): Handle
36911 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36912 ix86_builtins_isa[fcode].isa as a requirement of those
36913 flags and any other flag in the bitmask.
36914 (ix86_init_mmx_sse_builtins): Use 0 instead of
36915 ~OPTION_MASK_ISA_64BIT as mask.
36916 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36917 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36918 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36919 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36921 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36924 * lra-constraints.c (split_reg): Check requested split mode
36925 is supported by the register.
36927 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36929 * lra-constraints.c (simplify_operand_subreg): Remove early
36932 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36935 * lra-constraints.c (curr_insn_transform): Tighten condition
36936 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36938 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36941 * lra-constraints.c (curr_insn_transform): Handle
36942 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36944 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36947 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36949 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36951 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36954 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36957 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36959 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36962 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36964 * final.c (last_columnnum, override_columnnum): New variables.
36965 (final_start_function): Set last_columnnum, pass it to begin_prologue
36966 hook and pass 0 to dwarf2out_begin_prologue.
36967 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36968 to source_line debug hook.
36969 (notice_source_line): Compute last_columnnum and for debug_column_info
36970 return true on column changes.
36971 * debug.h (struct gcc_debug_hooks): Add column argument to
36972 source_line and begin_prologue hooks.
36973 (debug_nothing_int_charstar_int_bool): Remove prototype.
36974 (debug_nothing_int_int_charstar,
36975 debug_nothing_int_int_charstar_int_bool): New prototypes.
36976 (dwarf2out_begin_prologue): Add column argument.
36977 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36978 begin_prologue hooks.
36979 (debug_nothing_int_charstar_int_bool): Remove.
36980 (debug_nothing_int_int_charstar,
36981 debug_nothing_int_int_charstar_int_bool): New functions.
36982 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36983 through to dwarf2out_source_line.
36984 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36985 (dwarf2out_source_line): Add column argument, emit it if requested.
36986 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36988 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36989 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36990 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36991 through to dwarf2out_begin_prologue.
36992 (vmsdbgout_source_line): Add column argument, pass it through to
36993 dwarf2out_source_line.
36994 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36995 dbxout_source_line caller.
36996 (dbxout_source_line): Add column argument.
36998 * common.opt (gno-column-info, gcolumn-info): New options.
36999 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37000 (check_die): Also test for multiple DW_AT_decl_column attributes.
37001 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37002 DW_AT_decl_column if requested.
37003 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37005 (gen_variable_die): Likewise.
37006 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37007 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37010 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37011 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37012 (ix86_handle_option): Handle OPT_m3dnowa.
37013 * doc/invoke.texi (-m3dnowa): Document.
37014 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37015 -m3dnowa instead of -m3dnow -march=athlon.
37018 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37019 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37021 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37024 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37025 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37026 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37027 generator for vsx_xxpermdi_<mode>_be.
37028 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37029 force big-endian semantics.
37030 (vsx_xxpermdi_<mode>_be): New define_expand with same
37031 implementation as previous version of vsx_xxpermdi_<mode>.
37033 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37035 PR tree-optimization/79327
37036 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37037 variable, its initialization and use.
37039 2017-02-17 Julia Koval <julia.koval@intel.com>
37041 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37042 (OPTION_MASK_ISA_PKU_UNSET): New.
37043 (ix86_handle_option): Handle -mrdpid.
37044 * config/i386/cpuid.h (bit_RDPID): New.
37045 * config/i386/driver-i386.c (host_detect_local_cpu):
37046 Detect RDPID feature.
37047 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37048 * config/i386/i386-c.c (ix86_target_macros_internal):
37050 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37051 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37052 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37053 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37054 * config/i386/i386.md (define_insn "rdpid"): New.
37055 * config/i386/i386.opt Add -mrdpid.
37056 * config/i386/immintrin.h (_rdpid_u32): New.
37058 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37060 PR rtl-optimization/79541
37061 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37062 instead of transforming it into USE.
37064 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37066 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37067 If HONOR_SNANS (SFmode) force the input to a register.
37068 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37069 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37070 an frsp or similar insn.
37072 2017-02-17 Martin Liska <mliska@suse.cz>
37074 PR rtl-optimization/79577
37075 * params.def (selsched-max-sched-times): Increase minimum to 1.
37077 2017-02-17 Martin Liska <mliska@suse.cz>
37079 PR rtl-optimization/79574
37080 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37082 2017-02-17 Martin Liska <mliska@suse.cz>
37084 PR tree-optimization/79529
37085 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37086 ssa_defined_default_def_p to handle cases which are implicitly
37088 * tree-ssa.c (ssa_defined_default_def_p): New function.
37089 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37090 which are implicitly defined.
37091 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37093 2017-02-17 Richard Biener <rguenther@suse.de>
37095 PR middle-end/79576
37096 * params.def (max-ssa-name-query-depth): Limit to 10.
37098 2017-02-17 Richard Biener <rguenther@suse.de>
37100 PR tree-optimization/79552
37101 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37104 2017-02-17 Richard Biener <rguenther@suse.de>
37107 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37109 2017-02-17 Marek Polacek <polacek@redhat.com>
37111 PR middle-end/79536
37112 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37113 (fold_negate_expr): New wrapper.
37115 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37117 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37118 Correct terminology and de-emphasize pre-standard behavior.
37120 2017-02-16 Alan Modra <amodra@gmail.com>
37122 PR rtl-optimization/79286
37123 * ira.c (def_dominates_uses): New function.
37124 (update_equiv_regs): Don't create an equivalence for insns that
37125 may trap where the register def does not dominate the use.
37127 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37129 PR rtl-optimization/78127
37130 * lra.c (lra): Call lra_eliminate before finish the loop after
37133 2017-02-16 Richard Biener <rguenther@suse.de>
37135 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37137 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37138 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37139 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37140 (isl_val_int_from_wi): New function.
37141 (extract_affine_gmp): Rename to ...
37142 (extract_affine_wi): ... this, take a widest_int.
37143 (extract_affine_int): Just wrap extract_affine_wi.
37144 (add_param_constraints): Use isl_val_int_from_wi.
37145 (add_loop_constraints): Likewise, and extract_affine_wi.
37147 2017-02-15 Jeff Law <law@redhat.com>
37149 PR middle-end/79521
37150 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37151 ira_init_register_move_cost_if_necessary.
37153 2017-02-15 Martin Sebor <msebor@redhat.com>
37155 PR middle-end/32003
37156 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37157 removed in a prior commit.
37159 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37161 PR tree-optimization/79347
37162 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37163 counters during peeling.
37165 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37167 * Makefile.in (site.exp): Remove "set ISLVER".
37169 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37172 * real.c (real_from_integer): Call real_convert even for decimal.
37174 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37177 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37179 2017-02-14 Andrew Pinski <apinski@cavium.com>
37181 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37182 cores and change the partno/implementer to be correct.
37183 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37184 the 'B" as the implementer.
37185 * config/aarch64/aarch64-tune.md: Regenerate.
37187 2017-02-14 Carl Love <cel@us.ibm.com>
37189 * config/rs6000/rs6000.c: Add case statement entry to make the
37190 xvcvuxdsp built-in argument unsigned.
37191 * config/rs6000/vsx.md: Fix the source and return operand types so they
37192 match the instruction definitions from the ISA document. Fix typo
37193 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37196 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37199 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37200 member early_clobber_alts.
37201 * lra-lives.c (reg_early_clobber_p): New.
37202 (process_bb_lives): Use it.
37203 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37204 (debug_operand_data): Initialize early_clobber_alts.
37205 (setup_operand_alternative): Set up early_clobber_alts.
37206 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37207 alternatives to new_insn_reg.
37208 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37210 (lra_update_insn_regno_info): Pass the new arg.
37212 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37214 PR middle-end/79505
37215 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37216 (new_oacc_loop_raw): Don't clear already cleared fields.
37219 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37220 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37221 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37223 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37226 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37228 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37231 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37232 the extra instruction to the right place to store 128-bit constant
37235 2017-02-14 Martin Sebor <msebor@redhat.com>
37237 PR middle-end/79448
37238 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37239 warning for strings of unknown length.
37241 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37243 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37245 2017-02-14 Jeff Law <law@redhat.com>
37248 * ira-costs.c (scan_one_insn): Initialize register move costs
37249 for pseudos seen in USE/CLOBBER insns.
37251 PR tree-optimization/79095
37252 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37253 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37254 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37255 if the operands are known to be not equal, then the resulting range
37257 (intersect_ranges): If the new range is ~[0,0] and the old range is
37258 wide, then prefer ~[0,0].
37259 * tree-vrp.c (overflow_comparison_p_1): New function.
37260 (overflow_comparison_p): New function.
37261 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37262 if NAME is used in an overflow test.
37263 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37264 overflow check that can be expressed as an equality test, then adjust
37265 ops to be that equality test.
37267 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37269 * config/s390/s390-builtin-types.def: Remove flags argument.
37270 * config/s390/s390.c (s390_init_builtins): Likewise.
37272 2017-02-14 Martin Liska <mliska@suse.cz>
37274 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37275 vector. Fix trailing white spaces.
37277 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37279 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37282 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37284 PR rtl-optimization/68664
37285 * config/arm/arm.c (arm_sched_can_speculate_insn):
37286 New function. Declare prototype.
37287 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37289 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37291 PR rtl-optimization/68664
37292 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37294 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37296 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37298 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37299 max skip bytes for function, loop and jump.
37301 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37303 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37304 ABS_EXPR for gimple dump.
37306 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37309 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37311 PR tree-optimization/79408
37312 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37313 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37314 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37315 also if rhs1 is INTEGER_CST.
37317 2017-02-14 Richard Biener <rguenther@suse.de>
37319 PR middle-end/79432
37320 * tree-into-ssa.c (insert_phi_nodes): When the function can
37321 have abnormal edges rewrite SSA names with broken use-def
37322 dominance out of SSA and register them for PHI insertion.
37324 2017-02-13 Martin Sebor <msebor@redhat.com>
37326 PR middle-end/79496
37327 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37328 clearing info.nowrite flag when snprintf size argument is a range.
37330 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37332 * cprop.c (cprop_jump): Add missing space in string literal.
37333 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37334 (get_constraint_for_component_ref): Likewise.
37335 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37336 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37337 * lra-constraints.c (process_alt_operands): Likewise.
37338 * ipa-inline.c (inline_small_functions): Likewise.
37339 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37340 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37341 * trans-mem.c (diagnose_tm_1_op): Likewise.
37342 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37343 (grid_parallel_clauses_gridifiable): Likewise.
37345 * config/nvptx/mkoffload.c (process): Add space in between
37348 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37349 "MOD4_SSE_REGS" and "ALL_REGS".
37351 * spellcheck.c (test_data): Add , in between "foo" and "food".
37353 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37356 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37357 boundary crossing check and subsequent code generation agree.
37359 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37361 * config/aarch64/aarch64.c (has_memory_op): Delete.
37362 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37365 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37367 PR rtl-optimization/79388
37368 PR rtl-optimization/79450
37369 * combine.c (distribute_notes): When removing TEM_INSN for which
37370 corresponding dest has last value recorded, invalidate that last
37373 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37375 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37376 of explicit '@'. Add missing assembly comment marker on branch costs
37379 2017-02-13 Nathan Sidwell <nathan@acm.org>
37381 * gengtype-lex.l (<in_struct>): Add '/'.
37383 2017-02-13 Martin Liska <mliska@suse.cz>
37386 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37388 2017-02-13 Richard Biener <rguenther@suse.de>
37390 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37392 * configure: Re-generate.
37393 * config.in: Likewise.
37394 * graphite-dependences.c: Simplify as if
37395 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37396 * graphite-isl-ast-to-gimple.c: Likewise.
37397 * graphite-optimize-isl.c: Likewise.
37398 * graphite-poly.c: Likewise.
37399 * graphite-sese-to-poly.c: Likewise.
37400 * graphite.h: Likewise.
37401 * toplev.c: Include isl/version.h and use isl_version () for
37402 printing the ISL version.
37403 * doc/install.texi: Update ISL requirement.
37405 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37407 * doc/standards.texi (Standards): Update reference to
37410 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37412 * doc/extend.texi (Named Address Spaces): sourceware.org now
37414 * doc/install.texi (Binaries): Ditto.
37417 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37419 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37420 standard terminology "stringize"/"stringizing" throughout.
37421 * doc/cppinternals.texi: Likewise.
37423 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37425 * doc/extend.texi: Fix some spelling mistakes and typos.
37426 * doc/invoke.texi: Likewise.
37428 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37431 * params.def (inline-min-speedup) Change from 10 to 8.
37433 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37435 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37438 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37441 * ipa-inline-analysis.c (get_minimal_bb): New function.
37442 (record_modified): Use it.
37443 (remap_edge_change_prob): Handle also ancestor functions.
37445 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37447 * doc/contrib.texi (Contributors): Remove broken link into
37448 the Mauve CVS repository.
37450 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37452 PR middle-end/79454
37453 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37454 result computation whenever lhs doesn't have vector mode, not
37455 just when it has BLKmode.
37457 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37459 * doc/makefile.texi (profiledbootstrap): Refer to the
37460 installation instructions only in textual form.
37462 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37465 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37467 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37469 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37470 (Specific): Update mingw-w64 reference.
37472 (Specific): Remove broken link to Renesas RX processor.
37474 2017-02-10 Richard Biener <rguenther@suse.de>
37476 * toplev.c (process_options): Do not mention obsolete graphite
37477 options when printing sorry message about missing graphite support.
37478 Mention -floop-nest-optimize.
37480 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37482 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37483 (vtst_p16): Likewise.
37484 (vtstq_p8): Likewise.
37485 (vtstq_p16): Likewise.
37487 (vtstq_p64): Likewise.
37488 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37489 (vset_lane_p64): New.
37490 (vsetq_lane_p64): New.
37492 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37494 PR tree-optimization/79411
37495 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37496 stmt operands are SSA_NAMEs used in abnormal phis.
37497 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37500 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37503 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37506 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37508 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37510 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37512 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37516 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37517 not arbitrary TREE_CONSTANT.
37520 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37521 "omp declare target link" attribute unless is_global_var.
37522 * omp-offload.c (find_link_var_op): Likewise.
37524 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37525 Chung-Lin Tang <cltang@codesourcery.com>
37527 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37529 (gimplify_adjust_omp_clauses): Don't delete TILE.
37530 (gimplify_omp_for): Deal with TILE.
37531 * internal-fn.c (expand_GOACC_TILE): New function.
37532 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37534 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37535 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37537 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37538 avoid DIV for outermost collapse var.
37539 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37540 Remove out of date comments, fix whitespace.
