1 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
4 support for Cortex-A55 and Cortex-A75.
6 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8 * common/config/arm/arm-common.c (compare_opt_names): Add function
9 comment. Use strcmp instead of manual loop.
11 2017-12-21 Martin Liska <mliska@suse.cz>
14 * gcov-dump.c (dump_gcov_file): Do not read info about
15 support_unexecuted_blocks for gcda files.
17 2017-12-21 Jakub Jelinek <jakub@redhat.com>
19 PR rtl-optimization/82973
20 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
21 (valid_for_const_vector_p): ... this.
22 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
23 (valid_for_const_vector_p): ... this. Adjust function comment.
24 (gen_vec_duplicate): Adjust caller.
25 * optabs.c (expand_vector_broadcast): Likewise.
26 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
27 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
29 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
30 instead of GET_CODE == CONST_FIXED.
31 (simplify_subreg): Use CONST_FIXED_P macro instead of
32 GET_CODE == CONST_FIXED.
35 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
36 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
37 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
38 (ix86_option_override_internal): Adjust for
39 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
40 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
41 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
42 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
43 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
44 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
45 def_builtin2 for CET builtins.
46 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
47 up formatting in IX86_BUILTIN_RDPID code.
48 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
49 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
50 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
51 ix86_isa_flags2 to ix86_isa_flags.
52 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
54 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
55 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
56 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
58 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
59 Or in OPTION_MASK_ISA_AVX512F_SET.
60 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
61 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
62 (ix86_handle_option): Adjust for
63 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
64 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
66 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
67 Alan Hayward <alan.hayward@arm.com>
68 David Sherwood <david.sherwood@arm.com>
70 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
71 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
72 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
73 Track polynomial offsets.
75 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
76 Alan Hayward <alan.hayward@arm.com>
77 David Sherwood <david.sherwood@arm.com>
79 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
80 Treat drb->init as a poly_int. Fail if its misalignment wrt
81 vector_alignment isn't known.
83 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
84 Alan Hayward <alan.hayward@arm.com>
85 David Sherwood <david.sherwood@arm.com>
87 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
89 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
90 (vect_loop_versioning): Take the loop versioning threshold as a
92 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
93 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
95 (vect_analyze_loop_2): Compute the loop versioning threshold
96 whenever loop versioning is needed, and store it in the new
97 field rather than combining it with the cost model threshold.
98 (vect_transform_loop): Update call to vect_loop_versioning.
99 Try to combine the loop versioning and cost thresholds here.
101 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
102 Alan Hayward <alan.hayward@arm.com>
103 David Sherwood <david.sherwood@arm.com>
105 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
106 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
107 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
108 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
109 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
110 (group_compare_offset): Update accordingly.
111 (split_small_address_groups_p): Likewise.
112 (record_use): Take addr_offset as a poly_uint64 rather than
113 an unsigned HOST_WIDE_INT.
114 (strip_offset): Return the offset as a poly_uint64 rather than
115 an unsigned HOST_WIDE_INT.
116 (record_group_use, split_address_groups): Track polynomial offsets.
117 (add_iv_candidate_for_use): Likewise.
118 (addr_offset_valid_p): Take the offset as a poly_int64 rather
119 than a HOST_WIDE_INT.
120 (strip_offset_1): Return the offset as a poly_int64 rather than
123 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
124 Alan Hayward <alan.hayward@arm.com>
125 David Sherwood <david.sherwood@arm.com>
127 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
128 rather than a HOST_WIDE_INT.
129 * tree.c (get_binfo_at_offset): Likewise.
131 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
132 Alan Hayward <alan.hayward@arm.com>
133 David Sherwood <david.sherwood@arm.com>
135 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
136 rather than a HOST_WIDE_INT.
137 * tree-sra.c (build_ref_for_offset): Likewise.
139 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
140 Alan Hayward <alan.hayward@arm.com>
141 David Sherwood <david.sherwood@arm.com>
143 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
145 * tree.c (mem_ref_offset): Likewise.
146 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
147 * builtins.c (get_object_alignment_2): Likewise.
148 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
149 * gimple-fold.c (get_base_constructor): Likewise.
150 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
151 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
153 * ipa-polymorphic-call.c
154 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
155 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
156 (get_ancestor_addr_info): Likewise.
157 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
158 * match.pd: Likewise.
159 * tree-data-ref.c (dr_analyze_innermost): Likewise.
160 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
161 * tree-eh.c (tree_could_trap_p): Likewise.
162 * tree-object-size.c (addr_object_size): Likewise.
163 * tree-ssa-address.c (copy_ref_info): Likewise.
164 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
165 (indirect_refs_may_alias_p): Likewise.
166 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
167 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
168 (non_rewritable_mem_ref_base): Likewise.
169 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
170 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
171 * varasm.c (decode_addr_const): Likewise.
173 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
174 Alan Hayward <alan.hayward@arm.com>
175 David Sherwood <david.sherwood@arm.com>
177 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
178 offsets for MEM_REFs.
180 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
181 Alan Hayward <alan.hayward@arm.com>
182 David Sherwood <david.sherwood@arm.com>
184 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
185 as a poly_uint64 rather than an unsigned int.
186 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
188 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
189 Alan Hayward <alan.hayward@arm.com>
190 David Sherwood <david.sherwood@arm.com>
192 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
194 (decode_addr_const): Update accordingly.
196 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
197 Alan Hayward <alan.hayward@arm.com>
198 David Sherwood <david.sherwood@arm.com>
200 * tree.h (bit_field_size, bit_field_offset): New functions.
201 * hsa-gen.c (gen_hsa_addr): Use them.
202 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
203 (simplify_vector_constructor): Likewise.
204 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
205 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
206 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
207 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
210 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
211 Alan Hayward <alan.hayward@arm.com>
212 David Sherwood <david.sherwood@arm.com>
214 * expr.h (emit_group_load, emit_group_load_into_temps)
215 (emit_group_store): Take the size as a poly_int64 rather than an int.
216 * expr.c (emit_group_load_1, emit_group_load): Likewise.
217 (emit_group_load_into_temp, emit_group_store): Likewise.
219 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
220 Alan Hayward <alan.hayward@arm.com>
221 David Sherwood <david.sherwood@arm.com>
223 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
224 an unsigned int to a poly_uint64.
225 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
226 sizes as poly_uint64s rather than unsigned ints.
227 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
230 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
231 Alan Hayward <alan.hayward@arm.com>
232 David Sherwood <david.sherwood@arm.com>
234 * lra-constraints.c (emit_inc): Change inc_amount from an int
237 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
238 Alan Hayward <alan.hayward@arm.com>
239 David Sherwood <david.sherwood@arm.com>
241 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
243 (add_stack_var, stack_var_cmp, partition_stack_vars)
244 (dump_stack_var_partition): Update accordingly.
245 (alloc_stack_frame_space): Take the size as a poly_int64 rather
246 than a HOST_WIDE_INT.
247 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
248 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
249 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
250 rather than a HOST_WIDE_INT.
252 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
253 Alan Hayward <alan.hayward@arm.com>
254 David Sherwood <david.sherwood@arm.com>
256 * target.def (return_pops_args): Treat both the input and output
257 sizes as poly_int64s rather than HOST_WIDE_INTS.
258 * targhooks.h (default_return_pops_args): Update accordingly.
259 * targhooks.c (default_return_pops_args): Likewise.
260 * doc/tm.texi: Regenerate.
261 * emit-rtl.h (incoming_args): Change pops_args, size and
262 outgoing_args_size from int to poly_int64_pod.
263 * function.h (expr_status): Change x_pending_stack_adjust and
264 x_stack_pointer_delta from int to poly_int64.
265 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
266 (ARGS_SIZE_RTX): Update accordingly.
267 * calls.c (highest_outgoing_arg_in_use): Change from int to
269 (stack_usage_watermark, stored_args_watermark): New variables.
270 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
271 (emit_call_1): Change the stack_size and rounded_stack_size
272 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
274 (save_fixed_argument_area): Check stack_usage_watermark.
275 (initialize_argument_information): Change old_pending_adj from
276 a HOST_WIDE_INT * to a poly_int64_pod *.
277 (compute_argument_block_size): Return the size as a poly_int64
279 (finalize_must_preallocate): Track polynomial argument sizes.
280 (compute_argument_addresses): Likewise.
281 (internal_arg_pointer_based_exp): Track polynomial offsets.
282 (mem_overlaps_already_clobbered_arg_p): Rename to...
283 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
284 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
285 Check stored_args_used_watermark.
286 (load_register_parameters): Update accordingly.
287 (check_sibcall_argument_overlap_1): Likewise.
288 (combine_pending_stack_adjustment_and_call): Take the unadjusted
289 args size as a poly_int64 rather than an int. Return a bool
290 indicating whether the optimization was possible and return
291 the new adjustment by reference.
292 (check_sibcall_argument_overlap): Track polynomail argument sizes.
293 Update stored_args_watermark.
294 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
295 (expand_call): Likewise. Maintain stack_usage_watermark and
296 stored_args_watermark. Update calls to
297 combine_pending_stack_adjustment_and_call.
298 (emit_library_call_value_1): Handle polynomial argument sizes.
299 Call stack_region_maybe_used_p and mark_stack_region_used.
300 Maintain stack_usage_watermark.
301 (store_one_arg): Likewise. Update call to
302 mem_overlaps_already_clobbered_arg_p.
303 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
305 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
306 * config/microblaze/microblaze.c (microblaze_function_prologue):
308 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
309 TARGET_RETURN_POPS_ARGS interface.
310 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
312 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
313 * config/i386/i386.c (ix86_return_pops_args): Update for new
314 TARGET_RETURN_POPS_ARGS interface.
315 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
316 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
317 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
318 TARGET_RETURN_POPS_ARGS interface.
319 * config/vax/vax.c (vax_return_pops_args): Likewise.
320 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
321 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
322 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
323 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
324 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
325 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
326 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
327 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
328 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
329 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
330 and x_stack_pointer_delta from int to poly_int64.
331 * dojump.c (do_pending_stack_adjust): Update accordingly.
332 * explow.c (allocate_dynamic_stack_space): Handle polynomial
333 stack_pointer_deltas.
334 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
335 (pad_to_arg_alignment): Track polynomial offsets.
336 (assign_parm_find_stack_rtl): Likewise.
337 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
338 * toplev.c (output_stack_usage): Update reference to
339 current_function_pushed_stack_size.
341 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
342 Alan Hayward <alan.hayward@arm.com>
343 David Sherwood <david.sherwood@arm.com>
345 * function.c (in_arg_offset, var_offset, dynamic_offset)
346 (out_arg_offset, cfa_offset): Change from int to poly_int64.
347 (instantiate_new_reg): Return the new offset as a poly_int64_pod
348 rather than a HOST_WIDE_INT.
349 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
350 (instantiate_virtual_regs_in_insn): Likewise.
352 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
353 Alan Hayward <alan.hayward@arm.com>
354 David Sherwood <david.sherwood@arm.com>
356 * rtl.h (get_args_size, add_args_size_note): New functions.
357 (find_args_size_adjust): Return a poly_int64 rather than a
359 (fixup_args_size_notes): Likewise. Make the same change to the
360 end_args_size parameter.
361 * rtlanal.c (get_args_size, add_args_size_note): New functions.
362 * builtins.c (expand_builtin_trap): Use add_args_size_note.
363 * calls.c (emit_call_1): Likewise.
364 * explow.c (adjust_stack_1): Likewise.
365 * cfgcleanup.c (old_insns_match_p): Update use of
366 find_args_size_adjust.
367 * combine.c (distribute_notes): Track polynomial arg sizes.
368 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
369 end_true_args_size, beg_delay_args_size and end_delay_args_size
370 from HOST_WIDE_INT to poly_int64.
371 (add_cfi_args_size): Take the args_size as a poly_int64 rather
372 than a HOST_WIDE_INT.
373 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
374 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
375 polynomial arg sizes.
376 * emit-rtl.c (try_split): Use get_args_size.
377 * recog.c (peep2_attempt): Likewise.
378 * reload1.c (reload_as_needed): Likewise.
379 * expr.c (find_args_size_adjust): Return the adjustment as a
380 poly_int64 rather than a HOST_WIDE_INT.
381 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
382 to a poly_int64 and change the return type in the same way.
383 (emit_single_push_insn): Track polynomial arg sizes.
385 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
386 Alan Hayward <alan.hayward@arm.com>
387 David Sherwood <david.sherwood@arm.com>
389 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
390 from HOST_WIDE_INT to poly_int64.
391 * expr.c (push_block, emit_push_insn): Likewise.
393 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
394 Alan Hayward <alan.hayward@arm.com>
395 David Sherwood <david.sherwood@arm.com>
397 * function.h (frame_space): Change start and length from HOST_WIDE_INT
399 (get_frame_size): Return the size as a poly_int64 rather than a
401 (frame_offset_overflow): Take the offset as a poly_int64 rather
402 than a HOST_WIDE_INT.
403 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
404 (assign_stack_temp): Likewise for the size.
405 * function.c (get_frame_size): Return a poly_int64 rather than
407 (frame_offset_overflow): Take the offset as a poly_int64 rather
408 than a HOST_WIDE_INT.
409 (try_fit_stack_local): Take the start, length and size as poly_int64s
410 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
411 rather than a HOST_WIDE_INT.
412 (add_frame_space): Take the start and end as poly_int64s rather than
414 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
415 (assign_stack_temp): Likewise for the size.
416 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
418 (find_temp_slot_from_address): Handle polynomial offsets.
419 (combine_temp_slots): Likewise.
420 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
422 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
423 poly_int64 rather than a HOST_WIDE_INT.
424 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
425 than a HOST_WIDE_INT.
426 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
427 polynomial frame offsets.
428 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
429 as a poly_int64 rather than an int.
430 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
431 * config/v850/v850-protos.h (compute_frame_size): Likewise.
432 * config/v850/v850.c (compute_frame_size): Likewise.
433 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
434 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
435 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
436 * config/pa/pa.c (pa_compute_frame_size): Likewise.
437 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
438 rather than a HOST_WIDE_INT.
439 * explow.c (get_dynamic_stack_base): Likewise.
440 * final.c (final_start_function): Use the constant lower bound
441 of the frame size for -Wframe-larger-than.
442 * ira.c (do_reload): Adjust for new get_frame_size return type.
443 * lra.c (lra): Likewise.
444 * reload1.c (reload): Likewise.
445 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
446 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
447 * rtlanal.c (get_initial_register_offset): Return the offset as
448 a poly_int64 rather than a HOST_WIDE_INT.
450 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
451 Alan Hayward <alan.hayward@arm.com>
452 David Sherwood <david.sherwood@arm.com>
454 * reload1.c (elim_table): Change initial_offset, offset and
455 previous_offset from HOST_WIDE_INT to poly_int64_pod.
456 (offsets_at): Change the target array's element type from
457 HOST_WIDE_INT to poly_int64_pod.
458 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
459 (elimination_costs_in_insn, update_eliminable_offsets)
460 (verify_initial_elim_offsets, set_offsets_for_label)
461 (init_eliminable_invariants): Update after above changes.
463 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
464 Alan Hayward <alan.hayward@arm.com>
465 David Sherwood <david.sherwood@arm.com>
467 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
468 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
469 (decomposition): Change start and end from HOST_WIDE_INT
471 (decompose, immune_p): Update accordingly.
472 (find_inc_amount): Return a poly_int64 rather than an int.
473 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
476 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
477 Alan Hayward <alan.hayward@arm.com>
478 David Sherwood <david.sherwood@arm.com>
480 * tree.h (get_inner_reference): Return the bitsize and bitpos
481 as poly_int64_pods rather than HOST_WIDE_INT.