37541 * omp-general.c (omp_extract_for_data): Deal with tiling.
37542 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37543 adjust OLF_DIM_BASE value.
37544 (struct omp_for_data): Add tiling field.
37545 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37546 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37547 for auto loops. Remove default auto determining, moved to
37548 oacc_loop_fixed_partitions.
37549 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37550 stmts, add e_mask field.
37551 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37552 (oacc_thread_numbers): Use oacc_dim_call.
37553 (oacc_xform_tile): New.
37554 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37555 (finish_oacc_loop): Adjust for ifns vector.
37556 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37557 add case for GOACC_TILE fns.
37558 (oacc_loop_xform_loop): Delete.
37559 (oacc_loop_process): Iterate over call list directly, and add
37560 handling for GOACC_TILE fns.
37561 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37563 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37564 vector partitioning to outer loops. Assign 2 partitions to loops
37565 when available. Add TILE handling.
37566 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37567 (execite_oacc_device_lower): Process GOACC_TILE fns,
37568 ignore unknown specs.
37569 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37570 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37571 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37573 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37575 * configure.ac (ACX_BUGURL): Update.
37576 * configure: Regenerate.
37578 2017-02-09 Richard Biener <rguenther@suse.de>
37580 PR tree-optimization/69823
37581 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37582 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37584 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37586 * config/arc/arc-c.def: Add __NPS400__ definition.
37587 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37588 (TARGET_NPS400): Define.
37590 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37592 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37594 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37595 pointer, arch_info.
37596 (arc_cpu_types): Fill the arch_info field with a pointer into the
37597 arc_arch_types table.
37598 (arc_selected_cpu): Declare.
37599 * config/arc/arc.c (arc_selected_cpu): Make global.
37600 (arc_selected_arch): Delete.
37601 (arc_base_cpu): Delete.
37602 (arc_override_options): Remove references to deleted variables,
37603 update access to arch information.
37604 (ARC_OPT): Update access to arch information.
37605 (ARC_OPTX): Likewise.
37606 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37607 (TARGET_ARC600): Update access to arch information.
37608 (TARGET_ARC601): Likewise.
37609 (TARGET_ARC700): Likewise.
37610 (TARGET_EM): Likewise.
37611 (TARGET_HS): Likewise.
37612 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37615 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37618 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37619 condition/operands for integer GE/LE/GEU/LEU operations.
37621 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37623 PR translation/79397
37624 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37627 2017-02-08 Martin Jambor <mjambor@suse.cz>
37630 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37631 whether allocation happened.
37632 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37633 nothing was allocated.
37635 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37637 PR tree-optimization/79408
37638 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37639 constant, but SSA_NAME with a known integer range, use the minimum
37640 of that range instead of op1 to determine if modulo can be replaced
37641 with its first operand.
37643 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37645 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37647 2017-02-08 Richard Biener <rguenther@suse.de>
37649 PR tree-optimization/71824
37650 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37651 Check all loops contained in the merged region.
37653 2017-02-07 Andrew Pinski <apinski@cavium.com>
37655 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37657 2017-02-07 Andrew Pinski <apinski@cavium.com>
37659 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37660 (thunderxt88): Likewise.
37661 (thunderxt81): Disable LSE and change v8.1 to v8.
37662 (thunderxt83): Likewise.
37664 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37665 Richard Biener <rguenther@suse.de>
37667 PR middle-end/79399
37668 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37669 type from int to size_t.
37670 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37672 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37674 PR rtl-optimization/79386
37675 * cprop.c (bypass_conditional_jumps): Initialize
37676 bypass_last_basic_block already before splitting bbs after
37677 unconditional traps...
37678 (bypass_conditional_jumps): ... rather than here.
37681 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37682 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37683 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37684 fix -masm=intel patterns.
37686 2017-02-07 Richard Biener <rguenther@suse.de>
37688 PR tree-optimization/79256
37689 PR middle-end/79278
37690 * builtins.c (get_object_alignment_2): Use min_align_of_type
37691 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37692 and ADJUST_FIELD_ALIGN.
37694 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37696 * doc/tm.texi: Regenerate.
37697 * stor-layout.c (layout_decl): Adjust.
37698 (update_alignment_for_field): Likewise.
37699 (place_field): Likewise.
37700 (min_align_of_type): Likewise.
37701 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37702 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37703 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37704 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37705 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37706 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37707 * config/i386/i386.c (x86_field_alignment): Likewise.
37708 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37709 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37710 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37711 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37712 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37713 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37717 2017-01-30 Richard Biener <rguenther@suse.de>
37719 PR tree-optimization/79256
37720 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37721 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37724 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37726 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37727 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37728 builtins to SImode and emit a zero-extend, if necessary.
37730 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37732 * docs/invoke.texi (RISC-V Options): Alphabetize.
37734 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37736 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37739 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37741 * config/riscv/riscv.c: New file.
37742 * common/config/riscv/riscv-common.c: Likewise.
37743 * config.gcc: Likewise.
37744 * config/riscv/constraints.md: Likewise.
37745 * config/riscv/elf.h: Likewise.
37746 * config/riscv/generic.md: Likewise.
37747 * config/riscv/linux.h: Likewise.
37748 * config/riscv/multilib-generator: Likewise.
37749 * config/riscv/peephole.md: Likewise.
37750 * config/riscv/pic.md: Likewise.
37751 * config/riscv/predicates.md: Likewise.
37752 * config/riscv/riscv-builtins.c: Likewise.
37753 * config/riscv/riscv-c.c: Likewise.
37754 * config/riscv/riscv-ftypes.def: Likewise.
37755 * config/riscv/riscv-modes.def: Likewise.
37756 * config/riscv/riscv-opts.h: Likewise.
37757 * config/riscv/riscv-protos.h: Likewise.
37758 * config/riscv/riscv.h: Likewise.
37759 * config/riscv/riscv.md: Likewise.
37760 * config/riscv/riscv.opt: Likewise.
37761 * config/riscv/sync.md: Likewise.
37762 * config/riscv/t-elf-multilib: Likewise.
37763 * config/riscv/t-linux: Likewise.
37764 * config/riscv/t-linux-multilib: Likewise.
37765 * config/riscv/t-riscv: Likewise.
37766 * configure.ac: Likewise.
37767 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37768 Waterman as RISC-V maintainers.
37769 * doc/install.texi: Add RISC-V entries.
37770 * doc/invoke.texi: Add RISC-V options section.
37771 * doc/md.texi: Add RISC-V constraints section.
37772 * configure: Regenerated.
37774 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37777 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37778 false values to be constant vectors with all 0 or all 1 bits set.
37779 (vcondu<mode><mode>): Likewise.
37780 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37782 (fpmask_comparison_operator): Update comment.
37783 (vecint_comparison_operator): New predicate.
37784 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37785 vector conditionals when the true and false values are constant
37786 vectors with all 0 bits or all 1 bits set.
37788 2017-02-06 Martin Sebor <msebor@redhat.com>
37790 PR tree-optimization/79376
37791 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37793 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37795 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37796 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37797 to simplify split condition.
37799 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37801 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37802 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37805 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37807 PR rtl-optimization/68664
37808 * target.def (can_speculate_insn): New hook.
37809 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37810 * doc/tm.texi: Regenerate.
37811 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37812 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37813 (rs6000_sched_can_speculate_insn): New function.
37815 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37817 PR tree-optimization/79284
37818 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37819 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37820 vectorizable_mask_load_store, vectorizable_operation,
37821 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37822 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37823 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37824 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37825 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37826 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37827 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37828 is_gimple_assign (stmt). Replace another such test with
37829 is_gimple_assign (stmt).
37831 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37834 * config/avr/avr.c (rtl-iter.h): Include it.
37835 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37836 (avr_legitimate_combined_insn): ...and implementation.
37838 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37840 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37841 * config/s390/s390.c (s390_const_operand_ok)
37842 (s390_canonicalize_comparison, s390_extract_part)
37843 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37844 (s390_contiguous_bitmask_p, s390_rtx_costs)
37845 (legitimize_pic_address): Likewise.
37846 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37847 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37848 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37849 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37850 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37852 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37854 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37855 REGNO($0) == REGNO($1).
37857 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37859 * config/s390/linux.h(SIZE_TYPE): Add comment.
37861 2017-02-06 Julian Brown <julian@codesourcery.com>
37862 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37863 Virendra Pathak <virendra.pathak@broadcom.com>
37865 * config/aarch64/aarch64-cores.def: Change the scheduler
37867 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37868 * config/aarch64/thunderx2t99.md: New file.
37870 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37872 * doc/standards.texi (Go Language): Update link to language
37875 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37877 * tree-eh.c (lower_resx): Sanitize profile.
37878 (cleanup_empty_eh_move_lp): Likewise.
37880 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37883 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37885 * cfgloopmanip.h (loop_version): Update prototype.
37886 * modulo-sched.c (sms_schedule): Update call of loop_version.
37887 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37888 * tree-parloops.c (gen_parallel_loop): Likewise.
37889 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37890 * tree-ssa-loop-split.c (split_loop): Likewise.
37891 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37892 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37894 2017-02-05 Martin Liska <mliska@suse.cz>
37897 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37899 (print_operand_address): Initialize a struct to zero.
37901 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37903 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37904 garbage collector only in textual form.
37906 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37908 * doc/extend.texi (x86 specific memory model extensions for
37909 transactional memory): Simplify a phrase.
37911 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37914 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37915 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37916 (atomic_storedi_1): Likewise.
37918 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37920 PR tree-optimization/79338
37921 * tree-parloops.c (gather_scalar_reductions): Don't call
37922 vect_analyze_loop_form for loop->inner before destroying loop's
37925 2017-02-03 Martin Sebor <msebor@redhat.com>
37927 PR tree-optimization/79327
37928 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37929 when precision has resulted in leading zeros.
37930 (format_integer): Adjust the likely counter to assume an unknown
37931 argument that may be zero is non-zero.
37933 2017-02-03 Jason Merrill <jason@redhat.com>
37936 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37937 avoid copying non-taken branch.
37939 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37941 PR tree-optimization/79340
37942 * tree-vect-loop.c (vectorizable_reduction): Release
37943 vec_defs elements after safe_splicing them into other vectors.
37946 PR tree-optimization/79327
37947 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37948 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37950 (format_integer): Use wide_int_to_tree instead of build_int_cst
37951 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37952 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37953 of shortest and longest sequence.
37955 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37957 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37958 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37960 2017-02-03 Walter Lee <walt@tilera.com>
37963 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37964 after initial stackframe link reg save.
37965 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37967 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37970 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37971 wu for stxssp alternative.
37973 2017-02-03 Martin Sebor <msebor@redhat.com>
37975 PR tree-optimization/79352
37976 * gimple-fold.c (get_range_strlen): Add argument.
37977 (get_range_strlen): Change return type to bool.
37978 (get_maxval_strlen): Pass in a dummy argument.
37979 * gimple-fold.h (get_range_strlen): Change return type to bool.
37980 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37981 * tree.h (array_at_struct_end_p): Add argument.
37982 * tree.c (array_at_struct_end_p): Handle it.
37984 2017-02-03 Martin Liska <mliska@suse.cz>
37987 * multiple_target.c (create_dispatcher_calls): Redirect edge
37988 from a caller of a dispatcher.
37989 (expand_target_clones): Make the clones local.
37990 (ipa_target_clone): Do both target clones and resolvers.
37991 (ipa_dispatcher_calls): Remove the pass.
37992 (pass_dispatcher_calls::gate): Likewise.
37993 (make_pass_dispatcher_calls): Likewise.
37994 * passes.def (pass_target_clone): Put as very first IPA early
37997 2017-02-03 Martin Liska <mliska@suse.cz>
37999 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38000 in case of a function with ifunc attribute.
38002 2017-02-03 Martin Liska <mliska@suse.cz>
38004 * cgraph.c (cgraph_node::dump): Dump function version info.
38005 * symtab.c (symtab_node::dump_base): Add missing new line.
38007 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38009 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38010 (ifcombine_ifandif): Use it.
38012 2017-02-03 Martin Liska <mliska@suse.cz>
38014 * doc/invoke.texi: Document default value for
38015 use-after-scope-direct-emission-threshold.
38017 2017-02-03 Martin Liska <mliska@suse.cz>
38019 PR tree-optimization/79339
38020 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38021 (format_floating): Likewise.
38023 2017-02-03 Martin Liska <mliska@suse.cz>
38026 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38028 (ipa_node_params_t::remove): Likewise.
38029 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38030 initialization from removed ipa_node_params_t::insert.
38031 (ipa_node_params::~ipa_node_params): Move from removed
38032 ipa_node_params_t::release.
38033 * symbol-summary.h (symbol_summary::m_released): New member.
38034 Do not release a summary twice. Do not allow to call finalizer
38035 for types of a summary that live in GGC memory.
38037 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38039 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38042 2017-02-02 Martin Sebor <msebor@redhat.com>
38044 PR middle-end/79275
38045 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38046 (format_string): Tighten up the range of output for non-constant
38047 strings and correct the expected range for wide non-constant strings.
38049 2017-02-02 Martin Sebor <msebor@redhat.com>
38051 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38053 PR middle-end/32003
38054 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38056 (-fdump-tree-@var): Add to index and document how to come up
38057 with pass-specific option and dump file names.
38058 (-fdump-passes): Clarify where to look for output.
38060 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38062 PR middle-end/77445
38063 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38064 statistics of the analyzed path; allow threading for speed when
38065 any of BBs along the path are optimized for speed.
38067 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38069 PR middle-end/78468
38070 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38071 settings of the virtual registers.
38074 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38076 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38077 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38080 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38082 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38083 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38085 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38087 * config/s390/s390.md: Add missing comments with the expanded
38089 * config/s390/vector.md: Likewise.
38090 * config/s390/vx-builtins.md: Likewise.
38092 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38095 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38096 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38097 conditions on a single line.
38099 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38101 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38102 __S390_VX__ to __VX__.
38104 2017-02-01 Andrew Pinski <apinski@cavium.com>
38106 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38107 stmt_info to record_stmt_cost.
38108 (vect_get_known_peeling_cost): Pass stmt_info if known to
38110 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38111 cpu_vector_cost field into
38112 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38113 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38114 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38115 splitting of scalar_stmt_cost and vec_stmt_cost.
38116 (thunderx_vector_cost): Likewise.
38117 (cortexa57_vector_cost): LIkewise.
38118 (exynosm1_vector_cost): Likewise.
38119 (xgene1_vector_cost): Likewise.
38120 (thunderx2t99_vector_cost): Improve after the splitting of the two
38122 (aarch64_builtin_vectorization_cost): Update for the splitting of
38123 scalar_stmt_cost and vec_stmt_cost.
38125 2017-02-01 Torvald Riegel <triegel@redhat.com>
38126 Richard Henderson <rth@redhat.com>
38128 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38129 conditional on existance of a fast atomic load.
38130 * optabs-query.c (can_atomic_load_p): New function.
38131 * optabs-query.h (can_atomic_load_p): Declare it.
38132 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38133 no fast atomic load is available for the particular size of access.
38134 (expand_atomic_compare_and_swap): Likewise.
38135 (expand_atomic_load): Likewise.
38136 (expand_atomic_store): Likewise.
38137 (expand_atomic_fetch_op): Likewise.
38138 * testsuite/lib/target-supports.exp
38139 (check_effective_target_sync_int_128): Remove x86 because it provides
38140 no fast atomic load.
38141 (check_effective_target_sync_int_128_runtime): Likewise.
38143 2017-02-01 Richard Biener <rguenther@suse.de>
38145 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38146 (graphite_transform_loops): Provide opt-info for optimized nests.
38147 * tree-parloop.c (parallelize_loops): Provide opt-info for
38148 parallelized loops.
38150 2017-02-01 Richard Biener <rguenther@suse.de>
38152 PR middle-end/79315
38153 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38154 was not set before.
38156 2017-02-01 Richard Biener <rguenther@suse.de>
38158 PR tree-optimization/71824
38159 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38160 Verify the loops are valid in the merged SESE region.
38161 (scop_detection::can_represent_loop_1): Check analyzing the
38162 evolution of the number of iterations in the region succeeds.
38164 2017-01-31 Ian Lance Taylor <iant@golang.org>
38166 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38167 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38169 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38171 PR preprocessor/79210
38172 * input.c (get_substring_ranges_for_loc): Replace line_width
38173 assertion with error-handling.
38175 2017-01-31 Richard Biener <rguenther@suse.de>
38177 PR tree-optimization/77318
38178 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38179 (create_pw_aff_from_tree): Take loop parameter.
38180 (add_condition_to_pbb): Pass loop of the condition to
38181 create_pw_aff_from_tree.
38183 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38185 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38186 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38188 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38192 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38194 (convert_int_to_float128): Likewise.
38195 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38196 (convert_int_to_float128): Likewise.
38197 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38198 (UNSPEC_IEEE128_CONVERT): Likewise.
38199 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38200 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38201 Use local variables for IBM extended format.
38202 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38203 (fix_trunc<mode>si2_fprs): Likewise.
38204 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38205 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38206 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38207 to know that we can now have integers of all sizes in vector
38209 (fix<uns>_<mode>di2_hw): Likewise.
38210 (float<uns>_<mode>si2_hw): Likewise.
38211 (fix_<mode>si2_hw): Likewise.
38212 (fixuns_<mode>si2_hw): Likewise.
38213 (float<uns>_<mode>di2_hw): Likewise.
38214 (float_<mode>di2_hw): Likewise.
38215 (float_<mode>si2_hw): Likewise.
38216 (floatuns_<mode>di2_hw): Likewise.
38217 (floatuns_<mode>si2_hw): Likewise.
38218 (xscvqp<su>wz_<mode>): Delete, no longer used.
38219 (xscvqp<su>dz_<mode>): Likewise.
38220 (xscv<su>dqp_<mode>): Likewise.
38221 (ieee128_mfvsrd_64bit): Likewise.
38222 (ieee128_mfvsrd_32bit): Likewise.
38223 (ieee128_mfvsrwz): Likewise.
38224 (ieee128_mtvsrw): Likewise.
38225 (ieee128_mtvsrd_64bit): Likewise.
38226 (ieee128_mtvsrd_32bit): Likewise.
38228 2017-01-31 Martin Liska <mliska@suse.cz>
38231 * ipa-prop.c (ipa_free_all_node_params): Call release method
38232 instead of ~sumbol_summary to not to trigger double times
38235 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38237 PR tree-optimization/71691
38238 * bitmap.h (class auto_bitmap): New.
38239 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38240 is_maybe_undefined instead of ssa_undefined_value_p.
38242 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38244 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38245 __S390_ARCH_LEVEL__ to __ARCH__.
38247 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38249 PR tree-optimization/79267
38250 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38251 if should_remove_lhs_p is true.
38253 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38256 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38257 (add_alignment_attribute): New.
38258 (base_type_die): Add alignment attribute.
38259 (subrange_type_die): Likewise.
38260 (modified_type_die): Likewise.
38261 (gen_array_type_die): Likewise.
38262 (gen_descr_array_type_die: Likewise.
38263 (gen_enumeration_type_die): Likewise.
38264 (gen_subprogram_die): Likewise.
38265 (gen_variable_die): Likewise.
38266 (gen_field_die): Likewise.
38267 (gen_ptr_to_mbr_type_die): Likewise.
38268 (gen_struct_or_union_type_die): Likewise.
38269 (gen_subroutine_type_die): Likewise.
38270 (gen_typedef_die): Likewise.
38271 (base_type_cmp): Compare alignment attribute.
38273 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38276 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38277 (setb_unsigned) New pattern for setb with CCUNS.
38278 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38279 subfc./subfe sequence to avoid overflow problems. Generate a
38280 shorter sequence with cmpld/setb for power9.
38281 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38282 for generating subfc. instruction.
38283 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38284 now uses this instruction.
38286 2017-01-30 Ian Lance Taylor <iant@google.com>
38289 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38290 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38292 2017-01-30 Martin Sebor <msebor@redhat.com>
38294 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38295 Move constant to the right of a relational operator.
38296 (get_mpfr_format_length, format_character, format_string): Ditto.
38297 (should_warn_p, maybe_warn): Same.
38299 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38301 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38304 * asan.c (get_translation_unit_decl): Remove function.
38305 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38307 2017-01-30 Martin Liska <mliska@suse.cz>
38309 PR gcov-profile/79259
38310 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38311 -fprofile-generate.
38313 2017-01-30 Martin Liska <mliska@suse.cz>
38316 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38317 Initialize variables with NULL value.
38319 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38322 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38324 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38326 2017-01-30 Richard Biener <rguenther@suse.de>
38328 PR tree-optimization/79276
38329 * tree-vrp.c (process_assert_insertions): Properly adjust common
38330 when removing a duplicate.
38332 2017-01-30 Richard Biener <rguenther@suse.de>
38334 PR tree-optimization/79256
38335 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38336 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38338 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38340 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38343 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38344 ("*r<noxa>sbg_<mode>_sll_bitmask")
38345 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38346 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38347 Use contiguous_bitmask_nowrap_operand.
38349 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38352 * config/rs6000/altivec.h (vec_xl): Revise #define.
38353 (vec_xst): Likewise.
38355 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38357 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38359 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38361 PR rtl-optimization/79194
38362 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38363 traps before call to bypass_conditional_jumps.
38365 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38367 PR tree-optimization/71374
38368 * lra-constraints.c (check_conflict_input_operands): New.
38369 (match_reload): Use it.
38371 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38374 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38375 account to calculate conflict_set.
38377 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38379 PR rtl-optimization/78559
38380 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38381 other_insn in combine.
38383 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38385 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38386 uint16_type_node for BT_UINT16.
38388 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38390 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38391 "RTL Tests" to menu.
38392 (GIMPLE Tests): New node.
38393 (RTL Tests): New node.
38395 2017-01-27 Richard Biener <rguenther@suse.de>
38397 PR tree-optimization/79245
38398 * tree-loop-distribution.c (distribute_loop): Apply cost
38399 modeling also to detected patterns.
38401 2017-01-27 Richard Biener <rguenther@suse.de>
38403 PR tree-optimization/71433
38404 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38405 (compare_assert_loc): New function.
38406 (process_assert_insertions): Sort and optimize assert locations
38407 to remove duplicates and push down identical assertions on
38408 edges to their destination block.
38410 2017-01-27 Richard Biener <rguenther@suse.de>
38412 PR tree-optimization/79244
38413 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38414 out SSA names even if abnormal.
38416 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38418 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38419 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38420 instead of MPFR_RNDN.
38422 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38425 * arm.c (arm_option_override): Don't call build_target_option_node
38426 until after doing all option overrides.
38427 (arm_valid_target_attribute_tree): Likewise.
38429 2017-01-27 Martin Liska <mliska@suse.cz>
38431 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38432 for {cd}tors and C++ {cd}tors.
38434 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38436 * config/s390/s390.md ("*setmem_long_and")
38437 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38439 2017-01-26 Martin Sebor <msebor@redhat.com>
38441 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38444 2017-01-26 Martin Sebor <msebor@redhat.com>
38446 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38447 HAVE_DFmode before using XFmode or DFmode.
38448 (parse_directive): Avoid using the z length modifier to avoid
38449 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38451 PR middle-end/78703
38452 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38453 to accept adjustment as an array.
38454 (get_int_range): New function.
38455 (struct directive): Make width and prec arrays.
38456 (directive::set_width, directive::set_precision): Call get_int_range.
38457 (format_integer, format_floating): Handle width and precision ranges.
38458 (format_string, parse_directive): Same.
38460 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38463 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38464 comdat_type_p set, just clone them, but keep the children in the
38468 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38469 which have direct callers with -fvar-tracking-assignments enabled
38471 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38472 inside of type units.
38474 2017-01-26 Martin Sebor <msebor@redhat.com>
38476 PR middle-end/78703
38477 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38479 (struct format_result): Replace number_chars, number_chars_min,
38480 and number_chars_max with a single member of struct result_range.
38482 (format_result::operator+=): Adjust.
38483 (struct fmtresult): Remove bounded. Handle likely and unlikely
38485 (fmtresult::adjust_for_width_or_precision): New function.
38486 (fmtresult:type_max_digits): New function.
38487 (bytes_remaining): Handle likely and unlikely counters.
38488 (min_bytes_remaining): Remove.
38489 (format_percent): Simplify.
38490 (format_integer, format_floating): Set likely and unlikely counters.
38491 (get_string_length, format_character, format_string): Same.
38492 (format_plain, should_warn_p): New function.
38493 (maybe_warn): Call should_warn_p. Update diagnostic messages
38494 and handle those for all directives, including plain strings.
38495 (format_directive): Handle likely and unlikely counters.
38496 Remove unnecessary quoting from diagnostics. Add an informational
38498 (add_bytes): Remove.
38499 (pass_sprintf_length::compute_format_length): Simplify.
38500 (try_substitute_return_value): Handle likely and unlikely counters.
38502 2017-01-26 Carl Love <cel@us.ibm.com>
38504 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38505 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38507 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38510 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38511 endianess for subregs into account.
38512 * lra-constraints.c (lra_constraints): Do risky transformations
38513 always on the first iteration.
38514 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38515 last_call_used_reg_set.
38516 (process_bb_lives): Define and use last_call_used_reg_set.
38517 * lra.c (lra): Always continue after lra_constraints on the first
38520 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38522 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38524 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38526 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38528 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38529 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38530 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38531 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38532 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38533 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38534 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38535 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38536 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38538 2017-01-26 Marek Polacek <polacek@redhat.com>
38541 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38542 for the third operand.
38544 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38546 PR middle-end/79236
38547 * omp-low.c (struct omp_context): Add simt_stmt field.
38548 (scan_omp_for): Return omp_context *.
38549 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38550 context to the _simt_ SIMD stmt.
38551 (lower_omp_for): For combined SIMD with sibling _simt_
38552 SIMD, make sure to use the same decls in _looptemp_
38553 clauses as in the sibling.
38555 2017-01-26 David Sherwood <david.sherwood@arm.com>
38557 PR middle-end/79212
38558 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38561 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38564 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38565 emit fld b; fld a; if possible.
38567 * brig-builtins.def: Update copyright years.
38568 * config/arm/arm_acle_builtins.def: Update copyright years.
38570 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38573 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38574 constraint instead of o for the stxsd instruction.
38576 2017-01-25 Carl Love <cel@us.ibm.com>
38578 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38579 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38581 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38583 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38585 2017-01-25 Richard Biener <rguenther@suse.de>
38587 PR tree-optimization/69264
38588 * target.def (vector_alignment_reachable): Improve documentation.
38589 * doc/tm.texi: Regenerate.
38590 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38592 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38593 earlier changes with respect to TYPE_USER_ALIGN.
38594 (vector_alignment_reachable_p): Likewise. Improve dumping.
38596 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38599 * config/arm/arm.md (xordi3): Force constant operand into a register
38602 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38604 * doc/invoke.texi (-fstore-merging): Correct default optimization
38605 levels at which it is enabled.
38606 (-O): Move -fstore-merging from list to...
38609 2017-01-25 Richard Biener <rguenther@suse.de>
38612 * omp-expand.c: Include debug.h.
38613 (expand_omp_taskreg): Make sure to generate early debug before
38614 outlining anything from a function.
38615 (expand_omp_target): Likewise.
38616 (grid_expand_target_grid_body): Likewise.
38618 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38621 * asan.c (get_translation_unit_decl): New function.
38622 (asan_add_global): Extract modules file name from globals
38623 TRANSLATION_UNIT_DECL name.
38625 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38628 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38629 for long calls with APCS frame and VFP.
38631 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38633 * cfg.c (original_copy_tables_initialized_p): New function.
38634 * cfg.h (original_copy_tables_initialized_p): New decl.
38635 * cfgrtl.c (relink_block_chain): Guard the call to
38636 free_original_copy_tables with a call to
38637 original_copy_tables_initialized_p.
38638 * cgraph.h (symtab_node::native_rtl_p): New decl.
38639 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38640 (symtab_node::needed_p): Don't assert for early assembly output
38641 for __RTL functions.
38642 (cgraph_node::finalize_function): Set "force_output" for __RTL
38644 (cgraph_node::analyze): Bail out early for __RTL functions.
38645 (analyze_functions): Update assertion to support __RTL functions.
38646 (cgraph_node::expand): Bail out early for __RTL functions.
38647 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38649 * function.h (struct function): Update comment for field
38651 * gimple-expr.c: Include "tree-pass.h".
38652 (gimple_has_body_p): Return false for __RTL functions.
38653 * Makefile.in (OBJS): Add run-rtl-passes.o.
38654 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38656 (gcc::pass_manager::get_clean_slate): New accessor.
38657 * passes.c: Include "insn-addr.h".
38658 (should_skip_pass_p): Add logging. Update logic for running
38659 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38660 property-provider override so it is only done for gimple passes.
38662 (skip_pass): New function.
38663 (execute_one_pass): Call skip_pass when skipping passes.
38664 * read-md.c (md_reader::read_char): Support filtering
38665 the input to a subset of line numbers.
38666 (md_reader::md_reader): Initialize fields
38667 m_first_line and m_last_line.