482 * fold-const.h (ptr_difference_const): Return the pointer difference
483 as a poly_int64_pod rather than a HOST_WIDE_INT.
484 * expr.c (get_inner_reference): Return the bitsize and bitpos
485 as poly_int64_pods rather than HOST_WIDE_INT.
486 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
488 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
489 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
490 (optimize_bit_field_compare): Update call to get_inner_reference.
491 (decode_field_reference): Likewise.
492 (fold_unary_loc): Track polynomial offsets and sizes.
493 (split_address_to_core_and_offset): Return the bitpos as a
494 poly_int64_pod rather than a HOST_WIDE_INT.
495 (ptr_difference_const): Likewise for the pointer difference.
496 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
497 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
498 * dbxout.c (dbxout_expand_expr): Likewise.
499 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
500 (loc_list_from_tree_1, fortran_common): Likewise.
501 * gimple-laddress.c (pass_laddress::execute): Likewise.
502 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
503 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
504 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
505 * tree-affine.c (tree_to_aff_combination): Likewise.
506 (get_inner_reference_aff): Likewise.
507 * tree-data-ref.c (split_constant_offset_1): Likewise.
508 (dr_analyze_innermost): Likewise.
509 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
510 * tree-sra.c (ipa_sra_check_caller): Likewise.
511 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
512 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
513 (instrument_bool_enum_load, instrument_object_size): Likewise.
514 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
515 to get_inner_reference.
516 * hsa-gen.c (gen_hsa_addr): Likewise.
517 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
518 * tsan.c (instrument_expr): Likewise.
519 * match.pd: Update call to ptr_difference_const.
521 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
522 Alan Hayward <alan.hayward@arm.com>
523 David Sherwood <david.sherwood@arm.com>
525 * fold-const.c (fold_comparison): Track sizes and offsets as
526 poly_int64s rather than HOST_WIDE_INTs when folding address
529 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
530 Alan Hayward <alan.hayward@arm.com>
531 David Sherwood <david.sherwood@arm.com>
533 * expr.h (get_bit_range): Return the bitstart and bitend as
534 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
535 as a poly_int64 rather than a HOST_WIDE_INT.
536 * expr.c (get_bit_range): Likewise.
537 (expand_assignment): Update call accordingly.
538 * fold-const.c (optimize_bit_field_compare): Likewise.
540 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
541 Alan Hayward <alan.hayward@arm.com>
542 David Sherwood <david.sherwood@arm.com>
544 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
545 (aarch64_legitimate_address_p): Use it instead of an rtx code,
546 as an optional final parameter.
547 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
548 (aarch64_legitimate_address_p): Likewise.
549 (aarch64_print_address_internal): Take an aarch64_addr_query_type
550 instead of an rtx code.
551 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
552 (aarch64_legitimate_address_hook_p): Likewise.
553 (aarch64_print_ldpstp_address): Likewise.
554 (aarch64_print_operand_address): Likewise.
555 (aarch64_address_cost): Likewise.
556 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
557 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
558 (aarch64_mem_pair_lanes_operand): Likewise.
560 2017-12-20 Richard Biener <rguenther@suse.de>
562 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
563 update_stmt_if_modified.
565 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
567 PR tree-optimization/83491
568 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
569 before walking uses. Improve coding style and comments.
571 2017-12-20 Tom de Vries <tom@codesourcery.com>
573 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
575 2017-12-20 Jakub Jelinek <jakub@redhat.com>
578 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
580 (pass_ipa_free_fn_summary): Add small_p private data member,
581 initialize to false in the ctor.
582 (pass_ipa_free_fn_summary::clone,
583 pass_ipa_free_fn_summary::set_pass_param,
584 pass_ipa_free_fn_summary::gate): New methods.
585 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
586 | TODO_dump_symtab if small_p.
587 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
588 entry and add another instance of the pass with false parm after
590 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
591 ipa_free_fn_summary here.
593 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
595 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
597 2017-12-20 Martin Sebor <msebor@redhat.com>
600 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
601 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
602 (expand_builtin_strncmp): Same.
604 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
607 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
608 allowed debug stmts before labels.
609 (expand_gimple_basic_block): Likewise.
610 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
611 * gimple-iterator.h (gsi_after_labels): Likewise.
612 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
613 rename reused variable, and simplify using gsi_move_before.
614 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
615 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
616 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
617 (gimple_verify_flow_info, gimple_block_label): Likewise.
618 (make_blocks): Move debug markers after adjacent labels.
619 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
620 allowed debug insns outside blocks.
621 * df-scan.c (df_insn_delete): Likewise.
622 * lra-constraints.c (update_ebb_live_info): Likewise.
623 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
624 (vt_initialize, delete_vta_debug_insns): Likewise.
625 (reemit_marker_as_note): Drop BB parm. Adjust callers.
627 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
628 Alan Hayward <alan.hayward@arm.com>
629 David Sherwood <david.sherwood@arm.com>
631 * poly-int-types.h (round_down_to_byte_boundary): New macro.
632 (round_up_to_byte_boundary): Likewise.
633 * expr.h (get_bit_range): Add temporary shim.
634 * gimple-ssa-store-merging.c (store_operand_info): Change the
635 bitsize, bitpos, bitregion_start and bitregion_end fields from
636 unsigned HOST_WIDE_INT to poly_uint64.
637 (merged_store_group): Likewise load_align_base.
638 (compatible_load_p, compatible_load_p): Update accordingly.
639 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
640 (split_group, imm_store_chain_info::output_merged_store): Likewise.
641 (mem_valid_for_store_merging): Return the bitsize, bitpos,
642 bitregion_start and bitregion_end as poly_uint64s rather than
643 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
644 (handled_load): Take the bitsize, bitpos,
645 bitregion_start and bitregion_end as poly_uint64s rather than
646 unsigned HOST_WIDE_INTs.
647 (pass_store_merging::process_store): Update call to
648 mem_valid_for_store_merging.
650 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
651 Alan Hayward <alan.hayward@arm.com>
652 David Sherwood <david.sherwood@arm.com>
654 * builtins.c (get_object_alignment_2): Track polynomial offsets
655 and sizes. Update the alignment handling.
657 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
658 Alan Hayward <alan.hayward@arm.com>
659 David Sherwood <david.sherwood@arm.com>
661 * tree.h (get_inner_reference): Add a version that returns the
662 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
663 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
664 the case in which bitpos is not associated with the first byte.
666 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
667 Alan Hayward <alan.hayward@arm.com>
668 David Sherwood <david.sherwood@arm.com>
670 * tree-affine.h (get_inner_reference_aff): Return the size as a
672 * tree-affine.c (get_inner_reference_aff): Likewise.
673 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
674 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
676 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
677 Alan Hayward <alan.hayward@arm.com>
678 David Sherwood <david.sherwood@arm.com>
680 * fold-const.c (pointer_may_wrap_p): Take the offset as a
681 HOST_WIDE_INT rather than a poly_int64.
683 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
684 Alan Hayward <alan.hayward@arm.com>
685 David Sherwood <david.sherwood@arm.com>
687 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
688 HOST_WIDE_INT to poly_int64_pod.
689 (perform_symbolic_merge): Update accordingly.
690 (bswap_replace): Likewise.
692 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
693 Alan Hayward <alan.hayward@arm.com>
694 David Sherwood <david.sherwood@arm.com>
696 * tree-affine.h (aff_tree::offset): Change from widest_int
698 (wide_int_ext_for_comb): Delete.
699 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
700 constants as poly_widest_int rather than widest_int.
701 (aff_combination_constant_multiple_p): Return the multiplier
702 as a poly_widest_int.
703 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
705 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
706 version static and add an overload for poly_widest_int.
707 (aff_combination_const, aff_combination_add_cst)
708 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
709 the constants as poly_widest_int rather than widest_int.
710 (tree_to_aff_combination): Generalize INTEGER_CST case to
712 (aff_combination_to_tree): Track offsets as poly_widest_ints.
713 (aff_combination_add_product, aff_combination_mult): Handle
715 (aff_combination_constant_multiple_p): Return the multiplier
716 as a poly_widest_int.
717 * tree-predcom.c (determine_offset): Return the offset as a
719 (split_data_refs_to_components, suitable_component_p): Update
721 (valid_initializer_p): Update call to
722 aff_combination_constant_multiple_p.
723 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
724 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
725 as a poly_int64 rather than a HOST_WIDE_INT.
726 (get_address_cost): Handle polynomial offsets.
727 (iv_elimination_compare_lt): Likewise.
728 (rewrite_use_nonlinear_expr): Likewise.
730 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
731 Alan Hayward <alan.hayward@arm.com>
732 David Sherwood <david.sherwood@arm.com>
734 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
735 as a poly_int64_pod rather than a HOST_WIDE_INT.
736 (get_addr_base_and_unit_offset): Likewise.
737 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
738 (get_addr_base_and_unit_offset): Likewise.
739 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
740 to poly_int64 in example.
741 * fold-const.c (fold_binary_loc): Update call to
742 get_addr_base_and_unit_offset.
743 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
744 (maybe_canonicalize_mem_ref_addr): Likewise.
745 (gimple_fold_stmt_to_constant_1): Likewise.
746 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
748 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
749 * match.pd: Likewise.
750 * omp-low.c (lower_omp_target): Likewise.
751 * tree-sra.c (build_ref_for_offset): Likewise.
752 (build_debug_ref_for_model): Likewise.
753 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
754 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
755 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
756 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
757 (constant_pointer_difference): Likewise.
758 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
759 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
760 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
761 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
762 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
763 (set_ssa_val_to): Likewise.
764 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
765 (maybe_diag_stxncpy_trunc): Likewise.
766 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
767 * tree.c (build_simple_mem_ref_loc): Likewise.
768 (array_at_struct_end_p): Likewise.
770 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
771 Alan Hayward <alan.hayward@arm.com>
772 David Sherwood <david.sherwood@arm.com>
774 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
775 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
776 (get_ref_base_and_extent_hwi): Declare.
777 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
778 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
779 (get_ref_base_and_extent_hwi): New function.
780 * cfgexpand.c (expand_debug_expr): Update call to
781 get_ref_base_and_extent.
782 * dwarf2out.c (add_var_loc_to_decl): Likewise.
783 * gimple-fold.c (get_base_constructor): Return the offset as a
784 poly_int64_pod rather than a HOST_WIDE_INT.
785 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
786 * ipa-polymorphic-call.c
787 (ipa_polymorphic_call_context::set_by_invariant)
788 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
789 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
790 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
791 rather than get_ref_base_and_extent.
792 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
793 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
794 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
796 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
797 call to get_ref_base_and_extent.
798 * tree-sra.c (create_access, get_access_for_expr): Likewise.
799 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
800 (stmt_kills_ref_p): Likewise.
801 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
802 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
804 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
805 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
806 when calling native_encode_expr.
807 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
808 call to get_ref_base_and_extent.
809 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
810 get_ref_base_and_extent.
811 * var-tracking.c (track_expr_p): Likewise.
813 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
814 Alan Hayward <alan.hayward@arm.com>
815 David Sherwood <david.sherwood@arm.com>
817 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
818 HOST_WIDE_INT to poly_int64_pod.
819 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
820 polynomail parameter offsets.
822 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
823 Alan Hayward <alan.hayward@arm.com>
824 David Sherwood <david.sherwood@arm.com>
826 * gengtype.c (main): Handle poly_int64_pod.
827 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
828 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
829 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
830 from HOST_WIDE_INT to poly_int64_pod.
831 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
832 (copy_cfa): New function.
833 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
834 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
835 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
836 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
837 (initial_return_save): Treat offsets as poly_ints.
838 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
839 offset is nonconstant.
840 (reg_save): Take the offset as a poly_int64. Fall back to
841 DW_CFA_expression for nonconstant offsets.
842 (queue_reg_save): Take the offset as a poly_int64.
843 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
845 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
846 Alan Hayward <alan.hayward@arm.com>
847 David Sherwood <david.sherwood@arm.com>
849 * rtl.h (operand_subword, operand_subword_force): Take the offset
850 as a poly_uint64 an unsigned int.
851 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
853 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
854 Alan Hayward <alan.hayward@arm.com>
855 David Sherwood <david.sherwood@arm.com>
857 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
858 'p' format code. Use INT_LIST rather than SUBREG as the example of
859 a code with an XINT and an XEXP. Remove the implication that
860 accessing an rtx field using XINT is expected to work.
861 * rtl.def (SUBREG): Change format from "ei" to "ep".
862 * rtl.h (rtunion::rt_subreg): New field.
863 (XCSUBREG): New macro.
864 (SUBREG_BYTE): Use it.
865 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
866 Update constructor accordingly.
867 (subreg_shape::operator ==): Update accordingly.
868 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
869 than an unsigned int.
870 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
871 a poly_uint64 rather than an unsigned int.
872 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
873 than an unsigned int.
874 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
875 (subreg_size_highpart_offset): Return a poly_uint64 rather than
876 an unsigned int. Take the sizes as poly_uint64s.
877 (subreg_offset_from_lsb): Return a poly_uint64 rather than
878 an unsigned int. Take the shift as a poly_uint64 rather than
880 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
881 as a poly_uint64 rather than an unsigned int.
882 (simplify_subreg_regno): Likewise.
883 (byte_lowpart_offset): Return the memory offset as a poly_int64
885 (subreg_memory_offset): Likewise. Take the subreg offset as a
886 poly_uint64 rather than an unsigned int.
887 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
888 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
889 poly_uint64 rather than an unsigned int.
890 * rtl.c (rtx_format): Describe 'p' in comment.
891 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
892 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
893 offset as a poly_uint64 rather than an unsigned int.
894 (byte_lowpart_offset): Return the memory offset as a poly_int64
896 (subreg_memory_offset): Likewise. Take the subreg offset as a
897 poly_uint64 rather than an unsigned int.
898 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
899 mode sizes as poly_uint64s rather than unsigned ints. Return a
900 poly_uint64 rather than an unsigned int.
901 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
902 (copy_insn_1): Handle 'p'.
903 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
904 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
905 an unsigned int. Return the shift in the same way.
906 (subreg_lsb): Return the shift as a poly_uint64 rather than an
908 (subreg_size_offset_from_lsb): Take the sizes and shift as
909 poly_uint64s rather than unsigned ints. Return the offset as
911 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
912 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
914 * rtlhash.c (add_rtx): Handle 'p'.
915 * genemit.c (gen_exp): Likewise.
916 * gengenrtl.c (type_from_format, gendef): Likewise.
917 * gensupport.c (subst_pattern_match, get_alternatives_number)
918 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
919 (subst_dup): Likewise.
920 * gengtype.c (adjust_field_rtx_def): Likewise.
921 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
922 (match_pattern_2): Likewise.
923 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
924 (rtx_test::subreg_field): New function.
925 (operator ==, safe_to_hoist_p, transition_parameter_type)
926 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
927 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
928 * genpeep.c (match_rtx): Likewise.
929 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
930 (rtx_writer::print_rtx_operand): Handle 'p'.
931 (print_value): Handle SUBREG.
932 * read-rtl.c (apply_int_iterator): Likewise.
933 (rtx_reader::read_rtx_operand): Handle 'p'.
934 * alias.c (rtx_equal_for_memref_p): Likewise.
935 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
936 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
938 * calls.c (expand_call): Likewise.
939 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
940 (make_extraction, gen_lowpart_for_combine): Likewise.
941 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
943 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
944 rather than an unsigned int. Treat subreg offsets as poly_ints.
945 (exp_equiv_p): Handle 'p'.
946 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
947 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
948 * dse.c (find_shift_sequence): Likewise.
949 * dwarf2out.c (rtl_for_decl_location): Likewise.
950 * expmed.c (extract_low_bits): Likewise.
951 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
952 (expand_expr_real_2): Likewise.