38668 (md_reader::read_file_fragment): New function.
38669 * read-md.h (md_reader::read_file_fragment): New decl.
38670 (md_reader::m_first_line): New field.
38671 (md_reader::m_last_line): New field.
38672 * read-rtl-function.c (function_reader::create_function): Only
38673 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38674 curr_properties. Set DECL_INITIAL to a dummy block.
38675 (read_rtl_function_body_from_file_range): New function.
38676 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38678 * run-rtl-passes.c: New file.
38679 * run-rtl-passes.h: New file.
38681 2017-01-24 Jeff Law <law@redhat.com>
38683 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38686 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38688 PR tree-optimization/79159
38689 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38690 (record_nonwrapping_iv): Improve boundary using above function if no
38691 value range information.
38693 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38694 Martin Jambor <mjambor@suse.cz>
38696 * brig-builtins.def: New file.
38697 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38698 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38699 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38700 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38701 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38702 * builtin-types.def (BT_INT8): New.
38703 (BT_INT16): Likewise.
38704 (BT_UINT8): Likewise.
38705 (BT_UINT16): Likewise.
38706 (BT_FN_ULONG): Likewise.
38707 (BT_FN_UINT_INT): Likewise.
38708 (BT_FN_UINT_ULONG): Likewise.
38709 (BT_FN_UINT_LONG): Likewise.
38710 (BT_FN_UINT_PTR): Likewise.
38711 (BT_FN_ULONG_PTR): Likewise.
38712 (BT_FN_INT8_FLOAT): Likewise.
38713 (BT_FN_INT16_FLOAT): Likewise.
38714 (BT_FN_UINT32_FLOAT): Likewise.
38715 (BT_FN_UINT16_FLOAT): Likewise.
38716 (BT_FN_UINT8_FLOAT): Likewise.
38717 (BT_FN_UINT64_FLOAT): Likewise.
38718 (BT_FN_UINT16_UINT32): Likewise.
38719 (BT_FN_UINT32_UINT16): Likewise.
38720 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38721 (BT_FN_INT_PTR_INT): Likewise.
38722 (BT_FN_UINT_PTR_UINT): Likewise.
38723 (BT_FN_LONG_PTR_LONG): Likewise.
38724 (BT_FN_ULONG_PTR_ULONG): Likewise.
38725 (BT_FN_VOID_UINT64_UINT64): Likewise.
38726 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38727 (BT_FN_INT8_INT8_INT8): Likewise.
38728 (BT_FN_INT16_INT16_INT16): Likewise.
38729 (BT_FN_INT_INT_INT): Likewise.
38730 (BT_FN_UINT_FLOAT_UINT): Likewise.
38731 (BT_FN_FLOAT_UINT_UINT): Likewise.
38732 (BT_FN_ULONG_UINT_UINT): Likewise.
38733 (BT_FN_ULONG_UINT_PTR): Likewise.
38734 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38735 (BT_FN_UINT_UINT_UINT): Likewise.
38736 (BT_FN_VOID_UINT_PTR): Likewise.
38737 (BT_FN_UINT_UINT_PTR: Likewise.
38738 (BT_FN_UINT32_UINT64_PTR): Likewise.
38739 (BT_FN_INT_INT_UINT_UINT): Likewise.
38740 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38741 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38742 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38743 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38744 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38745 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38746 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38747 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38748 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38749 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38750 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38751 * doc/frontends.texi: List BRIG FE.
38752 * doc/install.texi (Testing): Add BRIG tesring requirements.
38753 * doc/invoke.texi (Overall Options): Mention BRIG.
38754 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38756 2017-01-24 Richard Biener <rguenther@suse.de>
38758 PR translation/79208
38759 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38761 2017-01-24 Martin Jambor <mjambor@suse.cz>
38764 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38765 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38766 and known_contexts.
38768 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38770 PR middle-end/79123
38771 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38772 casts from signed to unsigned really don't have a range.
38774 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38776 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38777 GMP_RNDx for compatiblity.
38779 2017-01-24 Martin Liska <mliska@suse.cz>
38782 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38783 that would prevent us to call alloca with -1 as argument.
38785 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38787 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38788 Avoid -Wformat-security warning.
38790 2017-01-23 Andrew Pinski <apinski@cavium.com>
38792 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38795 2017-01-23 Martin Sebor <msebor@redhat.com>
38797 PR middle-end/78703
38798 * gimple-ssa-sprintf.c (warn_level): New global.
38799 (format_integer): Use it here and throughout the rest of the file.
38800 Use the same switch to compute sign as base.
38801 (maybe_warn): New function.
38802 (format_directive): Factor out warnings into maybe_warn.
38803 Add debugging output. Use warn_level.
38804 (add_bytes): Use warn_level.
38805 (pass_sprintf_length::compute_format_length): Add debugging output.
38806 (try_substitute_return_value): Same.
38807 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38809 PR middle-end/78703
38810 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38811 (struct fmtresult, format_integer, format_floating): Adjust.
38812 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38813 (get_string_length, format_string,format_directive): Same.
38814 (pass_sprintf_length::compute_format_length): Same.
38815 (try_substitute_return_value): Simplify slightly.
38817 PR middle-end/78703
38818 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38819 (fmtresult::operator+=): Outlined.
38820 (struct fmtresult): Add ctors.
38821 (struct conversion_spec): Rename...
38822 (struct directive): ...to this. Add and remove data members.
38823 (directive::set_width, directive::set_precision): New functions.
38824 (format_percent): Use fmtresult ctor.
38825 (get_width_and_precision): Remove.
38826 (format_integer): Make naming changes. Avoid computing width and
38828 (format_floating): Same. Adjust indentation.
38829 (format_character, format_none): New functions.
38830 (format_string): Moved character handling to format_character.
38831 (format_directive): Remove arguments, change return type.
38832 (parse_directive): New function.
38833 (pass_sprintf_length::compute_format_length): Move directive
38834 parsing to parse_directive.
38836 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38838 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38839 (assign_assembler_name_if_needed): ... this.
38840 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38841 (assign_assembler_name_if_needed): ... this.
38842 (free_lang_data_in_cgraph): Adjust callers.
38843 * cgraphunit.c (cgraph_node::analyze): Likewise.
38844 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38847 2017-01-23 Richard Biener <rguenther@suse.de>
38849 PR tree-optimization/79088
38850 PR tree-optimization/79188
38851 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38852 resetting loop bounds after last path deletion. Reset loop
38853 bounds of the target loop, make code match the comments.
38854 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38855 Make sure loops need no fixups.
38857 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38859 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38860 exponent support with double type for first argument.
38861 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38862 type returned by __builtin_vec_extract_sig,
38863 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38864 functions from "vector int" to "vector unsigned int" or from
38865 "vector long long int" to "vector unsigned long long int".
38866 Changed type returned by __builtin_vec_extract_exp,
38867 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38868 functions from "vector int" to "vector unsigned int" or from
38869 "vector long long int" to "vector unsigned long long int".
38870 Changed return type of __builtin_vec_test_data_class,
38871 __builtin_vec_test_data_class_sp, and
38872 __builtin_vec_test_data_class_dp from "vector int" to
38873 "vector bool int" or from "vector long long int" to "vector bool
38874 long long int" and changed second argument type from "unsigned
38875 int" to "int". Added new overloaded function forms "vector float
38876 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38877 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38878 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38879 double, vector unsigned long long int)" and "vector double
38880 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38881 long int)". Changed return type of
38882 __builtin_scalar_test_data_class and
38883 __builtin_scalar_test_data_class_sp and
38884 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38885 int" and changed second argument from "unsigned int" to "int".
38886 Changed type returned by __builtin_scalar_test_neg,
38887 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38888 from "int" to "bool int". Added new overloaded function form
38889 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38890 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38891 exponent double-precision with floating point first argument.
38892 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38893 documentation of scalar_test_data_class, scalar_test_neg,
38894 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38895 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38896 vec_test_data_class built-in functions to reflect refinements in
38897 their type signatures.
38899 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38901 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38903 (aarch64_elf_asm_destructor): Likewise.
38905 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38907 PR rtl-optimization/78634
38908 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38909 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38910 * ifcvt.c (noce_try_cmove): Add missing cost check.
38912 PR rtl-optimization/71724
38913 * combine.c (if_then_else_cond): Look for situations where it is
38914 beneficial to undo the work of one of the recursive calls.
38916 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38918 PR tree-optimization/70754
38919 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38920 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38921 combined stmt before it if not NULL.
38922 (combine_chains): Process refs reversely and compute dominance point
38925 2017-01-23 Martin Liska <mliska@suse.cz>
38927 PR tree-optimization/79196
38928 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38929 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38931 (strlen_optimize_stmt): Call the renamed function.
38933 2017-01-23 Michael Matz <matz@suse.de>
38935 PR tree-optimization/78384
38936 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38938 2017-01-23 Richard Biener <rguenther@suse.de>
38940 PR tree-optimization/79186
38941 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38942 both incoming edges before moving an assert.
38944 2017-01-23 Martin Jambor <mjambor@suse.cz>
38946 * ipa-prop.c (load_from_param_1): Removed.
38947 (load_from_unmodified_param): Bits from load_from_param_1 put back
38949 (load_from_param): Removed.
38950 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38951 with stmt. Reverted back to use of load_from_unmodified_param.
38953 2017-01-23 Martin Jambor <mjambor@suse.cz>
38956 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38957 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38958 field a pointer to garbage collected vector, mark lattices and
38959 ipcp_orig_node with GTY((skip)).
38960 (ipa_get_param_count): Adjust to descriptors being a pointer.
38961 (ipa_get_param): Likewise.
38962 (ipa_get_type): Likewise.
38963 (ipa_get_param_move_cost): Likewise.
38964 (ipa_set_param_used): Likewise.
38965 (ipa_get_controlled_uses): Likewise.
38966 (ipa_set_controlled_uses): Likewise.
38967 (ipa_is_param_used): Likewise.
38968 (ipa_node_params_t): Move into garbage collector. New methods insert
38970 (ipa_node_params_sum): Annotate wth GTY(()).
38971 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38973 (ipa_load_from_parm_agg): Adjust declaration.
38974 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38975 * ipa-profile.c (ipa_profile): Likewise.
38976 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38977 (ipa_populate_param_decls): Make descriptors parameter garbage
38979 (ipa_dump_param): Adjust to descriptors being a pointer.
38980 (ipa_alloc_node_params): Likewise.
38981 (ipa_initialize_node_params): Likewise.
38982 (load_from_param_1): Make descriptors parameter garbage collected.
38983 (load_from_unmodified_param): Likewise.
38984 (load_from_param): Likewise.
38985 (ipa_load_from_parm_agg): Likewise.
38986 (ipa_node_params::~ipa_node_params): Removed.
38987 (ipa_free_all_node_params): Remove call to delete operator.
38988 (ipa_node_params_t::insert): New.
38989 (ipa_node_params_t::remove): Likewise.
38990 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38991 copy known_csts and known_contexts vectors.
38992 (ipa_read_node_info): Adjust to descriptors being a pointer.
38993 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38995 (ipcp_transform_function): Make descriptors variable garbage
38998 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39000 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39001 * config/i386/avx512dqintrin.h: Ditto.
39002 * config/i386/avx512fintrin.h: Ditto.
39003 * config/i386/i386.c: Handle new builtins.
39004 * config/i386/i386-builtin.def: Add new builtins.
39005 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39006 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39008 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39009 Martin Liska <mliska@suse.cz>
39011 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39012 * asan.c (asan_expand_poison_ifn): Support stores and use
39013 appropriate ASAN report function.
39014 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39015 * internal-fn.def (ASAN_POISON_USE): Declare.
39016 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39017 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39018 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39019 ASAN_POISON calls w/o LHS.
39020 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39021 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39022 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39023 * gimplify.c (asan_poison_variables): Add attribute
39024 use_after_scope_memory to variables that really needs to live
39026 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39027 having the attribute.
39029 2017-01-23 Martin Liska <mliska@suse.cz>
39031 * asan.c (create_asan_shadow_var): New function.
39032 (asan_expand_poison_ifn): Likewise.
39033 * asan.h (asan_expand_poison_ifn): New declaration.
39034 * internal-fn.c (expand_ASAN_POISON): Likewise.
39035 * internal-fn.def (ASAN_POISON): New builtin.
39036 * sanopt.c (pass_sanopt::execute): Expand
39037 asan_expand_poison_ifn.
39038 * tree-inline.c (copy_decl_for_dup_finish): Make function
39040 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39041 * tree-ssa.c (is_asan_mark_p): New function.
39042 (execute_update_addresses_taken): Rewrite local variables
39043 (identified just by use-after-scope as addressable) into SSA.
39045 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39047 * doc/install.texi (Specific): opensource.apple.com uses https
39048 now. Remove trailing slash.
39050 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39052 * README.Portability: Remove note on an Irix compatibility issue.
39054 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39056 * gcov.c (INCLUDE_ALGORITHM): Define.
39057 (INCLUDE_VECTOR): Define.
39058 No longer include <vector> and <algorithm> directly.
39060 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39062 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39064 * doc/invoke.texi (Code Gen Options): Ditto.
39066 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39069 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39071 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39073 rtl-optimization/79125
39074 * cprop.c (local_cprop_pass): Handle cases where we make an
39075 unconditional trap.
39077 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39081 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39082 read from, for big endian.
39084 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39086 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39087 register pauth builtins for LP64 only.
39089 2017-01-20 Marek Polacek <polacek@redhat.com>
39092 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39095 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39097 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39099 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39100 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39101 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39103 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39106 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39107 in big-endian mode when they are not a single duplicated value.
39109 2017-01-20 Richard Biener <rguenther@suse.de>
39111 * BASE-VER: Bump to 7.0.1.
39113 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39115 * omp-low.c (omplow_simd_context): New struct. Use it...
39116 (lower_rec_simd_input_clauses): ...here and...
39117 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39118 references to idx, lane, max_vf, is_simt.
39120 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39122 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39125 2017-01-20 Martin Jambor <mjambor@suse.cz>
39127 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39128 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39130 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39131 (GTFILES): Rename hsa.c to hsa-common.c.
39132 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39133 * hsa-dump.c: Likewise.
39134 * hsa-gen.c: Likewise.
39135 * hsa-regalloc.c: Likewise.
39136 * ipa-hsa.c: Likewise.
39137 * omp-expand.c: Likewise.
39138 * omp-low.c: Likewise.
39139 * toplev.c: Likewise.
39141 2017-01-20 Marek Polacek <polacek@redhat.com>
39144 * doc/invoke.texi: Document -Wduplicated-branches.
39145 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39146 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39147 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39148 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39149 return 0 only when not OEP_LEXICOGRAPHIC.
39150 (fold_build_cleanup_point_expr): Use the expression
39151 location when building CLEANUP_POINT_EXPR.
39152 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39153 * tree.c (add_expr): Handle error_mark_node.
39155 2017-01-20 Martin Liska <mliska@suse.cz>
39158 * tree-profile.c (init_ic_make_global_vars): Do not call
39160 (gimple_init_gcov_profiler): Likewise.
39162 2017-01-20 Martin Liska <mliska@suse.cz>
39165 * cgraph.h (maybe_create_reference): Remove argument and
39167 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39169 * ipa-cp.c (create_specialized_node): Likewise.
39170 * symtab.c (symtab_node::maybe_create_reference): Handle
39171 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39173 2017-01-20 Martin Liska <mliska@suse.cz>
39175 * read-rtl-function.c (function_reader::create_function): Use
39176 build_decl instread of build_decl_stat.
39178 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39180 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39181 * config/i386/avx512dqintrin.h: Ditto.
39182 * config/i386/avx512fintrin.h: Ditto.
39183 * config/i386/i386-builtin-types.def: Add new types.
39184 * config/i386/i386.c: Handle new types.
39185 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39186 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39187 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39188 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39189 (__builtin_ia32_kshiftridi): New.
39190 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39192 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39196 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39197 define to rs6000_init_stack_protect_guard.
39198 (rs6000_init_stack_protect_guard): New function.
39200 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39201 Yunqiang Su <yunqiang.su@imgtec.com>
39203 * config.gcc (supported_defaults): Add madd4.
39204 (with_madd4): Add validation.
39205 (all_defaults): Add madd4.
39206 * config/mips/mips.opt (mmadd4): New option.
39207 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39209 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39211 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39212 (ISA_HAS_FUSED_MADD4): Likewise.
39213 * doc/invoke.texi (-mmadd4): Document the new option.
39214 * doc/install.texi (--with-madd4): Document the new option.
39216 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39218 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39219 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39220 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39221 (aarch64_init_pauth_hint_builtins): New.
39222 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39223 (aarch64_expand_builtin): Expand new builtins.
39225 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39227 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39228 * combine-stack-adj.c (no_unhandled_cfa): Handle
39229 REG_CFA_TOGGLE_RA_MANGLE.
39230 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39231 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39232 info for return address signing.
39233 (aarch64_expand_epilogue): Likewise.
39235 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39237 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39238 * config/aarch64/aarch64-protos.h
39239 (aarch64_return_address_signing_enabled): New declaration.
39240 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39242 (aarch64_expand_prologue): Sign return address before it's pushed onto
39244 (aarch64_expand_epilogue): Authenticate return address fetched from
39246 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39247 (aarch64_attributes): New function attributes for "sign-return-address".
39248 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39249 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39250 ("*do_return"): Generate combined instructions according to key index.
39251 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39252 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39254 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39255 * config/aarch64/aarch64.opt (msign-return-address=): New.
39256 * doc/extend.texi (AArch64 Function Attributes): Documents
39257 "sign-return-address=".
39258 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39260 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39262 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39263 overall option summary.
39265 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39267 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39268 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39269 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39270 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39272 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39274 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39275 -mpower9-minmax by default for -mcpu=power9.
39276 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39277 128-bit floating point.
39279 2017-01-20 Alan Modra <amodra@gmail.com>
39281 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39282 optimizing for size.
39284 2017-01-20 Alan Modra <amodra@gmail.com>
39287 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39288 for strcmp and strncmp from corresponding builtin decl.
39290 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39292 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39293 instead of i386/rtems-64.h.
39294 * config/i386/rtems-64.h: Remove.
39296 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39300 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39302 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39304 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39306 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39307 Change int to HOST_WIDE_INT.
39308 * config/aarch64/aarch64-protos.h
39309 (aarch64_simd_gen_const_vector_dup): Likewise.
39310 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39312 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39314 * langhooks-def.h (lhd_type_for_size): New decl.
39315 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39316 * langhooks.c (lhd_type_for_size): New function, taken from
39319 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39321 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39322 define_bypass for CR latency.
39323 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39324 (power9-alu2): Add define_bypass for CR latency.
39326 (power9-mul): Update insn latency.
39327 (power9-mul-compare): Update insn latency, bypass latency and remove
39330 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39332 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39334 * config/aarch64/aarch64.md
39335 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39336 aarch64_nopcrelative_literal_loads.
39337 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39339 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39341 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39342 TARGET_LOONGSON_3A.
39343 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39345 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39348 * config.gcc (supported_defaults): Add lxc1-sxc1.
39349 (with_lxc1_sxc1): Add validation.
39350 (all_defaults): Add lxc1-sxc1.
39351 * config/mips/mips.opt (mlxc1-sxc1): New option.
39352 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39354 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39355 __mips_no_lxc1_sxc1.
39356 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39357 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39358 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39360 2017-01-19 Richard Biener <rguenther@suse.de>
39362 PR tree-optimization/72488
39363 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39364 sure to restore SSA info.
39365 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39367 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39369 PR rtl-optimization/79121
39370 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39371 of the inner type when shifting an extended value.
39373 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39376 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39377 interposable aliases.
39379 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39382 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39383 Use the evmergelohi instruction.
39384 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39385 (mov_sitf_e500_subreg8_2_be): Likewise.
39386 (mov_sitf_e500_subreg12_2_le): Likewise.
39387 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39388 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39389 (mov_sitf_e500_subreg8_2_le): Likewise.
39390 (mov_sitf_e500_subreg12_2_be): Likewise.
39392 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39394 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39395 attribute from vecsimple to vecperm.
39396 (altivec_vbpermq2): Likewise.
39398 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39401 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39403 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39404 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39405 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39406 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39407 case where N arg is SIZE_MAX.
39408 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39409 (cmpstrsi): Add pattern.
39411 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39413 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39414 __builtin_vec_revb builtins.
39415 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39416 built-in functions to support generation of the ISA 3.0 XXBR<x>
39417 vector byte reverse instructions.
39418 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39419 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39420 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39421 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39422 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39423 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39424 (P9V_BUILTIN_VEC_REVB): Likewise.
39425 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39426 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39427 (p9_xxbrq_v16qi): Likewise.
39428 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39429 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39430 (p9_xxbrh_v8hi): Likewise.
39431 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39432 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39433 vec_revb built-in functions.
39435 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39437 PR rtl-optimization/78952
39438 * config/i386/i386.md (any_extract): New code iterator.
39439 (*insvqi_2): Use any_extract for source operand.
39440 (*insvqi_3): Use any_shiftrt for source operand.
39442 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39444 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39446 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39448 2017-01-18 Matthias Klose <doko@ubuntu.com>
39450 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39452 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39454 * config/rs6000/altivec.h (vec_bperm): Change #define.
39455 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39456 (altivec_vbpermq2): New define_insn.
39457 (altivec_vbpermd): Likewise.
39458 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39459 function interface.
39460 (VBPERMD): Likewise.
39461 (VBPERM): New polymorphic function interface.
39462 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39463 Add entries for P9V_BUILTIN_VEC_VBPERM.
39464 * doc/extend.texi: Add interfaces for vec_bperm.
39466 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39468 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39469 first letter of error messages.
39470 (s390_resolve_overloaded_builtin): Likewise.
39471 * config/s390/s390.c (s390_expand_builtin): Likewise.
39472 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39473 (s390_valid_target_attribute_inner_p): Likewise.
39474 * config/s390/s390.md ("tabort"): Likewise.
39476 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39478 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39479 (ISA_AVOID_DIV_HILO): New macro.
39480 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39481 (ISA_HAS_DDIV): Likewise.
39483 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39485 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39487 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39489 * doc/invoke.texi (fabi-version): Spelling fix.
39491 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39494 * doc/invoke.texi (fabi-version): Mention mangling fix for
39497 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39500 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39502 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39505 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39506 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39508 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39510 (stack_protect_test): Ditto.
39511 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39512 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39514 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39515 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39516 -mstack-protector-guard-offset=.
39517 (RS/6000 and PowerPC Options): Ditto.
39519 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39521 * config/i386/i386.h (MASK_CLASS_P): New define.
39522 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39523 there are no registers from different register sets also when
39524 mask registers are used. Update function comment.
39525 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39526 to (*k/*r) and (*k/*km) alternatives.
39528 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39530 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39531 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39532 (EH_RETURN_HANDLER_RTX): New define.
39533 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39534 Force frame pointer in EH return functions.
39535 (aarch64_expand_epilogue): Add barrier for eh_return.
39536 (aarch64_final_eh_return_addr): Remove.
39537 (aarch64_eh_return_handler_rtx): New function.
39538 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39540 (aarch64_eh_return_handler_rtx): New prototype.
39542 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39544 * config/rs6000/altivec.h (vec_rlmi): New #define.
39545 (vec_vrlnm): Likewise.
39546 (vec_rlnm): Likewise.
39547 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39548 (UNSPEC_VRLNM): Likewise.
39549 (VIlong): New mode iterator.
39550 (altivec_vrl<VI_char>mi): New define_insn.
39551 (altivec_vrl<VI_char>nm): Likewise.
39552 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39554 (VRLDNM): Likewise.
39555 (RLNM): New polymorphic function entry.
39556 (VRLWMI): New monomorphic function entry.
39557 (VRLDMI): Likewise.
39558 (RLMI): New polymorphic function entry.
39559 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39560 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39561 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39564 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39567 * dwarf2out.c (field_byte_offset): Restore the
39568 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39569 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39570 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39573 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39576 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39578 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39581 * dwarf2out.c (add_data_member_location_attribute): For constant
39582 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39583 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39584 DW_AT_byte_size attributes.
39586 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39588 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39589 after forcing to constant memory when the code model is medium.
39591 2017-01-17 Julia Koval <julia.koval@intel.com>
39594 * config/i386/avx512fintrin.h
39595 (_mm512_i32gather_ps): Change __addr type to void const*.
39596 (_mm512_mask_i32gather_ps): Ditto.
39597 (_mm512_i32gather_pd): Ditto.
39598 (_mm512_mask_i32gather_pd): Ditto.
39599 (_mm512_i64gather_ps): Ditto.
39600 (_mm512_mask_i64gather_ps): Ditto.
39601 (_mm512_i64gather_pd): Ditto.
39602 (_mm512_mask_i64gather_pd): Ditto.
39603 (_mm512_i32gather_epi32): Ditto.
39604 (_mm512_mask_i32gather_epi32): Ditto.
39605 (_mm512_i32gather_epi64): Ditto.
39606 (_mm512_mask_i32gather_epi64): Ditto.
39607 (_mm512_i64gather_epi32): Ditto.
39608 (_mm512_mask_i64gather_epi32): Ditto.
39609 (_mm512_i64gather_epi64): Ditto.
39610 (_mm512_mask_i64gather_epi64): Ditto.
39611 (_mm512_i32scatter_ps): Change __addr type to void*.
39612 (_mm512_mask_i32scatter_ps): Ditto.
39613 (_mm512_i32scatter_pd): Ditto.
39614 (_mm512_mask_i32scatter_pd): Ditto.
39615 (_mm512_i64scatter_ps): Ditto.
39616 (_mm512_mask_i64scatter_ps): Ditto.
39617 (_mm512_i64scatter_pd): Ditto.
39618 (_mm512_mask_i64scatter_pd): Ditto.
39619 (_mm512_i32scatter_epi32): Ditto.
39620 (_mm512_mask_i32scatter_epi32): Ditto.
39621 (_mm512_i32scatter_epi64): Ditto.
39622 (_mm512_mask_i32scatter_epi64): Ditto.
39623 (_mm512_i64scatter_epi32): Ditto.
39624 (_mm512_mask_i64scatter_epi32): Ditto.
39625 (_mm512_i64scatter_epi64): Ditto.
39626 (_mm512_mask_i64scatter_epi64): Ditto.
39627 * config/i386/avx512pfintrin.h
39628 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39629 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39630 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39631 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39632 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39633 (_mm512_prefetch_i32scatter_ps): Ditto.
39634 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39635 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39636 (_mm512_prefetch_i64scatter_pd): Ditto.
39637 (_mm512_prefetch_i64scatter_ps): Ditto.
39638 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39639 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39640 * config/i386/avx512vlintrin.h
39641 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39642 (_mm_mmask_i32gather_ps): Ditto.
39643 (_mm256_mmask_i32gather_pd): Ditto.
39644 (_mm_mmask_i32gather_pd): Ditto.
39645 (_mm256_mmask_i64gather_ps): Ditto.
39646 (_mm_mmask_i64gather_ps): Ditto.
39647 (_mm256_mmask_i64gather_pd): Ditto.
39648 (_mm_mmask_i64gather_pd): Ditto.
39649 (_mm256_mmask_i32gather_epi32): Ditto.
39650 (_mm_mmask_i32gather_epi32): Ditto.
39651 (_mm256_mmask_i32gather_epi64): Ditto.
39652 (_mm_mmask_i32gather_epi64): Ditto.
39653 (_mm256_mmask_i64gather_epi32): Ditto.
39654 (_mm_mmask_i64gather_epi32): Ditto.
39655 (_mm256_mmask_i64gather_epi64): Ditto.
39656 (_mm_mmask_i64gather_epi64): Ditto.
39657 (_mm256_i32scatter_ps): Change __addr type to void*.
39658 (_mm256_mask_i32scatter_ps): Ditto.
39659 (_mm_i32scatter_ps): Ditto.
39660 (_mm_mask_i32scatter_ps): Ditto.
39661 (_mm256_i32scatter_pd): Ditto.
39662 (_mm256_mask_i32scatter_pd): Ditto.
39663 (_mm_i32scatter_pd): Ditto.
39664 (_mm_mask_i32scatter_pd): Ditto.
39665 (_mm256_i64scatter_ps): Ditto.
39666 (_mm256_mask_i64scatter_ps): Ditto.
39667 (_mm_i64scatter_ps): Ditto.
39668 (_mm_mask_i64scatter_ps): Ditto.
39669 (_mm256_i64scatter_pd): Ditto.
39670 (_mm256_mask_i64scatter_pd): Ditto.
39671 (_mm_i64scatter_pd): Ditto.
39672 (_mm_mask_i64scatter_pd): Ditto.
39673 (_mm256_i32scatter_epi32): Ditto.
39674 (_mm256_mask_i32scatter_epi32): Ditto.
39675 (_mm_i32scatter_epi32): Ditto.
39676 (_mm_mask_i32scatter_epi32): Ditto.
39677 (_mm256_i32scatter_epi64): Ditto.
39678 (_mm256_mask_i32scatter_epi64): Ditto.