953 * final.c (alter_subreg): Likewise.
954 (leaf_renumber_regs_insn): Handle 'p'.
955 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
956 Treat subreg offsets as poly_ints.
957 * fwprop.c (forward_propagate_and_simplify): Likewise.
958 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
959 * ira.c (get_subreg_tracking_sizes): Likewise.
960 * ira-conflicts.c (go_through_subreg): Likewise.
961 * ira-lives.c (process_single_reg_class_operands): Likewise.
962 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
963 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
964 as a poly_uint64 rather than an unsigned int.
965 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
966 subreg offsets as poly_ints.
967 * lra-constraints.c (operands_match_p): Handle 'p'.
968 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
969 * lra-spills.c (assign_mem_slot): Likewise.
970 * postreload.c (move2add_valid_value_p): Likewise.
971 * recog.c (general_operand, indirect_operand): Likewise.
972 * regcprop.c (copy_value, maybe_mode_change): Likewise.
973 (copyprop_hardreg_forward_1): Likewise.
974 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
975 (record_subregs_of_mode): Likewise.
976 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
977 * reload.c (operands_match_p): Handle 'p'.
978 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
979 * reload1.c (alter_reg, choose_reload_regs): Likewise.
980 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
981 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
982 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
983 (simplify_const_poly_int_tests<N>::run): Likewise.
984 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
985 a poly_uint64 rather than an unsigned int.
986 * valtrack.c (debug_lowpart_subreg): Likewise.
987 * var-tracking.c (var_lowpart): Likewise.
988 (loc_cmp): Handle 'p'.
990 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
991 Alan Hayward <alan.hayward@arm.com>
992 David Sherwood <david.sherwood@arm.com>
994 * ira.c (get_subreg_tracking_sizes): New function.
995 (init_live_subregs): Take an integer size rather than a register.
996 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
997 to init_live_subregs.
999 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1000 Alan Hayward <alan.hayward@arm.com>
1001 David Sherwood <david.sherwood@arm.com>
1003 * expr.c (store_constructor_field): Change bitsize from a
1004 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1005 HOST_WIDE_INT to a poly_int64.
1006 (store_constructor): Change size from a HOST_WIDE_INT to
1008 (store_field): Likewise bitsize and bitpos.
1010 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1011 Alan Hayward <alan.hayward@arm.com>
1012 David Sherwood <david.sherwood@arm.com>
1014 * expmed.h (store_bit_field): Change bitregion_start and
1015 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1016 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1017 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1018 (store_fixed_bit_field, store_split_bit_field): Likewise.
1019 * expr.c (store_constructor_field, store_field): Likewise.
1020 (optimize_bitfield_assignment_op): Likewise. Make the same change
1021 to bitsize and bitpos.
1022 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1023 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1024 same change in the constructor arguments.
1025 (get_best_mode): Change bitregion_start and bitregion_end from
1026 unsigned HOST_WIDE_INT to poly_uint64.
1027 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1028 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1030 (bit_field_mode_iterator::next_mode): Update for new types
1031 of m_bitregion_start and m_bitregion_end.
1032 (get_best_mode): Change bitregion_start and bitregion_end from
1033 unsigned HOST_WIDE_INT to poly_uint64.
1035 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1036 Alan Hayward <alan.hayward@arm.com>
1037 David Sherwood <david.sherwood@arm.com>
1039 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1040 accepts poly_uint64 offsets.
1041 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1042 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1043 * expmed.c (lowpart_bit_field_p): Likewise.
1044 (extract_bit_field_as_subreg): New function, split out from...
1045 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1046 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1047 extractions, check that BITSIZE matches the size of the extracted
1048 value and that BITNUM is an exact multiple of that size.
1049 If all else fails, try forcing the value into memory if
1050 BITNUM is variable, and adjusting the address so that the
1051 offset is constant. Split the part that can only handle constant
1052 bitsize and bitnum out into...
1053 (extract_integral_bit_field): ...this new function.
1054 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1055 rather than unsigned HOST_WIDE_INTs.
1057 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1058 Alan Hayward <alan.hayward@arm.com>
1059 David Sherwood <david.sherwood@arm.com>
1061 * expmed.h (store_bit_field): Take bitsize and bitnum as
1062 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1063 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1064 that returns the byte size.
1065 (store_bit_field_1): Take bitsize and bitnum as
1066 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1067 to simple_mem_bitfield_p. Split the part that can only handle
1068 constant bitsize and bitnum out into...
1069 (store_integral_bit_field): ...this new function.
1070 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1071 than unsigned HOST_WIDE_INTs.
1072 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1074 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1075 Alan Hayward <alan.hayward@arm.com>
1076 David Sherwood <david.sherwood@arm.com>
1078 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1079 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1081 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1082 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1083 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1084 offset as a poly_int64 rather than an int.
1085 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1086 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1087 * lra-constraints.c (equiv_address_substitution): Track offsets
1089 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1090 (curr_insn_transform): Handle the new form of sp_offset.
1091 * lra-eliminations.c (lra_elim_table): Change previous_offset
1092 and offset from HOST_WIDE_INT to poly_int64.
1093 (print_elim_table, update_reg_eliminate): Update accordingly.
1094 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1095 (get_elimination): Update accordingly.
1096 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1097 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1098 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1099 poly_int64 offsets generally.
1100 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1101 (mark_not_eliminable, init_elimination): Update accordingly.
1102 (remove_reg_equal_offset_note): Return a bool and pass the new
1103 offset back by pointer as a poly_int64.
1104 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1105 rather than a HOST_WIDE_INT.
1106 (do_remat): Track offsets poly_int64s.
1107 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1109 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1110 Alan Hayward <alan.hayward@arm.com>
1111 David Sherwood <david.sherwood@arm.com>
1113 * rtl.h (mem_attrs): Add a default constructor. Change size and
1114 offset from HOST_WIDE_INT to poly_int64.
1115 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1116 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1117 (widen_memory_access): Take the sizes and offsets as poly_int64s
1118 rather than HOST_WIDE_INTs.
1119 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1120 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1122 (adjust_offset_for_component_ref): Change the offset from a
1123 HOST_WIDE_INT to a poly_int64.
1124 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1125 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1126 * dce.c (find_call_stack_args): Likewise.
1127 * dse.c (record_store): Likewise.
1128 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1129 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1130 * read-rtl-function.c (test_loading_mem): Likewise.
1131 * rtlanal.c (may_trap_p_1): Likewise.
1132 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1133 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1134 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1135 (mem_attrs::mem_attrs): New function.
1136 (set_mem_attributes_minus_bitpos): Change bitpos from a
1137 HOST_WIDE_INT to poly_int64.
1138 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1139 (clear_mem_offset, clear_mem_size, change_address)
1140 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1141 initialize mem_attrs.
1142 (set_mem_offset, set_mem_size, adjust_address_1)
1143 (adjust_automodify_address_1, offset_address, widen_memory_access):
1144 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
1146 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1147 Alan Hayward <alan.hayward@arm.com>
1148 David Sherwood <david.sherwood@arm.com>
1150 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1151 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1152 rather than 0 to represent an unknown size. Assert that the size
1153 is known when the mode isn't BLKmode.
1154 (may_trap_p_1): Use -1 for unknown sizes.
1155 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1157 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1158 Alan Hayward <alan.hayward@arm.com>
1159 David Sherwood <david.sherwood@arm.com>
1161 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1162 to poly_int64. Update commentary for positions_needed.large.
1163 (read_info_type): Change offset and width from HOST_WIDE_INT
1165 (set_usage_bits): Likewise.
1166 (canon_address): Return the offset as a poly_int64 rather than
1167 a HOST_WIDE_INT. Use strip_offset_and_add.
1168 (set_all_positions_unneeded, any_positions_needed_p): Use
1169 positions_needed.large to track stores with non-constant widths.
1170 (all_positions_needed_p): Likewise. Take the offset and width
1171 as poly_int64s rather than ints. Assert that rhs is nonnull.
1172 (record_store): Cope with non-constant offsets and widths.
1173 Nullify the rhs of an earlier store if we can't tell which bytes
1175 (find_shift_sequence): Take the access_size and shift as poly_int64s
1177 (get_stored_val): Take the read_offset and read_width as poly_int64s
1178 rather than HOST_WIDE_INTs.
1179 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1180 non-constant offsets and widths.
1182 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1183 Alan Hayward <alan.hayward@arm.com>
1184 David Sherwood <david.sherwood@arm.com>
1186 * inchash.h (inchash::hash::add_poly_int): New function.
1187 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1188 Use poly_int64 rather than HOST_WIDE_INT.
1189 (ao_ref::max_size_known_p): New function.
1190 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1191 rather than HOST_WIDE_INT.
1192 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1193 to temporaries until its interface is adjusted to match.
1194 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1195 (aliasing_component_refs_p, decl_refs_may_alias_p)
1196 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1197 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1198 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1200 * alias.c (ao_ref_from_mem): Likewise.
1201 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1202 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1203 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1204 (maybe_trim_complex_store, maybe_trim_constructor_store)
1205 (live_bytes_read, dse_classify_store): Likewise.
1206 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1207 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1208 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1209 (vn_reference_lookup_3): Likewise.
1210 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1212 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1213 Alan Hayward <alan.hayward@arm.com>
1214 David Sherwood <david.sherwood@arm.com>
1216 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1217 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1218 instead of ranges_overlap_p.
1220 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1221 Alan Hayward <alan.hayward@arm.com>
1222 David Sherwood <david.sherwood@arm.com>
1224 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1225 sizes as poly_int64s rather than HOST_WIDE_INTs.
1227 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1228 Alan Hayward <alan.hayward@arm.com>
1229 David Sherwood <david.sherwood@arm.com>
1231 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1232 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1233 * gimple-fold.c (fold_ctor_reference): Likewise.
1235 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1236 Alan Hayward <alan.hayward@arm.com>
1237 David Sherwood <david.sherwood@arm.com>
1239 * target.def (dwarf_poly_indeterminate_value): New hook.
1240 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1241 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1242 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1243 * doc/tm.texi: Regenerate.
1244 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1245 offset as a poly_int64.
1246 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1247 offset as a poly_int64.
1248 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1249 Take the offset as a poly_int64.
1250 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1251 (frame_pointer_fb_offset): Change to a poly_int64.
1252 (int_loc_descriptor): Take the offset as a poly_int64. Use
1253 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1254 (based_loc_descr): Take the offset as a poly_int64.
1255 Use strip_offset_and_add to handle (plus X (const)).
1256 Use new_reg_loc_descr instead of an open-coded version of the
1257 previous implementation.
1258 (mem_loc_descriptor): Handle CONST_POLY_INT.
1259 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1260 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1262 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1263 Alan Hayward <alan.hayward@arm.com>
1264 David Sherwood <david.sherwood@arm.com>
1266 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1267 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1268 * inchash.h (inchash::hash::add_poly_hwi): New function.
1269 * gengtype.c (main): Register poly_int64.
1270 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1271 offset as a poly_int.
1272 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1273 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1274 offset as a poly_int64.
1275 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1276 * print-rtl.c (print_poly_int): New function.
1277 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1279 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1281 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1282 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1283 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1284 Enforce track_offset_p for parts of a PARALLEL.
1285 (vt_add_function_parameter): Use const_offset for the final
1286 offset to track. Use get_tracked_reg_offset for the parts
1289 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1290 Alan Hayward <alan.hayward@arm.com>
1291 David Sherwood <david.sherwood@arm.com>
1293 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1294 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1295 * doc/tm.texi: Regenerate.
1296 * hooks.h (hook_bool_uint_uint_true): Delete.
1297 (hook_bool_puint64_puint64_true): Declare.
1298 * hooks.c (hook_bool_uint_uint_true): Delete.
1299 (hook_bool_puint64_puint64_true): New function.
1300 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1301 instead of unsigned ints.
1302 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1303 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1305 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1306 Alan Hayward <alan.hayward@arm.com>
1307 David Sherwood <david.sherwood@arm.com>
1309 * optabs.h (expand_operand): Add an int_value field.
1310 (create_expand_operand): Add an int_value parameter and use it
1311 to initialize the new expand_operand field.
1312 (create_integer_operand): Replace with a declaration of a function
1313 that accepts poly_int64s. Move the implementation to...
1314 * optabs.c (create_integer_operand): ...here.
1315 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1316 the mode preserves the value of int_value, instead of calling
1317 const_int_operand on the rtx. Use gen_int_mode to generate
1320 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1321 Alan Hayward <alan.hayward@arm.com>
1322 David Sherwood <david.sherwood@arm.com>
1324 * dumpfile.h (dump_dec): Declare.
1325 * dumpfile.c (dump_dec): New function.
1326 * pretty-print.h (pp_wide_integer): Turn into a function and
1327 declare a poly_int version.
1328 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1330 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1331 Alan Hayward <alan.hayward@arm.com>
1332 David Sherwood <david.sherwood@arm.com>
1334 * doc/generic.texi (POLY_INT_CST): Document.
1335 * tree.def (POLY_INT_CST): New tree code.
1336 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1337 * tree-core.h (tree_poly_int_cst): New struct.
1338 (tree_node): Add a poly_int_cst field.
1339 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1340 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1341 instead of a wide_int_ref.
1342 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1344 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1345 instead of an unsigned HOST_WIDE_INT.
1346 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1347 (ptrdiff_tree_p): Declare.
1348 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1349 extern inline implementations if the target doesn't use POLY_INT_CST.
1350 (poly_int_tree_p): New function.
1351 (wi::unextended_tree): New class.
1352 (wi::int_traits <unextended_tree>): New override.
1353 (wi::extended_tree): Add a default constructor.
1354 (wi::extended_tree::get_tree): New function.
1355 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1356 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1357 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1358 (wi::tree_to_poly_wide_ref): New typedefs.
1359 (wi::ints_for): Provide overloads for extended_tree and
1361 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1362 (wi::to_wide): New functions.
1363 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1364 * tree.c (poly_int_cst_hasher): New struct.
1365 (poly_int_cst_hash_table): New variable.
1366 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1367 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1369 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1370 (init_ttree): Initialize poly_int_cst_hash_table.
1371 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1372 a poly_int64 instead of a HOST_WIDE_INT.
1373 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1374 instead of an unsigned HOST_WIDE_INT.
1375 (wide_int_to_tree): Rename to...
1376 (wide_int_to_tree_1): ...this.
1377 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1378 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1379 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1380 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1381 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1382 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1384 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1385 (streamer_read_tree_body): Likewise.
1386 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1387 (streamer_write_tree_body): Likewise.
1388 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1389 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1390 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1391 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1392 * gimple-expr.h (is_gimple_constant): Likewise.
1393 * gimplify.c (maybe_with_size_expr): Likewise.
1394 * print-tree.c (print_node): Likewise.
1395 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1396 * tree-pretty-print.c (dump_generic_node): Likewise.
1397 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1398 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1399 * tree-vrp.c (compare_values_warnv): Likewise.
1400 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1401 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1402 (force_expr_to_var_cost): Likewise.
1403 * tree-ssa-loop.c (for_each_index): Likewise.
1404 * fold-const.h (build_invariant_address, size_int_kind): Take a
1405 poly_int64 instead of a HOST_WIDE_INT.
1406 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1407 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1409 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1410 (int_const_binop_2): New function, split out from...
1411 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1412 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1413 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1414 * gimple-ssa-strength-reduction.c (slsr_process_add)
1415 (slsr_process_mul): Check for INTEGER_CSTs before using them
1417 * stor-layout.c (bits_from_bytes): New function.
1418 (bit_from_pos): Use it.
1419 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1420 by BITS_PER_UNIT to get the TYPE_SIZE.