39679 (_mm_i32scatter_epi64): Ditto.
39680 (_mm_mask_i32scatter_epi64): Ditto.
39681 (_mm256_i64scatter_epi32): Ditto.
39682 (_mm256_mask_i64scatter_epi32): Ditto.
39683 (_mm_i64scatter_epi32): Ditto.
39684 (_mm_mask_i64scatter_epi32): Ditto.
39685 (_mm256_i64scatter_epi64): Ditto.
39686 (_mm256_mask_i64scatter_epi64): Ditto.
39687 (_mm_i64scatter_epi64): Ditto.
39688 (_mm_mask_i64scatter_epi64): Ditto.
39689 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39690 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39691 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39692 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39693 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39694 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39695 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39696 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39697 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39698 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39699 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39700 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39701 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39702 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39703 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39704 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39705 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39706 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39707 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39708 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39709 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39710 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39711 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39712 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39713 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39714 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39715 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39716 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39717 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39718 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39719 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39720 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39721 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39722 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39723 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39724 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39725 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39726 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39727 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39728 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39729 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39730 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39731 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39732 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39733 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39734 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39735 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39736 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39737 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39738 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39739 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39740 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39741 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39742 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39743 definitions accordingly.
39745 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39746 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39749 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39752 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39755 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39756 hard regs for inner regno.
39758 2017-01-17 Martin Liska <mliska@suse.cz>
39761 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39762 assumption and add comment.
39764 2017-01-17 Nathan Sidwell <nathan@acm.org>
39766 * ipa-visibility.c (localize_node): New function, broken out of ...
39767 (function_and_variable_visibility): ... here. Call it.
39769 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39771 PR middle-end/77445
39772 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39773 correctly set frequency of oudgoing edge.
39774 (duplicate_thread_path): Fix profile updating.
39776 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39779 * configure.ac: Add GCC_BASE_VER.
39780 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39781 version from BASE-VER file.
39782 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39783 (gcc.o): Depend on $(BASEVER).
39784 * common.opt (dumpfullversion): New option.
39785 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39786 * doc/invoke.texi: Document -dumpfullversion.
39787 * doc/install.texi: Document --with-gcc-major-version-only.
39788 * configure: Regenerated.
39790 2017-01-17 Richard Biener <rguenther@suse.de>
39792 PR tree-optimization/71433
39793 * tree-vrp.c (register_new_assert_for): Merge same asserts
39794 on all incoming edges.
39795 (process_assert_insertions_for): Handle insertions at the
39798 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39800 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39801 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39803 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39806 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39809 2017-01-17 Alan Modra <amodra@gmail.com>
39812 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39813 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39814 symbolic stack limit when pic.
39816 2017-01-16 Martin Sebor <msebor@redhat.com>
39818 PR tree-optimization/78608
39819 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39821 2017-01-16 Jeff Law <law@redhat.com>
39824 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39825 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39826 for several include directories that may be relative to sysroot.
39827 * config/i386/x-mingw32 (gplus_includedir): Define.
39828 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39829 (native_system_includedir): Likewise.
39830 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39831 override if TARGET_SYSTEM_ROOT is defined.
39832 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39834 PR tree-optimization/79090
39835 PR tree-optimization/33562
39836 PR tree-optimization/61912
39837 PR tree-optimization/77485
39838 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39839 and computed trims into the dump file.
39841 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39843 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39845 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39848 * gimplify.c (gimplify_init_constructor): If want_value and
39849 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39853 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39854 sequence. Formatting fixes.
39855 (doloop_optimize): Formatting fixes.
39858 * gcc.c (debug_level_greater_than_spec_func): New function.
39859 (static_spec_functions): Add debug-level-gt spec function.
39860 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39862 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39863 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39864 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39865 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39866 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39867 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39869 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39871 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39872 QImode fixups to general and mask registers only.
39874 2017-01-16 Carl Love <cel@us.ibm.com>
39876 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39877 for built-in functions
39878 vector signed char vec_nabs (vector signed char)
39879 vector signed short vec_nabs (vector signed short)
39880 vector signed int vec_nabs (vector signed int)
39881 vector signed long long vec_nabs (vector signed long long)
39882 vector float vec_nabs (vector float)
39883 vector double vec_nabs (vector double)
39884 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39886 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39887 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39888 * doc/extend.texi: Update the documentation file for the new built-in
39891 2017-01-16 Martin Sebor <msebor@redhat.com>
39893 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39896 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39898 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39899 UNSPEC_VSX__XXSPLTD to require special splat handling.
39901 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39904 * system.h: Poison strndup.
39906 2017-01-16 Alan Modra <amodra@gmail.com>
39909 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39912 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39914 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39916 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39918 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39919 call recog here. Assert that INSN_CODE (insn) is non-negative.
39921 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39924 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39926 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39927 in the currently scheduled RTL fragment.
39929 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39931 PR rtl-optimization/78751
39932 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39935 2017-01-14 Jeff Law <law@redhat.com>
39937 PR tree-optimization/79090
39938 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39939 variable length stores.
39940 (compute_trims): Delete dead assignment to *trim_tail.
39941 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39944 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39946 PR rtl-optimization/78626
39947 PR rtl-optimization/78727
39948 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39949 of a block, and split such blocks after everything else is finished.
39951 2017-01-14 Alan Modra <amodra@gmail.com>
39954 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39955 target legitimate_combined_insn.
39956 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39957 (rs6000_legitimate_combined_insn): New function.
39958 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39960 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39961 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39962 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39964 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39966 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39968 2017-01-13 Jeff Law <law@redhat.com>
39970 PR tree-optimization/33562
39971 PR tree-optimization/61912
39972 PR tree-optimization/77485
39973 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39975 (delete_dead_assignment): Likewise.
39976 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39977 statement to delete_dead_call and delete_dead_assignment.
39979 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39982 * substring-locations.c (format_warning_va): Strengthen case 1 so
39983 that both endpoints of the substring must be within the format
39984 range for just the substring to be printed.
39986 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39988 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39989 * config/i386/i386.c (ix86_target_string): Add missing options
39990 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39991 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39992 flags_other and ix86_target_other to flags2_other. Display unknown
39994 (ix86_valid_target_attribute_inner_p): Add missing options and
39995 reorder options by implied ISAs, as in ix86_target_string.
39997 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39999 * hash-table.h (hash_table::too_empty_p): New function.
40000 (hash_table::expand): Use it.
40001 (hash_table::traverse): Likewise.
40002 (hash_table::empty_slot): Use sizeof (value_type) instead of
40003 sizeof (PTR) to convert bytes to elements. Shrink the table
40004 if the current size is excessive for the current number of
40007 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40009 * ira-costs.c (record_reg_classes): Break from the inner loop
40010 early once alt_fail is known to be true. Update outer loop
40011 handling accordingly.
40013 2017-01-13 Jeff Law <law@redhat.com>
40015 * tree-ssa-dse.c (decrement_count): New function.
40016 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40017 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40018 when we know the partially dead statement is a mem* function.
40020 PR tree-optimization/61912
40021 PR tree-optimization/77485
40022 * tree-ssa-dse.c: Include expr.h.
40023 (maybe_trim_constructor_store): New function.
40024 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40026 PR tree-optimization/33562
40027 PR tree-optimization/61912
40028 PR tree-optimization/77485
40029 * doc/invoke.texi: Document new dse-max-object-size param.
40030 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40031 * tree-ssa-dse.c: Include params.h.
40032 (dse_store_status): New enum.
40033 (initialize_ao_ref_for_dse): New, partially extracted from
40035 (valid_ao_ref_for_dse, normalize_ref): New.
40036 (setup_live_bytes_from_ref, compute_trims): Likewise.
40037 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40038 (maybe_trim_partially_dead_store): Likewise.
40039 (maybe_trim_complex_store): Likewise.
40040 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40041 Track what bytes live from the original store. Return tri-state
40042 for dead, partially dead or live.
40043 (dse_dom_walker): Add constructor, destructor and new private members.
40044 (delete_dead_call, delete_dead_assignment): New extracted from
40046 (dse_optimize_stmt): Make a member of dse_dom_walker.
40047 Use initialize_ao_ref_for_dse.
40049 PR tree-optimization/33562
40050 PR tree-optimization/61912
40051 PR tree-optimization/77485
40052 * sbitmap.h (bitmap_count_bits): Prototype.
40053 (bitmap_clear_range, bitmap_set_range): Likewise.
40054 * sbitmap.c (bitmap_clear_range): New function.
40055 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40057 2017-01-13 Martin Liska <mliska@suse.cz>
40060 * function.c (set_cfun): Add new argument force.
40061 * function.h (set_cfun): Likewise.
40062 * ipa-inline-transform.c (inline_call): Use the function when
40063 strict alising from is dropped for function we inline to.
40065 2017-01-13 Richard Biener <rguenther@suse.de>
40067 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40068 for dumping GIMPLE INTEGER_CSTs.
40070 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40072 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40073 to 201112L since C++17.
40075 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40078 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40079 if -fsanitize=kernel-address is present.
40081 2017-01-13 Richard Biener <rguenther@suse.de>
40083 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40084 as _Literal ( type ) number in case usual suffixes do not
40085 preserve all information.
40087 2017-01-13 Richard Biener <rguenther@suse.de>
40089 PR tree-optimization/77283
40090 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40091 and ssa-iterators.h.
40092 (is_feasible_trace): Implement a cost model based on joiner
40095 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40098 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40099 char or short to __float128/_Float128 directly.
40101 2017-01-12 Martin Sebor <msebor@redhat.com>
40103 to -Wformat-overflow.
40104 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40105 (min_bytes_remaining): Same.
40106 (get_string_length): Same.
40107 (format_string): Same.
40108 (format_directive): Same.
40110 (pass_sprintf_length::handle_gimple_call): Same.
40112 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40114 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40115 info.nowrite calls with no lhs that can't throw. Return bool
40116 whether gsi_remove has been called or not.
40117 (pass_sprintf_length::handle_gimple_call): Return bool whether
40118 try_substitute_return_value called gsi_remove. Formatting fix.
40119 (pass_sprintf_length::execute): Don't use gsi_remove if
40120 handle_gimple_call returned true.
40123 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40124 be removed due to side-effects, don't remove following barrier nor
40125 turn the successor edge into fallthru edge.
40127 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40130 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40131 element-reversing loads and stores as not swappable.
40133 2017-01-12 Nathan Sidwell <nathan@acm.org>
40134 Nicolai Stange <nicstange@gmail.com>
40136 * combine.c (try_combine): Don't ignore result of overlap checking
40137 loop. Combine overlap & asm check into single loop.
40139 2017-01-12 Richard Biener <rguenther@suse.de>
40141 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40142 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40144 2017-01-12 Richard Biener <rguenther@suse.de>
40146 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40147 and TS_TARGET_OPTION directly derive from TS_BASE.
40148 * tree-core.h (tree_optimization_option): Derive from tree_base.
40149 (tree_target_option): Likewise.
40151 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40153 * config/i386/i386.c (memory_address_length): Increase len
40154 only when rip_relative_addr_p returns false.
40156 2017-01-11 Julia Koval <julia.koval@intel.com>
40158 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40159 (OPTION_MASK_ISA_SGX_SET): New.
40160 (ix86_handle_option): Handle OPT_msgx.
40161 * config.gcc: Added sgxintrin.h.
40162 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40163 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40164 * config/i386/i386.c (ix86_target_string): Add -msgx.
40166 (ix86_option_override_internal): Handle new options.
40167 (ix86_valid_target_attribute_inner_p): Add sgx.
40168 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40169 * config/i386/i386.opt: Add msgx.
40170 * config/i386/sgxintrin.h: New file.
40171 * config/i386/x86intrin.h: Add sgxintrin.h.
40173 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40176 * fold-const.c (maybe_nonzero_address): Return 1 for function
40178 (tree_single_nonzero_warnv_p): Don't handle function local objects
40182 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40185 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40188 * opts.c: Include "spellcheck.h"
40189 (struct string_fragment): New struct.
40190 (struct edit_distance_traits<const string_fragment &>): New
40192 (get_closest_sanitizer_option): New function.
40193 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40195 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40197 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40199 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40200 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40201 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40202 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40203 for initial die_offset if dwarf_split_debug_info.
40204 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40205 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40207 (output_skeleton_debug_sections): Formatting fix. Use
40208 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40209 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40211 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40213 * config/arm/cortex-a53.md: Add bypasses for
40214 cortex_a53_r2f_cvt.
40215 (cortex_a53_r2f): Only use for transfers.
40216 (cortex_a53_f2r): Likewise.
40217 (cortex_a53_r2f_cvt): Add reservation for conversions.
40218 (cortex_a53_f2r_cvt): Likewise.
40220 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40222 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40223 to all inlined functions, change static to extern.
40225 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40228 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40230 (arm_assemble_integer): Likewise.
40232 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40234 * config.gcc: Use new awk script to check CPU, FPU and architecture
40235 parameters for --with-... options.
40236 * config/arm/parsecpu.awk: New file
40237 * config/arm/arm-cpus.in: New file.
40238 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40240 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40242 * config/arm/t-arm: Update dependency rules.
40243 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40244 of processing .def files.
40245 * config/arm/genopt.sh: Deleted.
40246 * config/arm/gentune.sh: Deleted.
40247 * config/arm/arm-cores.def: Deleted.
40248 * config/arm/arm-arches.def: Deleted.
40249 * config/arm/arm-fpus.def: Deleted.
40250 * config/arm/arm-tune.md: Regenerated.
40251 * config/arm/arm-tables.opt: Regenerated.
40252 * config/arm/arm-cpu.h: New generated file.
40253 * config/arm/arm-cpu-data.h: New generated file.
40254 * config/arm/arm-cpu-cdata.h: New generated file.
40256 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40259 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40261 (input_varpool_node): Unpack dynamically_initialized bit.
40263 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40265 PR rtl-optimization/79032
40266 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40267 the alignment of the adjusted memory reference against that of MODE,
40268 instead of the alignment of the original memory reference.
40270 2017-01-11 Martin Jambor <mjambor@suse.cz>
40272 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40274 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40275 decorated functions.
40277 2017-01-11 Richard Biener <rguenther@suse.de>
40279 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40280 set range/nonnull info for PHI results. Do not set it on
40281 stmts marked for removal.
40283 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40285 * expr.c (store_field): In the bitfield case, fetch the return value
40286 from the registers before applying a single big-endian adjustment.
40287 Always do a final load for a BLKmode value not larger than a word.
40289 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40292 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40293 that we correctly handle column numbers greater than
40294 LINE_MAP_MAX_COLUMN_NUMBER.
40296 2017-01-10 Martin Sebor <msebor@redhat.com>
40298 PR middle-end/78245
40299 * gimple-ssa-sprintf.c (get_destination_size): Call
40300 {init,fini}object_sizes.
40301 * tree-object-size.c (addr_object_size): Adjust.
40302 (pass_through_call): Adjust.
40303 (pass_object_sizes::execute): Adjust.
40304 * tree-object-size.h (fini_object_sizes): Declare.
40306 2017-01-10 Martin Sebor <msebor@redhat.com>
40308 PR tree-optimization/78775
40309 * builtins.c (get_size_range): Move...
40310 * calls.c: ...to here.
40311 (alloc_max_size): Accept zero argument.
40312 (operand_signed_p): Remove.
40313 (maybe_warn_alloc_args_overflow): Call get_size_range.
40314 * calls.h (get_size_range): Declare.
40316 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40318 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40319 from TI's devices.csv file as of September 2016.
40320 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40322 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40324 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40325 * doc/invoke.texi: Likewise.
40326 * doc/md.texi: Likewise.
40327 * doc/objc.texi: Likewise.
40329 2017-01-10 Joshua Conner <joshconner@google.com>
40331 * config/arm/fuchsia-elf.h: New file.
40332 * config/fuchsia.h: New file.
40333 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40334 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40336 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40338 2016-01-10 Richard Biener <rguenther@suse.de>
40340 PR tree-optimization/79034
40341 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40342 Propagate out degenerate PHIs in the joiner.
40344 2017-01-10 Martin Liska <mliska@suse.cz>
40346 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40347 (sort_congruence_classes_by_decl_uid): Likewise.
40348 (sort_congruence_class_groups_by_decl_uid): Likewise.
40349 (sem_item_optimizer::merge_classes): Sort class, groups in these
40350 classes and members in the groups by DECL_UID of declarations.
40351 This would make merge operations stable.
40353 2017-01-10 Martin Liska <mliska@suse.cz>
40355 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40356 usage of m_classes_vec.
40357 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40358 (sem_item_optimizer::get_group_by_hash): Likewise.
40359 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40360 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40361 (sem_item_optimizer::verify_classes): Likewise.
40362 (sem_item_optimizer::process_cong_reduction): Likewise.
40363 (sem_item_optimizer::dump_cong_classes): Likewise.
40364 (sem_item_optimizer::merge_classes): Likewise.
40365 * ipa-icf.h (congruence_class_hash): Rename from
40366 congruence_class_group_hash. Remove declaration of m_classes_vec.
40368 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40370 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40371 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40372 * config.gcc: Add avx512vpopcntdqintrin.h.
40373 * config/i386/avx512vpopcntdqintrin.h: New.
40374 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40375 * config/i386/i386-builtin-types.def: Add new types.
40376 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40377 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40378 __builtin_ia32_vpopcountq_v8di_mask): New.
40379 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40380 __AVX512VPOPCNTDQ__.
40381 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40382 (PTA_AVX512VPOPCNTDQ): Define.
40383 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40384 TARGET_AVX512VPOPCNTDQ_P): Define.
40385 * config/i386/i386.opt: Add mavx512vpopcntdq.
40386 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40387 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40389 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40391 PR middle-end/77484
40392 * predict.def (PRED_CALL): Set to 67.
40394 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40396 * expr.c (store_field): In the bitfield case, if the value comes from
40397 a function call and is of an aggregate type returned in registers, do
40398 not modify the field mode; extract the value in all cases if the mode
40399 is BLKmode and the size is not larger than a word.
40401 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40404 * config/i386/cpuid.h: Fix undefined behavior.
40406 2017-01-04 Jeff Law <law@redhat.com>
40408 PR tree-optimization/79007
40409 PR tree-optimization/67955
40410 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40411 conservative for pt.null when flag_non_call_exceptions is on.
40413 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40415 PR translation/79019
40416 PR translation/79020
40417 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40418 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40419 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40421 * config/avr/avr.opt (maccumulate-args): Likewise.
40422 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40423 * common.opt (freport-bug): Likewise.
40424 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40425 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40426 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40427 translatable string.
40428 * config/i386/i386.c (function_value_32): Likewise.
40429 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40430 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40432 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40433 * common/config/msp430/msp430-common.c (msp430_handle_option):
40435 * symtab.c (symtab_node::verify_base): Likewise.
40436 * opts.c (set_debug_level): Likewise.
40437 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40438 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40439 missing whitespace to translatable strings.
40440 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40441 * config/sh/superh.h: Likewise.
40442 * config/i386/xopintrin.h: Likewise.
40443 * config/i386/znver1.md: Likewise.
40444 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40445 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40446 * double-int.h (struct double_int): Likewise.
40447 * double-int.c (div_and_round_double): Likewise.
40448 * wide-int.cc: Likewise.
40449 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40450 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40451 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40452 (crossjumps_occurred): ... this.
40453 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40456 PR tree-optimization/78899
40457 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40458 returning bool return struct loop *, NULL for failure and the new
40460 (versionable_outer_loop_p): Don't version outer loop if it has
40461 dont_vectorized bit set.
40462 (tree_if_conversion): When versioning outer loop, ensure
40463 tree_if_conversion is performed also on the inner loop of the
40464 non-vectorizable outer loop copy.
40465 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40466 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40467 prevent vectorization of it.
40468 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40469 the outer loop vectorization of the non-scalar version is attempted
40470 before vectorization of the inner loop in scalar version. If
40471 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40472 vectorization of its inner loop.
40473 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40474 has 2 inner loops, rename also on edges from bb whose single pred
40475 is outer_loop->header. Fix typo in function comment.
40477 2017-01-09 Martin Sebor <msebor@redhat.com>
40480 * asan.c (asan_emit_stack_protection): Increase local buffer size
40481 to avoid snprintf truncation warning.
40483 2017-01-09 Andrew Pinski <apinski@cavium.com>
40485 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40486 to reference thunderx2t99 for the tuning structure
40487 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40489 (thunderx2t99_extra_costs): This.
40490 * config/aarch64/aarch64-tune.md: Regenerate.
40491 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40492 (vulcan_addrcost_table): This.
40493 (vulcan_regmove_cost): Rename to ...
40494 (thunderx2t99_regmove_cost): This.
40495 (vulcan_vector_cost): Rename to ...
40496 (thunderx2t99_vector_cost): this.
40497 (vulcan_branch_cost): Rename to ...
40498 (thunderx2t99_branch_cost): This.
40499 (vulcan_tunings): Rename to ...
40500 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40501 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40503 2017-01-09 Martin Jambor <mjambor@suse.cz>
40507 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40508 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40509 (propagate_vr_accross_jump_function): Use the above function for all
40510 value range computations for pass-through jump functions and type
40511 converasion from explicit value range values.
40512 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40513 parameters from TYPE_ARG_TYPES.
40514 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40515 (ipa_write_node_info): Stream type of the actual argument.
40516 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40518 2017-01-09 Martin Liska <mliska@suse.cz>
40521 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40522 (lookup_compiler): Do not show error message with have_E.
40524 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40526 PR tree-optimization/78938
40527 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40528 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40529 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40530 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40533 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40535 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40538 2017-01-09 Richard Biener <rguenther@suse.de>
40540 PR tree-optimization/78997
40541 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40542 name condition properly.
40544 2017-01-09 Richard Biener <rguenther@suse.de>
40547 * dwarf2out.c (is_cxx): New overload with context.
40548 (is_naming_typedef_decl): Use it.
40550 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40552 * invoke.texi (Option Summary): Correct spacing in option lists
40553 and add line breaks to fix over-long lines.
40555 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40557 PR middle-end/17660
40559 * extend.texi (Common Variable Attributes): Add xref to GCC
40560 Internals manual to explain mode attribute keywords.
40562 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40565 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40566 and Preprocessor Options.
40567 (Options for Linking): Document -pthread here....
40568 (RS/6000 and PowerPC Options): ...not here.
40569 (Solaris 2 Options): ...or here.
40570 * doc/cppopts.texi: Document -pthread.
40572 2017-01-08 Martin Sebor <msebor@redhat.com>
40574 PR middle-end/77708
40575 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40576 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40577 New member functions.
40578 (format_directive): Used them.
40580 (pass_sprintf_length::handle_gimple_call): Same.
40581 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40582 to avoid truncation for any argument.
40583 (extract_affine_mul): Same.
40584 * tree.c (get_file_function_name): Same.
40586 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40588 PR middle-end/77484
40589 * predict.def (PRED_INDIR_CALL): Set to 86.
40591 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40593 PR preprocessor/54124
40594 * doc/cppopts.texi: Reformat -d subtable to list the full name
40595 of the options. Add cross-reference to the docs for the general
40596 compiler -d options.
40597 * doc/invoke.texi (Developer Options): Add cross-reference to the
40598 preprocessor-specific -d option documentation.
40600 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40602 PR preprocessor/13498
40603 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40604 redudant material, and reflect new command-line options.
40605 (System Headers): Likewise.
40607 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40609 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40610 -isystem, and -idirafter. Copy-edit.
40611 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40612 default for -ftrack-macro-expansion. Delete obsolete and
40613 badly-formatted implementation details about -fdebug-cpp output.
40614 * doc/cppwarnopts.texi: Copy-edit.
40616 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40619 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40620 that the transition from a max line width >= 1<<10 to narrower
40621 lines works correctly.
40623 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40625 * doc/options.texi (PerFunction): New.
40626 * opt-functions.awk (switch_flags): Map both Optimization and
40627 PerFunction to CL_OPTIMIZATION.
40628 * opth-gen.awk: Test for PerFunction flag along with
40630 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40631 it only when the latter is present. Skip those that don't in
40632 the hash function generator.
40633 * common.opt (fvar-tracking): Mark as PerFunction instead of
40635 (fvar-tracking-assignments): Likewise.
40636 (fvar-tracking-assignments-toggle): Likewise.
40637 (fvar-tracking-uninit): Likewise.
40639 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40641 PR translation/79018
40642 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40645 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40648 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40649 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40650 TARGET_LONG_JUMP_TABLE_OFFSETS.
40651 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40652 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40653 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40654 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40655 * config/m68k/m68k.md (tablejump expander): Likewise.
40656 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40657 TARGET_LONG_JUMP_TABLE_OFFSETS.
40658 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40659 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40661 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40662 David Holsgrove <david.holsgrove@xilinx.com>
40664 * common/config/microblaze/microblaze-common.c
40665 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40666 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40668 * config/microblaze/microblaze.c (microblaze_must_save_register)
40669 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40671 (microblaze_eh_return): New function.
40672 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40673 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40674 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40675 * config/microblaze/microblaze.md (eh_return): New pattern.
40677 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40679 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40680 GCC_DIAGNOSTIC_STRINGIFY): Define.
40682 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40684 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40686 * config/arm/arm.md (<mcrr>): New.
40688 * config/arm/arm.c (arm_arch5te): New.
40689 (arm_option_override): Set arm_arch5te.
40690 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40692 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40693 (arm_mcrr_qualifiers): ... this. New.
40694 (MRRC_QUALIFIERS): Define to...
40695 (arm_mrrc_qualifiers): ... this. New.
40696 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40698 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40699 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40700 (MRRCI, mrrc, MRRC): New.
40701 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40702 VUNSPEC_MRRC2): New.
40704 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40706 * config/arm/arm.md (<mcr>): New.
40708 * config/arm/arm.c (arm_coproc_builtin_available): Add
40709 support for mcr, mrc, mcr2 and mrc2.
40710 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40711 (arm_mcr_qualifiers): ... this. New.
40712 (MRC_QUALIFIERS): Define to ...
40713 (arm_mrc_qualifiers): ... this. New.
40714 (MCR_QUALIFIERS): Define to ...
40715 (arm_mcr_qualifiers): ... this. New.
40716 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40718 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40719 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40720 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40721 VUNSPEC_MRC2): New.
40723 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40725 * config/arm/arm.md (*ldc): New.
40729 * config/arm/arm.c (arm_coproc_builtin_available): Add
40730 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40731 (arm_coproc_ldc_stc_legitimate_address): New.
40732 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40733 'qualifier_const_pointer'.
40734 (LDC_QUALIFIERS): Define to...
40735 (arm_ldc_qualifiers): ... this. New.
40736 (STC_QUALIFIERS): Define to...
40737 (arm_stc_qualifiers): ... this. New.
40738 * config/arm/arm-protos.h
40739 (arm_coproc_ldc_stc_legitimate_address): New.
40740 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40741 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40742 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40743 stc2, stcl, stc2l): New.
40744 * config/arm/constraints.md (Uz): New.
40745 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40746 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40747 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40748 VUNSPEC_STC2L): New.
40750 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40752 * config/arm/arm.md (<cdp>): New.
40753 * config/arm/arm.c (neon_const_bounds): Rename this ...
40754 (arm_const_bounds): ... this.
40755 (arm_coproc_builtin_available): New.
40756 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40757 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40758 (CDP_QUALIFIERS): Define to...
40759 (arm_cdp_qualifiers): ... this. New.
40761 (arm_expand_builtin_args): Add case for 6 arguments.
40762 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40763 (arm_const_bounds): ... this.
40764 (arm_coproc_builtin_available): New.
40765 * config/arm/arm_acle.h (__arm_cdp): New.
40767 * config/arm/arm_acle_builtins.def (cdp): New.
40769 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40770 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40771 'arm_const_bounds'.
40772 * config/arm/types.md (coproc): New.
40773 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40774 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40775 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40776 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40778 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40780 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40781 (UBINOP_QUALIFIERS): New.
40783 (acle_builtin_data): New. Change comment.
40784 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40785 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40786 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40787 arm_acle_builtins.def.
40788 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40789 (arm_init_acle_builtins): New.
40790 (CRC32_BUILTIN): Remove.
40791 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40792 crc32cb, crc32ch and crc32cw.
40793 (arm_init_crc32_builtins): Remove.
40794 (arm_init_builtins): Use arm_init_acle_builtins rather
40795 than arm_init_crc32_builtins.
40796 (arm_expand_acle_builtin): New.
40797 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40798 * config/arm/arm_acle_builtins.def: New.
40800 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40802 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40803 (arm_builtin_datum): ... this.
40804 (arm_init_neon_builtin): Rename to ...
40805 (arm_init_builtin): ... this. Add a new parameters PREFIX
40806 and USE_SIG_IN_NAME.
40807 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40808 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40809 'arm_builtin_datum'.
40810 (arm_init_vfp_builtins): Likewise.
40811 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40812 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40813 (arm_expand_neon_args): Rename to ...
40814 (arm_expand_builtin_args): ... this. Rename builtin_arg
40815 enum values and differentiate between ARG_BUILTIN_MEMORY
40816 and ARG_BUILTIN_NEON_MEMORY.