1421 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1422 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1424 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1425 Alan Hayward <alan.hayward@arm.com>
1426 David Sherwood <david.sherwood@arm.com>
1428 * doc/rtl.texi (const_poly_int): Document. Also document the
1429 rtl sharing behavior.
1430 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1431 * rtl.h (const_poly_int_def): New struct.
1432 (rtx_def::u): Add a cpi field.
1433 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1434 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1435 (wi::rtx_to_poly_wide_ref): New typedef
1436 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1437 (poly_int_rtx_p): New functions.
1438 (trunc_int_for_mode): Declare a poly_int64 version.
1439 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1440 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1442 (strip_offset): Declare.
1443 (strip_offset_and_add): New function.
1444 * rtl.def (CONST_POLY_INT): New rtx code.
1445 * rtl.c (rtx_size): Handle CONST_POLY_INT.
1446 (shared_const_p): Use poly_int_rtx_p.
1447 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
1449 (gen_int_shift_amount): Likewise.
1450 * emit-rtl.c (const_poly_int_hasher): New class.
1451 (const_poly_int_htab): New variable.
1452 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
1453 (const_poly_int_hasher::hash): New function.
1454 (const_poly_int_hasher::equal): Likewise.
1455 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
1456 (immed_wide_int_const): Rename to...
1457 (immed_wide_int_const_1): ...this and make static.
1458 (immed_wide_int_const): New function, taking a poly_wide_int_ref
1459 instead of a wide_int_ref.
1460 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
1461 (gen_lowpart_common): Handle CONST_POLY_INT.
1462 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
1463 * cselib.c (cselib_hash_rtx): Likewise.
1464 * dwarf2out.c (const_ok_for_output_1): Likewise.
1465 * expr.c (convert_modes): Likewise.
1466 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
1467 * rtlhash.c (add_rtx): Likewise.
1468 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
1469 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1470 Handle existing CONST_POLY_INT rtxes.
1471 * expmed.h (expand_shift): Take a poly_int64 instead of a
1473 * expmed.c (expand_shift): Likewise.
1474 * rtlanal.c (strip_offset): New function.
1475 (commutative_operand_precedence): Give CONST_POLY_INT the same
1476 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
1478 * rtl-tests.c (const_poly_int_tests): New struct.
1479 (rtl_tests_c_tests): Use it.
1480 * simplify-rtx.c (simplify_const_unary_operation): Handle
1482 (simplify_const_binary_operation): Likewise.
1483 (simplify_binary_operation_1): Fold additions of symbolic constants
1484 and CONST_POLY_INTs.
1485 (simplify_subreg): Handle extensions and truncations of
1487 (simplify_const_poly_int_tests): New struct.
1488 (simplify_rtx_c_tests): Use it.
1489 * wide-int.h (storage_ref): Add default constructor.
1490 (wide_int_ref_storage): Likewise.
1491 (trailing_wide_ints): Use GTY((user)).
1492 (trailing_wide_ints::operator[]): Add a const version.
1493 (trailing_wide_ints::get_precision): New function.
1494 (trailing_wide_ints::extra_size): Likewise.
1496 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1497 Alan Hayward <alan.hayward@arm.com>
1498 David Sherwood <david.sherwood@arm.com>
1500 * emit-rtl.h (gen_int_shift_amount): Declare.
1501 * emit-rtl.c (gen_int_shift_amount): New function.
1502 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
1504 * calls.c (shift_return_value): Likewise.
1505 * cse.c (fold_rtx): Likewise.
1506 * dse.c (find_shift_sequence): Likewise.
1507 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
1508 (expand_shift, expand_smod_pow2): Likewise.
1509 * lower-subreg.c (shift_cost): Likewise.
1510 * optabs.c (expand_superword_shift, expand_doubleword_mult)
1511 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
1512 (expand_vec_perm_var): Likewise.
1513 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1514 (simplify_binary_operation_1): Likewise.
1515 * combine.c (try_combine, find_split_point, force_int_to_mode)
1516 (simplify_shift_const_1, simplify_shift_const): Likewise.
1517 (change_zero_ext): Likewise. Use simplify_gen_binary.
1519 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1521 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
1523 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1525 * doc/invoke.texi (ARM Options): Document accepted extension options
1526 for -march=armv8.3-a.
1528 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
1531 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
1532 or --with-float=softfp, set the default CPU to arm10e.
1534 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
1536 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
1537 * config/visium/predicates.md (const_shift_operand): Likewise.
1538 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
1539 (visium_legitimize_reload_address): Likewise.
1541 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1544 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
1546 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1549 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
1551 2017-12-20 Julia Koval <julia.koval@intel.com>
1553 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
1554 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
1555 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
1556 * config.gcc: Include vpclmulqdqintrin.h.
1557 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
1558 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
1560 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
1561 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
1562 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
1563 * config/i386/i386.c (isa2_opts): Add -mcx16.
1564 (isa_opts): Add -mpclmulqdq, remove -mcx16.
1565 (ix86_option_override_internal): Move mcx16 to flags2.
1566 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
1567 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
1568 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
1569 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
1570 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
1571 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
1572 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
1573 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
1574 * doc/invoke.texi: Add -mvpclmulqdq.
1576 2017-12-20 Tom de Vries <tom@codesourcery.com>
1579 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
1580 * calls.c (rtx_for_static_chain): ... here. New function.
1581 * calls.h (rtx_for_static_chain): Declare.
1582 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
1583 instead of targetm.calls.static_chain.
1584 * df-scan.c (df_get_entry_block_def_set): Same.
1586 2017-12-19 Tom de Vries <tom@codesourcery.com>
1588 PR tree-optimization/83493
1589 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
1592 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1594 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
1595 inform with hardcoded english plural handling.
1597 2017-12-18 Jeff Law <law@redhat.com>
1599 PR tree-optimization/83477
1600 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
1601 a non-virtual PHI, always push a new range.
1603 2017-12-19 Martin Sebor <msebor@redhat.com>
1606 * builtins.c (compute_objsize): Handle non-constant offsets.
1608 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1610 PR tree-optimization/83444
1611 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
1612 character load case, if get_stridx on MEM_REF's operand doesn't
1613 look usable, retry with get_addr_stridx.
1615 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
1618 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
1619 (variable_tracking_main_1): Keep markers even when VTA fails.
1622 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
1623 even if there are markers before it.
1624 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
1626 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1628 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
1629 typical order conditions.
1630 * sel-sched.c (extract_new_fences_from): Likewise.
1631 * config/visium/constraints.md (J, K, L): Likewise.
1632 * config/visium/predicates.md (const_shift_operand): Likewise.
1633 * config/visium/visium.c (visium_legitimize_address,
1634 visium_legitimize_reload_address): Likewise.
1635 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
1636 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
1637 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
1638 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
1639 SET_DUMP_DETAIL): Likewise.
1640 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
1641 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
1642 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
1643 avr_set_core_architecture, avr_set_current_function,
1644 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
1645 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
1646 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
1647 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
1648 avr_map_decompose, avr_fold_builtin): Likewise.
1649 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
1650 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
1651 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
1652 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
1653 * config/m32c/m32c.c (m32c_conditional_register_usage,
1654 m32c_address_cost): Likewise.
1655 * config/m32c/predicates.md (shiftcount_operand,
1656 longshiftcount_operand): Likewise.
1657 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
1658 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
1659 can_use_cdx_ldstw): Likewise.
1660 * config/nios2/nios2.h (CDX_REG_P): Likewise.
1661 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
1663 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
1664 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
1665 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
1666 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
1667 * config/vax/vax.c (adjacent_operands_p): Likewise.
1668 * config/ft32/constraints.md (L, b, KA): Likewise.
1669 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
1671 * cfgexpand.c (expand_stack_alignment): Likewise.
1672 * gcse.c (insert_expr_in_table): Likewise.
1673 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
1674 * cgraphunit.c (cgraph_node::expand): Likewise.
1675 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
1676 * emit-rtl.c (add_insn): Likewise.
1677 * input.c (dump_location_info): Likewise.
1678 * passes.c (NEXT_PASS): Likewise.
1679 * read-rtl-function.c (parse_note_insn_name,
1680 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
1682 * sched-rgn.c (sched_rgn_init): Likewise.
1683 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
1684 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
1685 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
1686 * explow.c (eliminate_constant_term): Likewise.
1687 * final.c (leaf_renumber_regs_insn): Likewise.
1688 * cfgrtl.c (print_rtl_with_bb): Likewise.
1689 * genhooks.c (emit_init_macros): Likewise.
1690 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
1691 * tree-data-ref.c (conflict_fn): Likewise.
1692 * selftest.c (assert_streq): Likewise.
1693 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
1694 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
1695 fold_binary_loc, multiple_of_p): Likewise.
1696 * reload.c (push_reload, find_equiv_reg): Likewise.
1697 * et-forest.c (et_nca, et_below): Likewise.
1698 * dbxout.c (dbxout_symbol_location): Likewise.
1699 * reorg.c (relax_delay_slots): Likewise.
1700 * dojump.c (do_compare_rtx_and_jump): Likewise.
1701 * gengtype-parse.c (type): Likewise.
1702 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
1703 simplify_const_relational_operation): Likewise.
1704 * reload1.c (do_output_reload): Likewise.
1705 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
1706 * gengtype.c (type_for_name): Likewise.
1707 * gimple-ssa-sprintf.c (format_directive): Likewise.
1709 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1712 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
1713 accessing it. Adjust comment.
1715 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1718 * predict.c (zero_one_minusone): New function.
1719 (apply_return_prediction): Avoid return prediction for functions
1720 returning only -1, 0 and 1 values, unless they only return -1 and 0
1723 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1725 * config/arc/arc.c (legitimate_scaled_address_p): Clean
1726 fall-through warning.
1727 (arc_compute_frame_size): Remove unused variables.
1728 (arc_print_operand): Fix fprintif format.
1729 (arc_can_follow_jump): Clean fall-through warning.
1731 2017-12-19 Marek Polacek <polacek@redhat.com>
1734 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
1737 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
1739 * config/arc/arc.c (overriderregs): New variable.
1740 (arc_override_options): Track fixed/call saved/call options.
1741 (arc_conditional_register_usage): Check against overrideregs
1742 variable whenever we change register properties.
1744 2017-12-19 Nathan Sidwell <nathan@acm.org>
1746 * opts.c (finish_options): Don't prefix dump_base_name if it
1747 already contains directories.
1749 2017-12-19 Martin Liska <mliska@suse.cz>
1751 PR rtl-optimization/82675
1752 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
1753 more element in sbitmap.
1755 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
1757 * gimplify.c (gimplify_expr): Use error_operand_p.
1759 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
1762 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
1765 2017-12-19 Daniel Cederman <cederman@gaisler.com>
1767 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
1768 the jump is to a label.
1770 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1772 PR tree-optimization/83444
1773 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1778 * ipa-inline.c (flatten_remove_node_hook): New function.
1779 (ipa_inline): Keep only nodes with flatten attribute at the end of
1780 the array in the order from ipa_reverse_postorder, only walk that
1781 portion of array for flattening, if there is more than one such
1782 node, temporarily register a removal hook and ignore removed nodes.
1784 PR tree-optimization/80631
1785 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1786 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1787 IFN_REDUC_MAX or IFN_REDUC_MIN.
1789 2017-12-18 Jeff Law <law@redhat.com>
1791 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1792 record symbolic equivalences from backedges in the CFG.
1795 2017-11-19 Jeff Law <law@redhat.com>
1797 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1798 of degenerates resulting from ignoring an edge.
1800 2017-12-18 Martin Sebor <msebor@redhat.com>
1803 PR tree-optimization/78450
1804 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1805 (handle_builtin_strlen): Call it.
1807 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1809 PR rtl-optimization/83424
1810 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1812 2017-12-18 Marek Polacek <polacek@redhat.com>
1815 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1816 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1819 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1821 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1822 (umaddsidi4, umaddsidi_split): Likewise.
1824 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1826 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1829 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1832 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1834 2017-12-18 Richard Biener <rguenther@suse.de>
1836 PR tree-optimization/81877
1837 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1838 (outermost_indep_loop): Adjust.
1839 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1840 (can_sm_ref_p): Adjust.
1842 2017-12-18 Richard Biener <rguenther@suse.de>
1845 * tree.c (array_at_struct_end_p): Return true if the underlying
1846 object has space for at least one element in excess of what
1847 the array domain specifies.
1849 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1851 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1852 example to avoid overfull hbox.
1853 * doc/invoke.texi (Option Summary): Add missing @gol.
1854 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1857 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1858 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1860 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1861 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1862 (AArch64 Options): Document -mverbose-cost-dump.
1863 (ARM Options): Likewise, plus -mflip-thumb.
1865 2017-12-17 Martin Sebor <msebor@redhat.com>
1868 * gimple-ssa-warn-restrict.c
1869 (builtin_memref::offset_out_of_bounds): Correct the handling of
1872 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1874 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1876 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1878 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1880 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1882 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1885 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1887 * doc/invoke.texi: Fix some typos.
1889 2017-12-16 Martin Sebor <msebor@redhat.com>
1891 PR tree-optimization/78918
1892 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1893 * builtins.c (check_sizes): Rename...
1894 (check_access): ...to this. Rename function arguments for clarity.
1895 (check_memop_sizes): Adjust names.
1896 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1897 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1898 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1899 (check_strncat_sizes, expand_builtin_strncat): Same.
1900 (expand_builtin_strncpy, expand_builtin_memset): Same.
1901 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1902 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1903 (maybe_emit_sprintf_chk_warning): Same.
1904 (expand_builtin_strcpy): Adjust.
1905 (expand_builtin_stpcpy): Same.
1906 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1907 in pointer-checking forms of memcpy, memmove, and mempcpy.
1908 (gcall_to_tree_minimal, max_object_size): Define new functions.
1909 * builtins.h (max_object_size): Declare.
1910 * calls.c (alloc_max_size): Call max_object_size instead of
1911 hardcoding ssizetype limit.
1912 (get_size_range): Handle new argument.
1913 * calls.h (get_size_range): Add a new argument.
1914 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1915 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1916 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1918 (gimple_fold_builtin_memory_chk): Same.
1919 (gimple_fold_builtin_stxcpy_chk): New function.
1920 * gimple-ssa-warn-restrict.c: New source.
1921 * gimple-ssa-warn-restrict.h: New header.
1922 * gimple.c (gimple_build_call_from_tree): Propagate location.
1923 * passes.def (pass_warn_restrict): Add new pass.
1924 * tree-pass.h (make_pass_warn_restrict): Declare.
1925 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1927 (handle_builtin_strcat): Same.
1928 (strlen_optimize_stmt): Rename...
1929 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1930 stpncpy, strncpy, and their checking forms.
1932 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1934 PR rtl-optimization/82849
1935 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1936 and get_max_loop_iterations_int.
1938 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1939 Alan Hayward <alan.hayward@arm.com>
1940 David Sherwood <david.sherwood@arm.com>
1942 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1943 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1944 size as a poly_uint64.
1945 (mode_for_vector, mode_for_int_vector): Take the number of vector
1946 elements as a poly_uint64.
1947 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1948 size as a poly_uint64.
1949 (mode_for_vector, mode_for_int_vector): Take the number of vector
1950 elements as a poly_uint64.
1952 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1953 Alan Hayward <alan.hayward@arm.com>
1954 David Sherwood <david.sherwood@arm.com>
1956 * machmode.h (MACRO_MODE): New macro.
1957 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1958 * targhooks.c (default_libcall_value, default_secondary_reload)
1959 (default_memory_move_cost, default_register_move_cost)
1960 (default_class_max_nregs): Likewise.
1962 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1963 Alan Hayward <alan.hayward@arm.com>
1964 David Sherwood <david.sherwood@arm.com>
1966 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1967 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1969 * genautomata.c (main): Likewise.
1970 * genconditions.c (write_header): Likewise.
1971 * genemit.c (main): Likewise.
1972 * genextract.c (print_header): Likewise.
1973 * genopinit.c (main): Likewise.
1974 * genoutput.c (output_prologue): Likewise.
1975 * genpeep.c (main): Likewise.
1976 * genpreds.c (write_insn_preds_c): Likewise.
1977 * genrecog.c (writer_header): Likewise.
1978 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1979 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1980 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1981 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1982 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1983 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1984 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1985 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1986 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1987 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1988 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1989 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1990 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1991 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1992 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1993 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1994 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1995 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1996 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1997 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1998 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1999 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2000 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2001 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2002 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2003 * config/darwin.c (IN_TARGET_CODE): Likewise.
2004 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2005 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2006 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2007 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2008 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2009 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2010 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2011 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2012 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2013 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2014 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2015 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2016 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2017 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2018 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2019 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2020 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2021 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2022 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2023 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2024 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2025 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2026 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2027 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2028 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2029 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2030 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2031 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2032 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2033 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2034 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2035 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2036 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2037 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2038 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2039 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2040 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2041 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2042 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2043 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2044 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2045 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2046 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2047 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2048 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2049 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2050 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2051 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2052 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2053 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2054 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2055 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2056 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2057 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2058 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2059 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2060 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2061 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2062 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2063 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2064 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2065 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2066 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2067 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2068 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2069 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2070 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2071 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2072 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2073 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2074 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2075 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2076 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2077 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2078 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2079 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2080 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2081 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2082 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2083 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2084 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2085 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2086 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2087 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2088 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2089 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2090 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2091 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2092 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2093 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2094 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2095 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2096 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2097 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2098 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2099 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2100 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2101 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2102 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2103 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2104 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2105 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2106 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2107 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2108 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2109 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2111 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2112 Alan Hayward <alan.hayward@arm.com>
2113 David Sherwood <david.sherwood@arm.com>
2115 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2118 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2119 Alan Hayward <alan.hayward@arm.com>
2120 David Sherwood <david.sherwood@arm.com>
2122 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2123 * doc/md.texi (vec_series@var{m}): Document.
2124 * tree.def (VEC_SERIES_EXPR): New tree code.
2125 * tree.h (build_vec_series): Declare.
2126 * tree.c (build_vec_series): New function.
2127 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2128 * tree-pretty-print.c (dump_generic_node): Likewise.
2129 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2130 * tree-inline.c (estimate_operator_cost): Likewise.
2131 * expr.c (expand_expr_real_2): Likewise.
2132 * optabs-tree.c (optab_for_tree_code): Likewise.
2133 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2134 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2135 * expmed.c (make_tree): Handle VEC_SERIES.
2136 * optabs.def (vec_series_optab): New optab.
2137 * optabs.h (expand_vec_series_expr): Declare.
2138 * optabs.c (expand_vec_series_expr): New function.
2139 * tree-vect-generic.c (expand_vector_operations_1): Check that
2140 the operands also have vector type.
2142 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2143 Alan Hayward <alan.hawyard@arm.com>
2144 David Sherwood <david.sherwood@arm.com>
2146 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2147 (VEC_COND_EXPR): Add missing @tindex.
2148 * doc/md.texi (vec_duplicate@var{m}): Document.
2149 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2150 * tree.c (build_vector_from_val): Add stubbed-out handling of
2151 variable-length vectors, using VEC_DUPLICATE_EXPR.
2152 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2153 * cfgexpand.c (expand_debug_expr): Likewise.
2154 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2155 * tree-inline.c (estimate_operator_cost): Likewise.
2156 * tree-pretty-print.c (dump_generic_node): Likewise.
2157 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2158 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2159 (test_vec_duplicate_folding): New function.
2160 (fold_const_c_tests): Call it.
2161 * optabs.def (vec_duplicate_optab): New optab.
2162 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2163 * optabs.h (expand_vector_broadcast): Declare.
2164 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2165 vec_duplicate_optab.
2166 * expr.c (store_constructor): Try using vec_duplicate_optab for
2168 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2170 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2173 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2174 div/mod latencies a bit.
2176 2017-12-15 Jeff Law <law@redhat.com>
2178 PR tree-optimization/36550
2179 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2180 (mark_threaded_blocks): Rewrite code to avoid block copying when
2181 optimizing for size. Don't pessimize blocks which will be
2182 copied, but all the statements will be dead.
2184 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2186 PR tree-optimization/81165
2187 * tree-ssa-threadupdate.c (uses_in_bb): New.
2188 (estimate_threading_killed_stmts): New.
2189 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2190 * tree-ssa-threadedge.c
2191 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2194 2017-12-15 Jeff Law <law@redhat.com>
2196 PR tree-optimization/83410
2197 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2198 threads when parallelizing loops.
2200 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2202 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2204 * config/alpha/alpha.c (vms_attribute_table): Swap
2205 affects_type_identity and handler fields, adjust comments.
2206 * config/mips/mips.c (mips_attribute_table): Likewise.
2207 * config/visium/visium.c (visium_attribute_table): Likewise.
2208 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2209 * config/microblaze/microblaze.c (microblaze_attribute_table):
2211 * config/spu/spu.c (spu_attribute_table): Likewise.
2212 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2213 * config/arc/arc.c (arc_attribute_table): Likewise.
2214 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2215 * config/v850/v850.c (v850_handle_interrupt_attribute,
2216 v850_handle_data_area_attribute): Formatting fixes.
2217 (v850_attribute_table): Swap affects_type_identity and handler
2218 fields, adjust comments.
2219 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2220 * config/arm/arm.c (arm_attribute_table): Likewise.
2221 * config/avr/avr.c (avr_attribute_table): Likewise.
2222 * config/s390/s390.c (s390_attribute_table): Likewise.
2223 * config/sh/sh.c (sh_attribute_table): Likewise.
2224 * config/i386/i386.c (ix86_handle_cconv_attribute,
2225 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2226 (ix86_attribute_table): Swap affects_type_identity and handler
2227 fields, adjust comments.
2228 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2229 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2230 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2231 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2232 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2233 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2234 * config/rx/rx.c (rx_attribute_table): Likewise.
2235 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2236 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2237 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2238 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2239 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2240 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2241 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2242 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2243 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2244 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2245 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2247 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2249 (register_scoped_attributes, decl_attributes): Formatting fixes.
2251 PR tree-optimization/83269
2252 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2253 subtraction in arg0's type if type is signed and arg0 is unsigned.
2257 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2258 convert? on @0 instead of convert. Check type of @1, not @0.
2259 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2260 convert? on @0 instead of convert. Check type of @1, not @0.
2261 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2262 only optimize if either both @1 and @2 types are narrower
2263 precision, or both are wider or equal precision, and in the former
2264 case only if both have undefined overflow.
2266 2017-12-15 Richard Biener <rguenther@suse.de>
2269 * internal-fn.def (IFN_NOP): Add.
2270 * internal-fn.c (expand_NOP): Do nothing.
2271 * lto-streamer-in.c (input_function): Instead of removing
2272 sanitizer calls replace them with IFN_NOP calls.
2274 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2275 Alan Hayward <alan.hayward@arm.com>
2276 David Sherwood <david.sherwood@arm.com>
2278 * dse.c (store_info, read_info_type): Replace begin and end with
2280 (print_range): New function.
2281 (set_all_positions_unneeded, any_positions_needed_p)
2282 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2284 (record_store): Likewise. Optimize the case in which all positions
2286 (get_stored_val): Replace read_begin and read_end with read_offset
2288 (replace_read): Update call accordingly.
2290 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2292 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2293 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2294 (loop_cand::loop_cand): Initialize above members.
2295 (loop_cand::supported_operations): Delete.
2296 (loop_cand::can_interchange_p): Inline above function.
2297 (loop_cand::classify_simple_reduction): Record number of constant
2298 initialized simple reductions.
2299 (should_interchange_loops): New parameters. Check stmt cost of loops
2301 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2302 Update call to should_interchange_loops.
2303 (should_interchange_loop_nest): Update call to
2304 should_interchange_loops.
2306 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2309 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2310 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2311 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2313 2017-12-15 Julia Koval <julia.koval@intel.com>
2315 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2316 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
2317 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2318 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2319 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2321 2017-12-15 Julia Koval <julia.koval@intel.com>
2323 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2324 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2325 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2326 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2327 _mm_aesenc_epi128): New intrinsics.
2329 2017-12-15 Julia Koval <julia.koval@intel.com>
2331 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2332 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
2333 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2334 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2335 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2337 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2339 * gimple-ssa-strength-reduction.c (analyze_increments):
2340 Distinguish replacement costs for constant strides from those for
2343 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2345 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2347 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2349 * doc/invoke.texi: Document -Wcast-function-type.
2350 * recog.h (stored_funcptr): Change signature.
2351 * tree-dump.c (dump_node): Avoid warning.
2352 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2354 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2357 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2360 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2362 PR tree-optimization/83312
2363 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2364 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2365 "val" param, and to cope with arbitrary basic blocks.
2366 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2367 handle NULL_TREE for "val", dropping "bb" param.
2368 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2369 drop "bb" param. Handle NULL_TREE for "val".
2370 (find_case_label_for_value): Make "switch_stmt" param const.
2371 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2373 (vrp_prop::check_all_array_refs): Reimplement as...
2374 (check_array_bounds_dom_walker::before_dom_children): ...this new
2375 vfunc. Replace linear search through BB block list, excluding
2376 those with non-executable in-edges via dominator walk.
2378 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2380 * config/arm/arm.opt (mverbose-cost-dump): New option.
2381 * config/arm/arm.c (arm_rtx_costs): Use it.
2383 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2386 * reload1.c (emit_input_reload_insns): Skip debug markers.
2388 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2390 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2394 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2395 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2396 * config/sh/sh.c: Skip debug insns besides notes.
2397 * config/sh/sh.md: Likewise.
2398 * config/sh/sh_treg_combine.cc: Likewise.
2399 * config/sh/sync.md: Likewise.
2401 2017-12-14 Tom de Vries <tom@codesourcery.com>
2403 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2404 Add item for weak_undefined.
2406 2017-12-14 Richard Biener <rguenther@suse.de>
2408 PR tree-optimization/67842
2409 * sese.h (bb_in_region): Remove #if 0'ed code.
2411 2017-12-14 Richard Biener <rguenther@suse.de>
2413 PR tree-optimization/83326
2414 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2415 may_be_zero parameter and handle it by not marking the first
2416 peeled copy as not exiting the loop.
2417 (try_peel_loop): Likewise.
2418 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2419 to handle the case of constant or zero iterations and perform
2420 loop header copying on-the-fly.
2422 2017-12-14 Richard Biener <rguenther@suse.de>
2424 PR tree-optimization/83418
2425 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2426 Instead of asserting we don't get unfolded comparisons deal with
2429 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2432 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2433 basic blocks. Assert debug bind insns don't appear outside of bbs,
2434 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2437 PR tree-optimization/83198
2438 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2439 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
2440 value if arg is a REAL_CST with incompatible type.
2442 2017-12-14 Sudakshina Das <sudi.das@arm.com>
2443 Bin Cheng <bin.cheng@arm.com>
2446 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
2448 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
2451 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2453 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
2454 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
2456 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
2457 Alan Hayward <alan.hayward@arm.com>
2458 David Sherwood <david.sherwood@arm.com>
2460 * poly-int.h: New file.
2461 * poly-int-types.h: Likewise.
2462 * coretypes.h: Include them.
2463 (POLY_INT_CONVERSION): Define.
2464 * target.def (estimated_poly_value): New hook.
2465 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
2466 * doc/tm.texi: Regenerate.
2467 * doc/poly-int.texi: New file.
2468 * doc/gccint.texi: Include it.
2469 * doc/rtl.texi: Describe restrictions on subreg modes.
2470 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
2471 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
2472 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
2473 * targhooks.h (default_estimated_poly_value): Declare.
2474 * targhooks.c (default_estimated_poly_value): New function.
2475 * target.h (estimated_poly_value): Likewise.
2476 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
2477 (wi::unary_traits): Delete.
2478 (wi::binary_traits::signed_shift_result_type): Define for
2479 offset_int << HOST_WIDE_INT, etc.
2480 (generic_wide_int::operator <<=): Define for all types and use
2481 wi::lshift instead of <<.
2482 (wi::hwi_with_prec): Add a default constructor.
2483 (wi::ints_for): New class.
2484 (operator <<): Define for all wide-int types.
2485 (operator /): New function.
2486 (operator %): Likewise.
2487 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
2488 (ASSERT_MAYBE_NE_AT): New macros.
2490 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
2491 Dominik Vogt <vogt@linux.vnet.ibm.com>
2494 * emit-rtl.c (init_emit): Remove ??? comment.
2495 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2496 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
2497 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
2498 alignment of 3 virtual registers to BITS_PER_WORD.
2500 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
2502 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
2504 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
2505 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
2507 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2510 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
2512 2017-12-13 Jakub Jelinek <jakub@redhat.com>
2514 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
2515 with the exception of debug begin stmt markers appear before
2519 * final.c (rest_of_handle_final): Call variable_tracking_main only
2520 if !flag_var_tracking.
2522 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
2523 Jakub Jelinek <jakub@redhat.com>
2527 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
2528 labels debug stmts that can only appear after labels.
2530 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
2532 PR rtl-optimization/82398
2533 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
2534 EXPR_USEFULNESS in priority comparison.
2536 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
2538 PR rtl-optimization/83393
2539 * combine.c (move_deaths): If reg_stat points to a too new insn in
2540 last_death, do not use it: find the proper insn instead.
2542 2017-12-12 Jeff Law <law@redhat.com>
2544 PR tree-optimization/83298
2545 PR tree-optimization/83362
2546 PR tree-optimization/83383
2547 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
2548 push_value_range a public interface. Add new argument to
2549 record_ranges_from_stmt.
2550 * gimple-ssa-evrp-analyze.c
2551 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
2552 Update comments. Handle recording temporary equivalences.
2553 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
2554 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
2555 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
2556 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
2557 gimple-ssa-evrp-analyze.h.
2558 (record_temporary_equivalences_from_phis): Add new argument. When
2559 the PHI arg is an SSA_NAME, set the result's range to the range
2561 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
2562 from statements too.
2563 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
2564 Pass it down to children as needed.
2565 (thread_outgoing_edges): Likewise.
2566 (thread_across_edge): Likewise. Push/pop range state as needed.
2567 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
2569 2017-12-12 Julia Koval <julia.koval@intel.com>
2571 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
2572 (PTA_CANNONLAKE): Remove PTA_CLWB.
2574 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2577 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
2578 (vcondv2div2df): Likewise.
2579 (vconduv2dfv2di): Likewise.
2581 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2583 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
2584 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
2585 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
2586 ATTR_MATHFN_FPROUNDING_ERRNO.
2588 2017-12-12 Richard Biener <rguenther@suse.de>
2590 PR tree-optimization/83385
2591 * graphite-scop-detection.c (get_order, order): Remove.
2592 (bb_to_rpo): New global.
2594 (build_scops): Sort pbbs in RPO order.
2596 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
2598 * combine.c (simplify_set): Do not transform subregs to zero_extends
2599 if the destination mode is a vector mode.
2601 2017-12-12 Jakub Jelinek <jakub@redhat.com>
2603 PR tree-optimization/83359
2604 * tree-cfg.h (fold_loop_internal_call): Declare.
2605 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
2606 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
2607 (find_loop_dist_alias): New function.
2608 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
2609 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
2612 PR tree-optimization/80631
2613 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
2614 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
2615 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
2616 hardcoding zero as the value if COND_EXPR is never true. For
2617 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
2618 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
2619 hardcoding MAX_EXPR as the reduction operation.