40817 (arm_expand_neon_builtin_1): Rename to ...
40818 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40819 values, arm_expand_builtin_args and add bool parameter NEON.
40820 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40821 (arm_expand_vfp_builtin): Likewise.
40822 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40824 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40826 PR middle-end/77484
40827 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40828 * predict.c (tree_estimate_probability_bb): Reverse direction of
40829 polymorphic call predictor.
40831 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40833 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40834 (determine_pass_name_match): ...this new function and...
40835 (should_skip_pass_p): ...this new function.
40837 2017-01-06 Nathan Sidwell <nathan@acm.org>
40839 * ipa-visibility.c (function_and_variable_visibility): Reformat
40840 comments and long lines. Remove extrneous if.
40841 * symtab.c (symtab_node::make_decl_local): Fix code format.
40842 (symtab_node::set_section_for_node): Fix comment typo.
40844 2017-01-06 Martin Liska <mliska@suse.cz>
40847 * lra-constraints.c: Rename invariant to lra_invariant.
40848 * predict.c (set_even_probabilities): Initialize e to NULL.
40850 2017-01-05 Martin Sebor <msebor@redhat.com>
40852 PR tree-optimization/78910
40853 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40854 (format_integer): Correct off-by-one error in the handling
40855 of precision with negative numbers in signed conversions..
40857 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40859 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40861 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40863 PR tree-optimization/71016
40864 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40865 factor_out_conditional_conversion. Formatting fix.
40866 (factor_out_conditional_conversion): Add cond_stmt argument.
40867 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40868 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40871 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40873 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40874 read-rtl-function.o, and selftest-rtl.o.
40875 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40876 (selftest::aarch64_test_loading_full_dump): New function.
40877 (selftest::aarch64_run_selftests): New function.
40878 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40879 selftest::aarch64_run_selftests.
40880 * config/i386/i386.c
40881 (selftest::ix86_test_loading_dump_fragment_1): New function.
40882 (selftest::ix86_test_loading_call_insn): New function.
40883 (selftest::ix86_test_loading_full_dump): New function.
40884 (selftest::ix86_test_loading_unspec): New function.
40885 (selftest::ix86_run_selftests): Call the new functions.
40886 * emit-rtl.c (maybe_set_max_label_num): New function.
40887 * emit-rtl.h (maybe_set_max_label_num): New decl.
40888 * function.c (instantiate_decls): Guard call to
40889 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40890 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40892 * gensupport.c (gen_reader::gen_reader): Pass "false"
40893 for new "compact" param of rtx_reader.
40894 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40895 rather than an empty string for NULL strings.
40896 * read-md.c: Potentially include config.h rather than bconfig.h.
40897 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40898 (have_error): New global, copied from errors.c.
40899 (md_reader::read_name): Rename to...
40900 (md_reader::read_name_1): ...this, adding "out_loc" param,
40901 and converting "missing name or number" to returning false, rather
40903 (md_reader::read_name): Reimplement in terms of read_name_1.
40904 (md_reader::read_name_or_nil): New function.
40905 (md_reader::read_string): Handle "(nil)" by returning NULL.
40906 (md_reader::md_reader): Add new param "compact".
40907 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40908 (md_reader::read_file): New method.
40909 * read-md.h (md_reader::md_reader): Add new param "compact".
40910 (md_reader::read_file): New method.
40911 (md_reader::is_compact): New accessor.
40912 (md_reader::read_name): Convert return type from void to file_location.
40913 (md_reader::read_name_or_nil): New decl.
40914 (md_reader::read_name_1): New decl.
40915 (md_reader::m_compact): New field.
40916 (noop_reader::noop_reader): Pass "false" for new "compact" param
40918 (rtx_reader::rtx_reader): Add new "compact" param.
40919 (rtx_reader::read_rtx_operand): Make virtual and convert return
40920 type from void to rtx.
40921 (rtx_reader::read_until): New decl.
40922 (rtx_reader::handle_any_trailing_information): New virtual function.
40923 (rtx_reader::postprocess): New virtual function.
40924 (rtx_reader::finalize_string): New virtual function.
40925 (rtx_reader::m_in_call_function_usage): New field.
40926 (rtx_reader::m_reuse_rtx_by_id): New field.
40927 * read-rtl-function.c: New file.
40928 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40929 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40930 (selftest::verify_three_block_rtl_cfg): New decl.
40931 * read-rtl-function.h: New file.
40932 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40933 For host, include function.h, memmodel.h, and emit-rtl.h.
40934 (one_time_initialization): New function.
40935 (struct compact_insn_name): New struct.
40936 (compact_insn_names): New array.
40937 (find_code): Handle insn codes in compact dumps.
40938 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40939 (bind_subst_iter_and_attr): Likewise.
40940 (add_condition_to_string): Likewise.
40941 (add_condition_to_rtx): Likewise.
40942 (apply_attribute_uses): Likewise.
40943 (add_current_iterators): Likewise.
40944 (apply_iterators): Likewise.
40945 (initialize_iterators): Guard usage of apply_subst_iterator with
40946 #ifdef GENERATOR_FILE.
40947 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40948 (md_reader::read_mapping): Likewise.
40949 (add_define_attr_for_define_subst): Likewise.
40950 (add_define_subst_attr): Likewise.
40951 (read_subst_mapping): Likewise.
40952 (check_code_iterator): Likewise.
40953 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40955 (one_time_initialization): New function.
40956 (rtx_reader::read_until): New method.
40957 (read_flags): New function.
40958 (parse_reg_note_name): New function.
40959 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40960 Handle reuse_rtx ids.
40961 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40962 Add parsing support for RTL dumps, mirroring the special-cases in
40963 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40964 values, and calling handle_any_trailing_information.
40965 (rtx_reader::read_rtx_operand): Convert return type from void
40966 to rtx, returning return_rtx. Handle case 'e'. Call
40967 finalize_string on XSTR and XTMPL fields.
40968 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40969 "(nil)" values were omitted. Call the postprocess vfunc on the
40971 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40972 class ctor. Initialize m_in_call_function_usage. Call
40973 one_time_initialization.
40974 * rtl-tests.c (selftest::test_uncond_jump): Call
40975 set_new_first_and_last_insn.
40976 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40977 * selftest-rtl.c: New file.
40978 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40979 (selftest::get_insn_by_uid): New decl.
40980 * selftest-run-tests.c (selftest::run_tests): Call
40981 read_rtl_function_c_tests.
40982 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40983 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40986 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40988 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40989 operands in a special way. Assert that pos+len <= mode precision.
40991 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40993 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40994 3 argument Alias with unlimited for the negative form.
40995 (fno-vect-cost-model): Removed.
40997 2017-01-05 Martin Liska <mliska@suse.cz>
40999 * hsa-gen.c (gen_hsa_divmod): New function.
41000 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41002 2017-01-05 Martin Liska <mliska@suse.cz>
41005 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41008 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41010 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41011 small constant length operands.
41013 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41015 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41016 between loop iterations.
41018 2017-01-05 Martin Liska <mliska@suse.cz>
41021 * gimplify.c (gimplify_decl_expr): Compare to
41022 asan_poisoned_variables instread of checking flags.
41023 (gimplify_target_expr): Likewise.
41024 (gimplify_expr): Likewise.
41025 (gimplify_function_tree): Conditionally initialize
41026 asan_poisoned_variables.
41028 2017-01-04 Jeff Law <law@redhat.com>
41030 PR tree-optimizatin/78812
41031 * rtl.h (contains_mem_rtx_p): Prototype.
41032 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41033 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41034 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41035 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41036 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41038 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41040 * input.c (assert_char_at_range): Default-initialize actual_range.
41042 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41044 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41045 to match the caller.
41047 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41049 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41050 insns after final jump in test to emit dummy move.
41052 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41054 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41055 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41057 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41059 * multiple_target.c (create_dispatcher_calls): Init e_next.
41060 * tree-ssa-loop-split.c (split_loop): Init border.
41061 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41064 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41069 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41070 (altivec_register_operand): Do not return true if the operand
41071 contains a SUBREG mixing SImode and SFmode.
41072 (vsx_register_operand): Likewise.
41073 (vsx_reg_sfsubreg_ok): New predicate.
41074 (vfloat_operand): Do not return true if the operand contains a
41075 SUBREG mixing SImode and SFmode.
41076 (vint_operand): Likewise.
41077 (vlogical_operand): Likewise.
41078 (gpc_reg_operand): Likewise.
41079 (int_reg_operand): Likewise.
41080 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41081 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41082 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41084 (rs6000_emit_move_si_sf_subreg): New helper function.
41085 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41086 fixup SUBREGs involving SImode and SFmode.
41087 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41088 numbers for the new peephole2 optimization.
41089 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41090 the GLIBC math library that do AND/IOR/XOR operations on single
41091 precision floating point.
41092 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41093 target macros to say whether we need to avoid SUBREGs mixing
41095 (TARGET_ALLOW_SF_SUBREG): Likewise.
41096 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41097 (UNSPEC_SI_FROM_SF): Likewise.
41098 (iorxor): Change spacing.
41099 (and_ior_xor): New iterator for AND, IOR, and XOR.
41100 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41101 (movdi_from_sf_zero_ext): Likewise.
41102 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41103 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41104 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41105 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41106 (fms<mode>4): Likewise.
41107 (fnma<mode>4): Likewise.
41108 (fnms<mode>4): Likewise.
41109 (nfma<mode>4): Likewise.
41110 (nfms<mode>4): Likewise.
41112 2017-01-04 Marek Polacek <polacek@redhat.com>
41115 * doc/invoke.texi: Document -Wpointer-compare.
41117 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41119 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41122 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41123 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41126 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41129 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41130 documentation of the powerpc_popcntb_ok attribute.
41131 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41132 code to issue warning messages if a requested CPU configuration is
41133 not supported by the binary (assembler and loader) toolchain.
41134 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41135 made to define a built-in function that has been disabled.
41136 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41137 made to define a built-in function that has been disabled.
41138 (altivec_init_builtins): Add comment explaining why definition
41139 of the DST built-in functions is not preceded by an assertion
41140 check. Add assertions to prevent ICE if attempts are made to
41141 define an altivec predicate or an abs* built-in function that has
41143 (htm_init_builtins): Add comment explaining why definition of the
41144 htm built-in functions is not preceded by an assertion check.
41146 2017-01-04 Jeff Law <law@redhat.com>
41148 PR tree-optimizatin/67955
41149 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41150 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41151 the points-to solution does not include pt_null. Use DECL_PT_UID
41154 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41156 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41157 Use gen_int_mode instead of gen_lopwart for const_int operands.
41159 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41161 PR tree-optimization/71563
41162 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41163 out of range value - has low bits known to be zero.
41165 2017-01-04 Alan Modra <amodra@gmail.com>
41167 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41168 * configure: Regenerate.
41169 * config.in: Regenerate.
41171 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41174 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41175 a substring of the message, but strcmp with the whole message. Ifdef
41176 ENABLE_NLS, translate the message first using dgettext.
41178 2017-01-03 Jeff Law <law@redhat.com>
41180 PR tree-optimizatin/78856
41181 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41182 (mark_threaded_blocks): Remove code to truncate thread paths that
41183 cross multiple loop headers. Instead invalidate the cached loop
41184 iteration information and handle case of a thread path walking
41185 into an irreducible region.
41187 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41190 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41191 assertions. Add support for doing the signbit if the IEEE 128-bit
41192 floating point value is in a GPR.
41193 * config/rs6000/rs6000.md (Fsignbit): Delete.
41194 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41195 Update the length attribute if the value is in a GPR.
41196 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41197 the sign or zero extension instruction, since the value is always 0/1.
41198 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41201 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41202 extracting SImode to a GPR register so that we can generate a
41203 store, limit the vector to be in a traditional Altivec register
41204 for the vextuwrx instruction.
41206 2017-01-03 Ian Lance Taylor <iant@google.com>
41208 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41210 2017-01-03 Martin Sebor <msebor@redhat.com>
41212 PR tree-optimization/78696
41213 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41214 precision. Use MPFR for %f for greater fidelity. Correct handling
41216 (pass_sprintf_length::compute_format_length): Set width and precision
41217 specified by asrerisk to void_node for vararg functions.
41218 (try_substitute_return_value): Adjust dump output.
41220 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41222 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41224 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41226 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41227 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41228 -mlra/-mno-lra was passed to the compiler.
41230 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41232 PR rtl-optimization/65618
41233 * emit-rtl.c (try_split): Move initialization of "before" and
41234 "after" to just before the call to emit_insn_after_setloc.
41236 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41238 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41240 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41242 * dwarf2out.c (gen_enumeration_type_die): When
41243 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41245 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41247 PR tree-optimization/78965
41248 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41249 Change first argument from const call_info & to call_info &. For %n
41250 set info.nowrite to false.
41252 PR middle-end/78901
41253 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41254 possibly throwing calls.
41256 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41257 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41258 and fns handling, rather than in a separate case SSA_NAME.
41260 2017-01-02 Jeff Law <law@redhat.com>
41262 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41263 fixes for first_period and second_period variables.
41265 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41268 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41269 (*insvqi_1): New insn pattern.
41270 (*insvqi_1_mem_rex64): Ditto.
41271 (*insvqi_2): Ditto.
41272 (*insvqi_3): Rename from *insvqi.
41274 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41276 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41278 * doc/cfg.texi (Edges): Remove reference to Java.
41279 (Maintaining the CFG): Ditto.
41281 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41283 PR middle-end/77674
41284 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41285 transparent aliases.
41287 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41289 PR middle-end/77484
41290 * predict.def (PRED_CALL): Update hitrate.
41291 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41292 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41293 into direct/indirect/polymorphic variants.
41295 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41297 Update copyright years.
41299 * gcc.c (process_command): Update copyright notice dates.
41300 * gcov-dump.c (print_version): Ditto.
41301 * gcov.c (print_version): Ditto.
41302 * gcov-tool.c (print_version): Ditto.
41303 * gengtype.c (create_file): Ditto.
41304 * doc/cpp.texi: Bump @copying's copyright year.
41305 * doc/cppinternals.texi: Ditto.
41306 * doc/gcc.texi: Ditto.
41307 * doc/gccint.texi: Ditto.
41308 * doc/gcov.texi: Ditto.
41309 * doc/install.texi: Ditto.
41310 * doc/invoke.texi: Ditto.
41312 Copyright (C) 2017 Free Software Foundation, Inc.
41314 Copying and distribution of this file, with or without modification,
41315 are permitted in any medium without royalty provided the copyright
41316 notice and this notice are preserved.