2620 (is_nonwrapping_integer_induction): Allow negative step.
2621 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
2622 vect_create_epilog_for_reduction, if no value is suitable, don't
2623 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
2625 2017-12-12 Richard Biener <rguenther@suse.de>
2627 PR tree-optimization/81889
2628 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
2629 range info from the non-wrapping IV instead of just the range
2632 2017-12-12 Julia Koval <julia.koval@intel.com>
2634 * config.gcc: Add vaesintrin.h.
2635 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2637 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2638 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2640 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2641 * config/i386/immintrin.h: Include vaesintrin.h.
2642 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2643 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2644 _mm_aesdec_epi128): New intrinsics.
2646 2017-12-12 Julia Koval <julia.koval@intel.com>
2648 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
2649 OPTION_MASK_ISA_VAES_UNSET): New.
2650 (ix86_handle_option): Handle -mvaes.
2651 * config/i386/cpuid.h: Define bit_VAES.
2652 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
2653 * config/i386/i386-c.c (__VAES__): New.
2654 * config/i386/i386.c (ix86_target_string): Add -mvaes.
2655 (ix86_valid_target_attribute_inner_p): Ditto.
2656 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
2657 * config/i386/i386.opt: Add -mvaes.
2658 * doc/invoke.texi: Ditto.
2660 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
2662 * debug.h (gcc_debug_hooks): Add inline_entry.
2663 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
2664 * debug.c (do_nothing_debug_hooks): Likewise.
2665 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2666 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
2667 (dwarf2_lineno_debug_hooks): Likewise.
2669 * common.opt (gstatement-frontiers): New, setting
2670 debug_nonbind_markers_p.
2671 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
2672 * toplev.c (process_options): Autodetect value for debug statement
2674 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
2675 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
2677 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
2678 markers. Integrate source bind into debug stmt expand loop.
2679 (pass_expand::execute): Check debug marker limit. Avoid deep
2680 TER and expand debug locations for debug bind insns only.
2681 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
2683 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
2684 * final.c (reemit_insn_block_notes): Take current block from
2685 nonbind markers. Declare note where it's first set.
2686 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
2687 begin stmt markers if enabled.
2688 (notice_source_line): Handle nonbind markers. Fail if their
2689 location is unknown or that of builtins.
2690 (rest_of_handle_final): Convert begin stmt markers to notes if
2691 var-tracking didn't run.
2692 (rest_of_clean_state): Skip begin stmt markers.
2693 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
2695 * function.c (allocate_struct_function): Set begin_stmt_markers.
2696 * function.h (struct function): Add debug_marker_count counter
2697 and debug_nonbind_markers flag.
2698 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
2699 * gimple-low.c (lower_function_body): Adjust
2700 debug_nonbind_markers.
2701 (lower_stmt): Drop or skip gimple debug stmts.
2702 (lower_try_catch): Skip debug stmts.
2703 * gimple.c (gimple_build_debug_begin_stmt): New.
2704 (gimple_copy): Increment debug_marker_count if copying one.
2705 * gimple.h (gimple_build_debug_begin_stmt): Declare.
2706 * gimplify.c (rexpr_location): New.
2707 (rexpr_has_location): New.
2708 (warn_switch_unreachable_r): Handle gimple debug stmts.
2709 (shortcut_cond_r): Call expr_location.
2711 (find_goto_label): New.
2712 (shortcut_cond_expr): Call expr_has_location, expr_location, and
2714 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
2716 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
2717 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
2718 (LANG_HOOKS_INITIALIZER): ... this.
2719 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
2720 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
2722 (update_ebb_live_info): Skip debug insn markers.
2723 * lra.c (debug_insn_static_data): Rename to...
2724 (debug_bind_static_data): ... this.
2725 (debug_marker_static_data): New.
2726 (lra_set_insn_recog_data): Select one of the above depending
2728 (lra_update_isn_regno_info): Don't assume debug insns have
2730 (push_insns): Skip debug insns.
2731 * lto-streamer-in.c (input_function): Drop debug stmts
2732 depending on active options. Adjust debug_nonbind_markers.
2733 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
2734 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
2735 begin stmt marker notes.
2736 (print_insn): Likewise.
2737 * recog.c (extract_insn): Recognize rtl for debug markers.
2738 * rtl.def (DEBUG_MARKER): New.
2739 * tree-inline.c: Include params.h.
2740 (remap_gimple_stmt): Handle nonbind markers.
2741 (maybe_move_debug_stmts_to_successors): Likewise.
2742 (copy_debug_stmt): Likewise.
2743 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
2744 markers regardless of no side effects.
2745 (tsi_link_before): Don't update container's side effects when adding
2746 a begin stmt marker.
2747 (tsi_link_after): Likewise.
2748 (expr_first): Skip begin stmt markers.
2749 (expr_last): Likewise.
2750 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
2751 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2752 Disregard nonbind markers.
2753 * tree.c (make_node_stat): Don't set side effects for begin stmt
2755 (build1_stat): Likewise.
2756 * tree.def (DEBUG_BEGIN_STMT): New.
2757 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
2758 * var-tracking.c (delete_debug_insns): Renamed to...
2759 (delete_vta_debug_insns): ... this.
2760 (reemit_marker_as_note): New.
2761 (vt_initialize): Reemit markers.
2762 (delete_vta_debug_insns): Likewise.
2763 (vt_debug_insns_local): Reemit or delete markers.
2764 (variable_tracking_main_1): Likewise.
2765 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
2766 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
2767 (gimple_debug_nonbind_marker_p): New.
2768 (gimple_build_debug_bind): Adjust.
2769 (gimple_build_debug_begin_stmt): New.
2770 * doc/invoke.texi (max-debug-marker-count): New param.
2771 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2772 (debug_parameter_ref, debug_marker): New.
2773 (NOTE_INSN_BEGIN_STMT): New.
2774 (DEBUG_INSN): Describe begin stmt markers.
2776 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2777 without debug insns, we wouldn't, but clean up debug insns
2778 after a control flow insn nevertheless.
2780 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2781 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2782 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2783 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2784 (rtl_tidy_fallthru_edge): Likewise.
2785 (rtl_verify_fallthru): Likewise.
2786 (rtl_verify_bb_layout): Likewise.
2787 (skip_insns_after_block): Likewise.
2788 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2789 * dwarf2out.c: Include print-rtl.h.
2790 (dwarf2out_next_real_insn): New.
2791 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2792 Dump debug binds in asm comments.
2793 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2794 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2795 callers to use gsi_start_nondebug_bb instead.
2796 (gsi_after_labels): Skip gimple debug stmts.
2797 (gsi_start_nondebug): New.
2798 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2799 (proper_loop_form_for_interchange): Adjust.
2800 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2801 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2802 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2803 (collect_fallthrough_labels): Likewise.
2804 (should_warn_for_implicit_fallthrough): Likewise.
2805 (warn_implicit_fallthrough_r): Likewise.
2806 (expand_FALLTHROUGH_r): Likewise.
2807 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2808 (graphite_copy_stmts_from_block): Skip nonbind markers.
2809 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2810 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2811 * jump.c (clean_barriers): Skip debug insns.
2812 * omp-expand.c (expand_parallel_call): Skip debug insns.
2813 (expand_task_call): Likewise.
2814 (remove_exit_barrier): Likewise.
2815 (expand_omp_taskreg): Likewise.
2816 (expand_omp_for_init_counts): Likewise.
2817 (expand_omp_for_generic): Likewise.
2818 (expand_omp_for_static_nochunk): Likewise.
2819 (expand_omp_for_static_chunk): Likewise.
2820 (expand_omp_simd): Likewise.
2821 (expand_omp_taskloop_for_outer): Likewise.
2822 (expand_omp_taskloop_for_inner): Likewise.
2823 (expand_oacc_for): Likewise.
2824 (expand_omp_sections): Likewise.
2825 (expand_omp_single): Likewise.
2826 (expand_omp_synch): Likewise.
2827 (expand_omp_atomic_load): Likewise.
2828 (expand_omp_atomic_store): Likewise.
2829 (expand_omp_atomic_fetch_op): Likewise.
2830 (expand_omp_atomic_pipeline): Likewise.
2831 (expand_omp_atomic_mutex): Likewise.
2832 (expand_omp_target): Likewise.
2833 (grid_expand_omp_for_loop): Likewise.
2834 (grid_expand_target_grid_body): Likewise.
2835 (build_omp_regions_1): Likewise.
2836 * omp-low.c (check_combined_parallel): Skip debug stmts.
2837 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2838 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2839 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2841 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2842 (make_edges): Likewise.
2843 (cleanup_dead_labels): Likewise.
2844 (gimple_can_merge_blocks_p): Likewise.
2845 (stmt_starts_bb_p): Likewise.
2846 (gimple_block_label): Likewise.
2847 (gimple_redirect_edge_and_branch): Likewise.
2848 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2850 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2852 * tree-pretty-print (print_declaration): Omit initializer in slim
2854 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2856 (eliminate_unnecessary_stmts): Stabilize block removal order.
2857 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2858 * var-tracking.c (get_first_insn): New.
2859 (vt_emit_notes): Call it.
2860 (vt_initialize): Walk any insns before the first BB.
2861 (delete_debug_insns): Likewise.
2863 * gimple.h (enum gimple_debug_subcode): Add
2864 GIMPLE_DEBUG_BEGIN_STMT.
2865 (gimple_debug_begin_stmt_p): New.
2866 (gimple_debug_nonbind_marker_p): New.
2867 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2868 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2869 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2870 * insn-notes.def (BEGIN_STMT): New.
2871 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2872 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2873 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2874 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2875 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2876 (INSN_DEBUG_MARKER_KIND): New.
2877 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2878 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2879 (INSN_VAR_LOCATION_PTR): New.
2880 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2882 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2883 deps only if debug bind insns are enabled.
2884 (pass_expand::execute): Avoid deep TER and expand
2885 debug locations for debug bind insns only.
2886 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2887 debug stmts special handling down to debug bind stmts.
2888 * combine.c (try_combine): Narrow debug insns special handling
2889 down to debug bind insns.
2890 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2891 Narrow debug insns preexisting special handling down to debug
2893 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2894 handling down to debug bind insns.
2895 * function.c (instantiate_virtual_regs): Skip debug markers,
2896 adjust handling of debug binds.
2897 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2898 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2899 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2900 insns to debug bind insns.
2901 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2902 special handling of debug stmts to debug bind stmts.
2903 * ipa-split.c (split_function): Likewise.
2904 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2905 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2907 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2908 * regrename.c (build_def_use): Likewise.
2909 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2910 (pass_cprop_hardreg): Narrow special casing of debug insns to
2912 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2913 * reload1.c (reload): Likewise.
2914 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2915 casing of debug stmts to debug bind stmts.
2916 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2917 * ssa-iterators.h (num_imm_uses): Likewise.
2918 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2919 debug stmts to debug bind stmts.
2920 * tree-inline.c (tree_function_versioning): Narrow special casing
2921 of debug stmts to debug bind stmts.
2922 * tree-loop-distribution.c (generate_loops_for_partition):
2923 Narrow special casing of debug stmts to debug bind stmts.
2924 * tree-sra.c (analyze_access_subtree): Narrow special casing
2925 of debug stmts to debug bind stmts.
2926 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2927 stmts to debug bind stmts.
2928 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2929 casing of debug stmts to debug bind stmts.
2930 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2931 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2932 casing of debug stmts to debug bind stmts.
2933 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2935 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2936 debug stmts to debug bind stmts.
2937 (gimple_replace_ssa_lhs): Likewise.
2938 (insert_debug_temp_for_var_def): Likewise.
2939 (insert_debug_temps_for_defs): Likewise.
2940 (reset_debug_uses): Likewise.
2941 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2942 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2943 (adjust_debug_stmts): Likewise.
2944 (adjust_phi_and_debug_stmts): Likewise.
2945 (vect_do_peeling): Likewise.
2946 * tree-vect-loop.c (vect_transform_loop): Likewise.
2947 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2948 * var-tracking.c (adjust_mems): Narrow special casing of debug
2949 insns to debug bind insns.
2950 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2951 (compute_bb_dataflow, vt_find_locations): Likewise.
2952 (vt_expand_loc, emit_notes_for_changes): Likewise.
2953 (vt_init_cfa_base): Likewise.
2954 (vt_emit_notes): Likewise.
2955 (vt_initialize): Likewise.
2956 (vt_finalize): Likewise.
2958 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2959 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2960 (next_nonnote_nondebug_insn_bb): New.
2961 (prev_nonnote_nondebug_insn_bb): New.
2962 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2963 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2964 (prev_nonnote_nondebug_insn_bb): Declare.
2965 (next_nonnote_nondebug_insn_bb): Declare.
2966 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2967 * cfgrtl.c (get_last_bb_insn): Likewise.
2968 * lra.c (push_insns): Likewise.
2970 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2973 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2975 * selftest.c (selftest::test_runner::test_runner): New ctor.
2976 (selftest::test_runner::~test_runner): New dtor.
2977 * selftest.h (class selftest::test_runner): New class.
2979 2017-12-11 Carl Love <cel@us.ibm.com>
2981 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2982 vec_extract_fp32_from_shortl]): Add #defines.
2983 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2984 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2985 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2986 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2987 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2988 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2989 * doc/extend.texi: Add documentation for the added builtins.
2991 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2993 PR rtl-optimization/80693
2994 PR rtl-optimization/81019
2995 PR rtl-optimization/81020
2996 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2997 are not mentioned in i3. Place the REG_UNUSED note on i2,
2998 possibly modified to REG_DEAD, if it did not originate in i3.
3000 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3002 * recog.c (store_data_bypass_p_1): New function.
3003 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3004 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3006 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3008 PR rtl-optimization/83361
3009 * ifcvt.c (if_convert): Call fixup_partitions.
3011 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3013 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3014 early folding of splat_u{8,16,32}.
3016 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3018 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3019 output_operand_lossage first argument with capital letter.
3020 (aarch64_override_options): Don't start error and sorry first argument
3021 with capital letter.
3023 2017-12-11 Andi Kleen <ak@linux.intel.com>
3025 PR gcov-profile/83355
3026 * auto-profile.c (string_table::get_index_by_decl): Don't
3027 recurse when abstract origin points to itself.
3029 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3031 PR tree-optimization/83320
3032 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3033 (prune_datarefs_not_in_loop): Ditto.
3035 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3037 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3039 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3041 PR tree-optimization/83337
3042 * gimple-loop-interchange.cc (compute_access_stride): Handle
3043 bitfield DRs properly.
3045 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3047 PR tree-optimization/83338
3048 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3049 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3050 vector of unsigned integers to vector of signed integers.
3052 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3054 PR rtl-optimization/83317
3055 * lra-constraints.c (process_address_1): Add insn code check.
3057 2017-12-08 Michael Matz <matz@suse.de>
3059 Fix PR tree-optimization/83323
3060 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3061 head-controlled loops and loop BBs.
3062 * common.opt (funroll-and-jam): Remove, instead ...
3063 (floop-unroll-and-jam): ... reuse this option.
3064 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3065 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3066 (-floop-unroll-and-jam): ... this option.
3068 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3070 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3071 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3072 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3074 2017-12-08 Richard Biener <rguenther@suse.de>
3077 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3078 handle accesses outside of zero-sized vars.
3080 2017-12-08 Martin Jambor <mjambor@suse.cz>
3082 PR tree-optimization/83141
3083 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3084 test for MEM_REFs implicitely changing types with padding. Remove
3086 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3088 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3090 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3092 * config/avr/avr.c (avr_attribute_table): Likewise.
3093 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3094 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3095 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3096 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3097 of struct attribute_spec.
3098 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3100 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3103 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3105 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3108 * profile-count.c (profile_count::from_gcov_type): Move from
3109 profile-count.h; handle overflow.
3110 * profile-count.h (profile_count::from_gcov_type): Move offline.
3112 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3114 PR rtl-optimization/83304
3115 * combine.c (move_deaths): If we do not know where a register died,
3118 2017-12-08 Richard Biener <rguenther@suse.de>
3120 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3121 Provide -fopt-info-loop feedback when we interchange in a nest.
3123 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3125 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3126 for armv6 ARM CPU IDs.
3128 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3130 * common/config/arm/arm-common.c: Include <algorithm>.
3131 (INCLUDE_VECTOR): Define.
3132 (compare_opt_names): New function.
3133 (arm_rewrite_selected_arch): Only strip out extensions that can be
3134 expressed through -mfpu. Sort the remaining extensions
3137 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3139 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3140 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3141 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3142 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3144 2017-06-08 Tristan Gingold <gindold@adacore.com>
3147 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3148 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3150 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3152 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3153 of struct attribute_spec.
3155 2017-12-08 Julia Koval <julia.koval@intel.com>
3157 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3158 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3159 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3160 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3161 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3162 _mm_maskz_dpwssds_epi32): Ditto.
3164 2017-12-08 Richard Biener <rguenther@suse.de>
3166 PR tree-optimization/81303
3167 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3168 conditions try to create a comparison vector type matching
3169 the data vector type.
3170 (vectorizable_condition): Adjust.
3171 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3172 Leave invariant conditions alone in case we can vectorize those.
3174 2017-12-08 Julia Koval <julia.koval@intel.com>
3176 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3177 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3178 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3179 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3180 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3182 2017-12-08 Julia Koval <julia.koval@intel.com>
3184 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3185 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3186 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3187 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3188 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3189 _mm_maskz_dpbusds_epi32): New intrinsics.
3191 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3193 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3195 (ld<bh>io_signed): Likewise.
3196 (st<bhw>io): Likewise.
3197 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3198 SMALL_INT12 constant integer operand.
3200 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3202 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3204 (nios2_large_constant_allowed): ...this. Adjust uses.
3205 (nios2_plus_symbolic_constant_p): Rename to...
3206 (nios2_plus_large_constant_p): ...this. Adjust uses.
3207 (nios2_legitimate_address_p): Correct CONST_INT handling.
3208 (nios2_symbolic_memory_operand_p): Rename to...
3209 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3210 (nios2_large_constant_p): Check for large constant integers too.
3211 (nios2_split_large_constant): Handle constant integers.
3212 (nios2_split_symbolic_memory_operand): Rename to...
3213 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3214 (nios2_legitimize_constant_address): Handle constant integers.
3215 (r0rel_constant_p): Handle small constant integers.
3216 (nios2_print_operand_address): Handle r0-relative integer addresses.
3217 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3218 * config/nios2/nios2.md: Adjust for renamed functions.
3220 2017-12-07 Andrew Waterman <andrew@sifive.com>
3222 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3223 (TARGET_HAVE_SRODATA_SECTION): New define.
3224 (riscv_select_section): New function.
3226 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3227 Alexander Monakov <amonakov@ispras.ru>
3228 Jakub Jelinek <jakub@redhat.com>
3231 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3233 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3235 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3236 a bool success value. Don't call output_operand_lossage here.
3237 (aarch64_print_ldpstp_address): Return a bool success value.
3238 (aarch64_print_operand_address): Call output_addr_const if
3239 aarch64_print_address_internal fails.
3240 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3241 'y'; call output_operand_lossage instead. Call output_operand_lossage
3242 if aarch64_print_ldpstp_address fails.
3244 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3246 * tree-vector-builder.h
3247 (tree_vector_builder::binary_encoded_nelts): Declare.
3248 * tree-vector-builder.c
3249 (tree_vector_builder::binary_encoded_nelts): New function.
3250 * fold-const.c (negate_expr_p): Likewise.
3251 (operand_equal_p, fold_checksum_tree): Likewise.
3252 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3253 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3254 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3255 (uniform_vector_p): Likewise.
3256 * varasm.c (const_hash_1, compare_constant): Likewise.
3257 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3258 (valid_lattice_transition): Operate directly on the VECTOR_CST
3260 * ipa-icf.c: Include tree-vector-builder.h.
3261 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3262 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3264 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3266 * tree.c (build_vector): Delete.
3267 * tree.h (build_vector): Make static and move into the self-testing
3270 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3272 * vector-builder.h (vector_builder::derived): New const overload.
3273 (vector_builder::elt): New function.
3274 * tree-vector-builder.h (tree_vector_builder::type): New function.
3275 (tree_vector_builder::apply_step): Declare.
3276 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3278 * gimple-fold.h (tree_vector_builder): Declare.
3279 (gimple_build_vector): Take a tree_vector_builder instead of a
3280 type and vector of elements.
3281 * gimple-fold.c (gimple_build_vector): Likewise.
3282 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3284 (get_initial_defs_for_reduction): Likewise.
3285 (vectorizable_induction): Likewise.
3287 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3289 * tree-vector-builder.h
3290 (tree_vector_builder::new_binary_operation): Declare.
3291 * tree-vector-builder.c
3292 (tree_vector_builder::new_binary_operation): New function.
3293 * fold-const.c (fold_relational_const): Use it.
3294 (const_binop): Likewise. Check that both input vectors have
3295 the same number of elements, thus excluding things like WIDEN_SUM.
3296 Check whether it is possible to operate directly on the encodings
3299 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3301 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3302 new_unary_operation, operating only on the encoded elements.
3303 (const_unop): Likewise.
3304 (exact_inverse): Likewise.
3305 (distributes_over_addition_p): New function.
3306 (const_binop): Use tree_vector_builder and new_unary_operation
3307 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3308 on the encoded elements unless the encoding is strided and the
3309 operation does not distribute over addition.
3310 (fold_convert_const): Use tree_vector_builder and
3311 new_unary_operation. Operate only on the encoded elements
3312 for truncating integer conversions, or for non-stepped encodings.
3314 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3316 * config/sparc/sparc.c: Include tree-vector-builder.h.
3317 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3318 * expmed.c: Include tree-vector-builder.h.
3319 (make_tree): Use tree_vector_builder instead of build_vector.
3320 * fold-const.c: Include tree-vector-builder.h.
3321 (const_binop): Use tree_vector_builder instead of build_vector.
3322 (const_unop): Likewise.
3323 (native_interpret_vector): Likewise.
3324 (fold_vec_perm): Likewise.
3325 (fold_ternary_loc): Likewise.
3326 * gimple-fold.c: Include tree-vector-builder.h.
3327 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3329 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3330 (simplify_vector_constructor): Use tree_vector_builder instead
3332 * tree-vect-generic.c: Include tree-vector-builder.h.
3333 (add_rshift): Use tree_vector_builder instead of build_vector.
3334 (expand_vector_divmod): Likewise.
3335 (optimize_vector_constructor): Likewise.
3336 * tree-vect-loop.c: Include tree-vector-builder.h.
3337 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3338 of build_vector. Explicitly use a stepped encoding for
3340 * tree-vect-slp.c: Include tree-vector-builder.h.
3341 (vect_get_constant_vectors): Use tree_vector_builder instead
3343 (vect_transform_slp_perm_load): Likewise.
3344 (vect_schedule_slp_instance): Likewise.
3345 * tree-vect-stmts.c: Include tree-vector-builder.h.
3346 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3347 (vect_gen_perm_mask_any): Likewise.
3348 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3349 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3351 (build_vector_from_val): Likewise. Explicitly use a duplicate
3354 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3356 * doc/generic.texi (VECTOR_CST): Describe new representation of
3358 * vector-builder.h: New file.
3359 * tree-vector-builder.h: Likewise.
3360 * tree-vector-builder.c: Likewise.
3361 * Makefile.in (OBJS): Add tree-vector-builder.o.
3362 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3363 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3364 (tree_vector): Change the number of elements to
3365 vector_cst_encoded_nelts.
3366 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3367 (VECTOR_CST_ELTS): Delete.
3368 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3369 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3370 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3371 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3372 (VECTOR_CST_ENCODED_ELT): Likewise.
3373 (vector_cst_encoded_nelts): New function.
3374 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3375 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3376 (vector_cst_int_elt, vector_cst_elt): Declare.
3377 * tree.c: Include tree-vector-builder.h.
3378 (tree_code_size): Abort if passed VECTOR_CST.
3379 (tree_size): Update for new VECTOR_CST layout.
3380 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3381 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3382 (build_vector): Use tree_vector_builder.
3383 (vector_cst_int_elt, vector_cst_elt): New functions.
3384 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3385 encoded elements and then create the vector in the canonical form.
3386 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3387 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3388 (tree_c_tests): Call test_vector_cst_patterns.
3389 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3391 (hash_tree): Likewise.
3392 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3393 (streamer_write_tree_header): Likewise.
3394 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3395 (streamer_alloc_tree): Likewise. Update call to make_vector.
3396 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3398 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3400 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3401 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3402 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3404 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3405 Richard Biener <rguenther@suse.de>
3407 PR tree-optimization/81303
3408 * Makefile.in (gimple-loop-interchange.o): New object file.
3409 * common.opt (floop-interchange): Reuse the option from graphite.
3410 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3411 -floop-interchange and mention it for -O3.
3412 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3413 * gimple-loop-interchange.cc: New file.
3414 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3415 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3416 * passes.def (pass_linterchange): New pass.
3417 * timevar.def (TV_LINTERCHANGE): New time var.
3418 * tree-pass.h (make_pass_linterchange): New declaration.
3419 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3420 interchange. Record IV before/after increment in new parameters.
3421 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3422 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3424 (check_reduction_path): ...New function here.
3425 * tree-vectorizer.h (check_reduction_path): New declaration.
3427 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3430 PR rtl-optimization/80818
3431 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3432 always early clobbered.
3433 * lra-lives.c (process_bb_lives): Check input hard regs for early
3434 clobbered non-operand hard reg.
3436 2017-12-07 Jakub Jelinek <jakub@redhat.com>
3439 * tree-cfg.c (verify_gimple_assign_binary): Don't require
3440 types_compatible_p, just that TYPE_MODE is the same.
3442 2017-12-07 Martin Sebor <msebor@redhat.com>
3445 * attribs.c (empty_attribute_table): Initialize new member of
3446 struct attribute_spec.
3447 (decl_attributes): Add argument. Handle mutually exclusive
3448 combinations of attributes.
3449 (selftests::test_attribute_exclusions): New function.
3450 (selftests::attribute_c_tests): Ditto.
3451 * attribs.h (decl_attributes): Add default argument.
3452 * selftest.h (attribute_c_tests): Declare.
3453 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
3454 * tree-core.h (attribute_spec::exclusions, exclude): New type and
3456 * doc/extend.texi (Common Function Attributes): Update const and pure.
3457 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
3458 of struct attribute_spec.
3459 * config/arc/arc.c (arc_attribute_table): Same.
3460 * config/arm/arm.c (arm_attribute_table): Same.
3461 * config/avr/avr.c ( avr_attribute_table): Same.
3462 * config/bfin/bfin.c (bfin_attribute_table): Same.
3463 * config/cr16/cr16.c (cr16_attribute_table): Same.
3464 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
3465 * config/h8300/h8300.c (h8300_attribute_table): Same.
3466 * config/i386/i386.c (ix86_attribute_table): Same.
3467 * config/ia64/ia64.c (ia64_attribute_table): Same.
3468 * config/m32c/m32c.c (m32c_attribute_table): Same.
3469 * config/m32r/m32r.c (m32r_attribute_table): Same.
3470 * config/m68k/m68k.c (m68k_attribute_table): Same.
3471 * config/mcore/mcore.c (mcore_attribute_table): Same.
3472 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
3473 * config/mips/mips.c (mips_attribute_table): Same.
3474 * config/msp430/msp430.c (msp430_attribute_table): Same.
3475 * config/nds32/nds32.c (nds32_attribute_table): Same.
3476 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
3477 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
3478 * config/rl78/rl78.c (rl78__attribute_table): Same.
3479 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
3480 * onfig/rx/rx.c (rx_attribute_table): Same.
3481 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
3482 * config/sh/sh.c (sh_attribute_table): Same.
3483 * config/sparc/sparc.c (sparc_attribute_table): Same.
3484 * config/spu/spu.c (spu_attribute_table): Same.
3485 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
3486 * config/v850/v850.c (v850_attribute_table): Same.
3487 * config/visium/visium.c (visium_attribute_table): Same.
3489 2017-12-07 Tamar Christina <tamar.christina@arm.com>
3492 * config/arm/arm.c (INCLUDE_STRING): Define.
3493 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
3494 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
3497 2017-12-07 Michael Matz <matz@suse.de>
3499 Add unroll and jam pass
3501 * gimple-loop-jam.c: New file.
3502 * Makefile.in (OBJS): Add gimple-loop-jam.o.
3503 * common.opt (funroll-and-jam): New option.
3504 * opts.c (default_options_table): Add unroll-and-jam at -O3.
3505 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
3506 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
3507 * passes.def: Add pass_loop_jam.
3508 * timevar.def (TV_LOOP_JAM): Add.
3509 * tree-pass.h (make_pass_loop_jam): Declare.
3510 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
3511 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
3512 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
3513 to flow_loop_tree_node_add.
3514 (duplicate_subloops, copy_loops_to): Append to sibling list.
3515 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
3516 * doc/invoke.texi (-funroll-and-jam): Document new option.
3517 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
3519 2017-12-07 Richard Biener <rguenther@suse.de>
3521 PR tree-optimization/83296
3522 PR tree-optimization/67769
3523 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
3524 flow sensitive info in an unrelated BB.
3525 (value_replacement): Use reset_flow_sensitive_info.
3526 (minmax_replacement): Reset flow sensitive info on the def
3527 we move. Do not reset flow sensitive info in the whole BB
3528 we move the stmt to.
3529 (abs_replacement): Likewise.
3531 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3534 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3535 rs6000_cpu to the given -mcpu=, or to the default processor.
3537 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3539 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
3540 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
3541 instead of rs6000_cpu_attr.
3542 (rs6000_adjust_cost): Ditto.
3543 (is_microcoded_insn): Ditto.
3544 (rs6000_adjust_priority): Ditto.
3545 (rs6000_issue_rate): Ditto.
3546 (rs6000_use_sched_lookahead): Ditto.
3547 (rs6000_use_sched_lookahead_guard): Ditto.
3548 (rs6000_sched_reorder): Ditto.
3549 (force_new_group): Ditto.
3550 * config/rs6000/rs6000.md (cpu attribute): Ditto.
3551 (group_ending_nop): Ditto.
3553 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
3555 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
3556 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
3557 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
3558 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
3560 (rs6000_reassociation_width): Ditto.
3561 (rs6000_emit_epilogue): Ditto.
3562 (rs6000_adjust_cost): Ditto.
3563 (is_microcoded_insn): Ditto.
3564 (is_cracked_insn): Ditto.
3565 (rs6000_adjust_priority): Ditto.
3566 (rs6000_sched_reorder): Ditto.
3567 (rs6000_sched_reorder2): Ditto.
3568 (insn_must_be_first_in_group): Ditto.
3569 (insn_must_be_last_in_group): Ditto.
3570 (rs6000_register_move_cost): Ditto.
3571 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
3574 2017-12-07 Julia Koval <julia.koval@intel.com>
3576 * config.gcc: Add vaesintrin.h.
3577 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3579 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3580 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3582 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3583 * config/i386/immintrin.h: Include vaesintrin.h.
3584 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3585 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3586 _mm_aesdec_epi128): New intrinsics.
3588 2017-12-06 David Malcolm <dmalcolm@redhat.com>
3590 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
3591 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
3592 c-family/c-spellcheck.cc.
3593 (best_macro_match::best_macro_match): Likewise.
3595 (struct edit_distance_traits<cpp_hashnode *>): Move to
3596 c-family/c-spellcheck.h.
3597 (class best_macro_match): Likewise.
3599 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3601 PR tree-optimization/83293
3602 * gimple-ssa-strength-reduction.c (insert_initializers): Use
3603 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
3604 might insert into empty bb.
3607 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
3608 simplify for plus with :c added, and pointer_plus without that.
3609 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
3610 with undefined overflow and the conversion is not widening,
3611 perform negation in utype and only convert to type afterwards.
3612 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
3613 simplify for plus with :c added, and pointer_plus without that.
3614 If type is integral with undefined overflow and the conversion is
3615 not widening, perform minus in utype and only convert to type
3616 afterwards. Move the last pointer_diff_expr simplify into the
3619 2017-12-06 Martin Sebor <msebor@redhat.com>
3621 PR tree-optimization/82646
3622 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
3623 strncpy, not maxlen.
3625 2017-12-06 Martin Sebor <msebor@redhat.com>
3627 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
3630 PR tree-optimization/83075
3631 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
3632 strncat/strncpy don't change length of source string.
3634 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
3637 2017-11-29 Martin Aberg <maberg@gaisler.com>
3639 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3640 to prevent b2bst errata sequence.
3641 (sqrtdf2_fix): Likewise.
3643 2017-12-06 Jakub Jelinek <jakub@redhat.com>
3645 PR tree-optimization/81945
3646 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
3647 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
3648 to dest_cfun has orig_loop_num set, either remap it to the new
3649 loop number if the loop got moved too, or clear it.
3651 2017-12-05 Steve Ellcey <sellcey@cavium.com>
3653 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
3655 (thunderx2t99_nothing): New insn reservation.
3656 (thunderx2t99_mrs): New insn reservation.
3657 (thunderx2t99_multiple): New insn reservation.
3658 (thunderx2t99_alu_basi): Add bfx to reservation.
3659 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
3661 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3664 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
3666 2017-12-05 Bin Cheng <bin.cheng@arm.com>
3668 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
3669 tree-ssa-pre.c::remove_dead_inserted_code.
3670 * tree-ssa-dce.h: New file.
3671 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
3672 (remove_dead_inserted_code): Move and rename to function
3673 tree-ssa-dce.c::simple_dce_from_worklist.
3674 (pass_pre::execute): Update use.
3676 2017-12-05 Richard Biener <rguenther@suse.de>
3678 PR tree-optimization/83277
3679 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
3680 to code-gen liveout vars.
3682 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
3684 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
3685 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
3686 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
3689 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
3691 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
3693 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
3694 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
3697 2017-12-05 Richard Biener <rguenther@suse.de>
3699 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
3700 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
3701 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
3702 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
3704 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3706 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
3708 2017-12-05 Martin Liska <mliska@suse.cz>
3709 Jakub Jelinek <jakub@redhat.com>
3711 * doc/invoke.texi: Document the options.
3712 * flag-types.h (enum sanitize_code): Add
3713 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3714 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
3715 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
3716 * opts.c: Define new sanitizer options.
3717 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
3718 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
3720 2017-12-05 Julia Koval <julia.koval@intel.com>
3722 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
3723 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
3724 (ix86_handle_option): Handle -mavx512vnni.
3725 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
3726 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
3727 * config/i386/i386-c (__AVX512VNNI__): New.
3728 * config/i386/i386.c (ix86_target_string): Handle new option.
3729 (ix86_valid_target_attribute_inner_p): Handle new option.
3730 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
3731 * config/i386/i386.opt (mavx512vnni): New option.
3733 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3736 * athlon.md: Disable for generic.
3737 * haswell.md: Enable for generic.
3738 * i386.c (ix86_sched_init_global): Add core hooks for generic.
3739 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
3741 (ix86_adjust_cost): Move generic to haswell path.
3743 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
3745 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
3746 instead of MEM_P in a couple more places. Fix formatting issues.
3748 2017-12-04 Jim Wilson <jimw@sifive.com>
3750 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
3751 instead of GP_REG_LAST-1.
3752 (riscv_adjust_libcall_cfi_prologue): Likewise.
3753 (riscv_adjust_libcall_cri_epilogue): Likewise.
3754 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
3757 2017-12-04 Luis Machado <luis.machado@linaro.org>
3759 * ipa-pure-const.c (check_decl): Add missing newline.
3760 (state_from_flags): Likewise.
3762 2017-12-04 Jeff Law <law@redhat.com>
3764 PR tree-optimizatin/78496
3765 * gimple-ssa-evrp-analyze.h
3766 (evrp_range_analyzer::get_vr_values): Simplify.
3767 * gimple-ssa-evrp-analyze.c: Corresponding changes.
3768 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3769 and gimple-ssa-evrp-analyze.h.
3770 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3771 (simplify_stmt_for_jump_threading): Copy a blob of code from
3772 tree-vrp.c to use ranges to simplify statements.
3773 (dom_opt_dom_walker::before_dom_children): Call
3774 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3775 (dom_opt_dom_walker::after_dom_children): Similarly for
3776 evrp_range_analyzer::leave.
3777 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3780 * gimple-ssa-evrp-analyze.c
3781 (evrp_range_analyzer::extract_range_from_stmt): Always use
3782 vr_values::update_value_range so preexisting range info is
3783 medged with new range info, even if the new range is VR_VARYING.
3785 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3787 * combine.c: Adjust comment.
3788 (use_crosses_set_p): Delete.
3789 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3790 (try_combine): Ditto.
3792 2017-12-04 Richard Biener <rguenther@suse.de>
3794 PR tree-optimization/83255
3795 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3796 Re-add zero-iteration check.
3798 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3800 PR rtl-optimization/83245
3801 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3802 hard registers as earlyclobber, also if not in an asm.
3804 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3808 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3811 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3812 rs6000_cpu based on cpu_index, not tune_index.
3814 2017-12-04 Richard Biener <rguenther@suse.de>
3816 PR tree-optimization/83238
3817 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3818 code match comment, rejecting invalid SESE regions.
3820 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3822 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3823 require base operand is a REG_POINTER prior to reload on targets
3824 with non-equivalent space registers.
3826 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3828 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3829 (update_specialized_profile): Fix updating of counts.
3830 (perhaps_add_new_callers): Likewise.
3832 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3835 * x86-tune.def: Remove obsolette FIXMEs.
3836 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3837 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3838 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3840 (X86_TUNE_PAD_RETURNS): Disable for generic.
3842 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3844 PR tree-optimization/83170
3845 PR tree-optimization/83241
3846 * gimple-ssa-store-merging.c
3847 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3848 gimple_vuse (ins_stmt) in case it has changed.
3849 (imm_store_chain_info::output_merged_store): Likewise.
3851 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3855 * tree-cfg.c (pass_warn_function_return::execute): Handle
3856 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3857 with BUILTINS_LOCATION.
3861 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3862 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3863 instead of DECL_MODE.
3865 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3866 last argument to rtx pointer.
3867 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3868 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3869 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3870 depending on the chosen ISAs.
3871 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3872 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3873 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3875 * config/i386/sse.md (mov<mode>_internal): Likewise.
3876 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3878 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3880 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3882 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3884 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3885 parameters from prototype.
3886 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3887 parameters. Don't print a comment.
3888 (emit_fusion_gpr_load): Adjust.
3889 (emit_fusion_load_store): Adjust.
3890 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3891 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3892 comment on the second line.
3894 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3897 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3898 rs6000_cpu based on cpu_index, not tune_index.
3900 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3902 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3903 which_alternative instead of which_alternative + 1.
3904 (output_asm_insn): Print an extra tab if the template is short.
3906 2017-12-01 Jim Wilson <jimw@sifive.com>
3908 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3910 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3911 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3913 * doc/tm.texi: Regenerate.
3915 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3918 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3919 whether we can allocate pseudos before trying to fix an address.
3920 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3921 memory address is indexed or indirect.
3922 (floatuns_<mode>si2_hw2): Likewise.
3924 2017-12-01 Jason Merrill <jason@redhat.com>
3926 * Makefile.in (TAGS): Add c-family/*.cc.
3928 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3930 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3931 (call_value_insn): Likewise.
3932 (sibcall_insn): Likewise.
3933 (sibcall_value_insn): Likewise.
3934 (movsi_aarch64): Likewise.
3935 (movdi_aarch64): Likewise.
3936 (add_losym_): Likewise.
3937 (ldr_got_small_): Likewise.
3938 (ldr_got_small_sidi): Likewise.
3939 (ldr_got_small_28k_): Likewise.
3940 (ldr_got_small_28k_sidi): Likewise.
3941 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3942 Move output_addr_const to symbolic case. Add error check.
3944 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3946 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3947 * config/s390/s390.md: Disable MVC merging peephole if it would
3948 disable operand forwarding.
3949 (new peephole2): Split MVCs if it would turn them into up to 2
3952 2017-12-01 Richard Biener <rguenther@suse.de>
3954 PR tree-optimization/83232
3955 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3956 detection of same access. Instead of breaking the group here
3957 do not consider the duplicate. Add comment explaining real fix.
3959 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3961 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3963 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3965 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3967 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3969 * function.h (struct function): Remove cilk_frame_decl,
3970 is_cilk_function and calls_cilk_spawn fields.
3971 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3973 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3974 cilk_elemental field.
3975 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3976 * target.def: Adjust comment.
3977 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3978 Don't test cilk_elemental.
3980 PR tree-optimization/83233
3981 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3982 bswap_stat name for the struct.
3985 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3986 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3987 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3988 SWITCH_BREAK_LABEL_P set on the label.
3989 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3990 added for default case if it was missing and not all cases covered.
3991 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3992 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3994 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3995 to avoid -Wimplicit-fallthrough warning.
3996 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3997 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3999 PR tree-optimization/83221
4000 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4002 (init_reassoc): Formatting fix.
4005 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4006 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4010 * tree-cfg.c: Include asan.h.
4011 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4013 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4015 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4017 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4019 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4020 vpcmpeqd instruction.
4022 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4024 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4025 registers type for 128bit mode.
4027 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4029 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4031 * gengtype-state.c (read_state_pair): Likewise.
4032 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4033 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4034 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4035 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4036 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4037 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4038 * builtins.c (fold_builtin_3): Likewise.
4039 * graphite-scop-detection.c
4040 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4041 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4044 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4047 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4048 parameter. Return true if ignore_decl_rtl_set_p is true and other
4049 conditions are satisfied.
4050 * asan.h (asan_protect_global): Add new parameter.
4051 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4052 to asan_protect_global calls.
4054 2017-11-30 Jim Wilson <jimw@sifive.com>
4056 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4057 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4058 -mstrict-align, add info on default value. Delete redundant lines for
4059 -mabi. Add missing -mexplicit-relocs docs.
4061 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4063 * config/arc/arc.md (trap): New pattern.
4065 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4067 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4068 instruction to end into a delay slot.
4069 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4070 can be placed into a delay slot against reg_note.
4072 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4074 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4075 labels number of usages.
4077 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4079 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4081 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4083 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4086 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4087 multiplication by power of 2 constant into two shifts + comparison.
4089 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4092 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4094 2017-11-30 Richard Biener <rguenther@suse.de>
4096 PR tree-optimization/83202
4097 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4098 allow_peel argument and guard peeling.
4099 (canonicalize_loop_induction_variables): Likewise.
4100 (canonicalize_induction_variables): Pass false.
4101 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4102 peeling from cunrolli.
4104 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4106 * combine.c (try_combine): Print a message to dump file whenever
4107 I0, I1, or I2 cannot be combined into I3.
4109 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4111 PR rtl-optimization/83156
4112 PR rtl-optimization/82621
4113 * combine.c (try_combine): Don't split an I2 if one of the dests is
4114 set again before I3. Allow unused dests.
4116 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4118 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4120 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4122 PR rtl-optimization/80818
4123 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4124 recursively. Use insn code for clobber.
4125 (lra_set_insn_recog_data): Pass the new arg to
4126 collect_non_operand_hard_regs.
4127 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4129 (lra_update_insn_regno_info): Pass insn to
4130 add_regs_to_insn_regno_info.
4132 2017-11-29 Jim Wilson <jimw@sifive.com>
4133 Andrew Waterman <andrew@sifive.com>
4135 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4136 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4137 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4138 use SINGLE_SHIFT_COST.
4139 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4140 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4142 2017-11-29 Julia Koval <julia.koval@intel.com>
4144 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4145 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4146 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4147 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4148 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4149 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4150 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4151 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4152 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4153 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4154 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4155 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4156 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4157 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4158 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4159 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4160 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4161 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4162 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4163 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4164 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4165 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4166 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4167 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4168 __builtin_ia32_vpshldv_v2di_mask,
4169 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4170 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4171 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4173 2017-11-29 Julia Koval <julia.koval@intel.com>
4175 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4176 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4177 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4178 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4179 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4180 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4181 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4182 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4183 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4184 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4185 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4186 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4187 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4188 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4189 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4190 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4191 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4192 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4193 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4194 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4195 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4197 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4199 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4200 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4203 2017-11-29 Martin Aberg <maberg@gaisler.com>
4205 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4206 to prevent b2bst errata sequence.
4207 (sqrtdf2_fix): Likewise.
4209 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4211 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4212 (div_sqrt_insn_p): New function.
4213 (sparc_do_work_around_errata): Insert NOP instructions to
4214 prevent sequences that could trigger the TN-0013 errata for
4215 certain LEON3 processors.
4216 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4217 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4218 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4219 (in_branch_delay): Prevent div and sqrt in delay slot if
4221 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4223 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4225 * config/sparc/sparc.c (atomic_insn_p): New function.
4226 (sparc_do_work_around_errata): Insert NOP instructions to
4227 prevent sequences that could trigger the TN-0010 errata for
4229 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4230 instruction referable in atomic_insns_p.
4232 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4234 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4235 (atomic_compare_and_swap_leon3_1): Likewise.
4238 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4240 * config/sparc/sparc.c (fpop_insn_p): New function.
4241 (sparc_do_work_around_errata): Insert NOP instructions to
4242 prevent sequences that could trigger the TN-0012 errata for
4244 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4245 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4246 (in_branch_annul_delay): Prevent floating-point instructions
4247 in delay slot of annulled integer branch.
4249 2017-11-29 Richard Biener <rguenther@suse.de>
4251 PR tree-optimization/83202
4252 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4254 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4255 nodes vectorized to the same stmts multiple times.
4256 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4257 (vect_analyze_slp_instance): Adjust.
4258 (scalar_stmts_to_slp_tree_map_t): New typedef.
4259 (vect_schedule_slp_instance): Add a map recording the SLP node
4260 representing the vectorized stmts for a set of scalar stmts.
4261 Avoid code-generating redundancies.
4262 (vect_schedule_slp): Allocate map and pass it down.
4264 2017-11-29 Nathan Sidwell <nathan@acm.org>
4267 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4270 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4272 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4273 cases for printing LDP/STP memory addresses.
4274 (aarch64_print_address_internal): Renamed from
4275 aarch64_print_operand_address, added parameter, add Pmode check.
4276 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4277 (aarch64_print_operand_address): Indirect to
4278 aarch64_print_address_internal.
4279 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4280 'y' operand output specifier.
4282 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4285 * tree.c (build_simple_mem_ref_loc): Handle
4286 get_addr_base_and_unit_offset returning a MEM_REF.
4289 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4292 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4295 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4297 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4299 (layout::move_to_column): Likewise.
4301 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4304 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4305 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4307 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4308 Martin Jambor <mjambor@suse.cz>
4311 * tree.h (expr_type_first_operand_type_p): Declare
4312 * tree.c (expr_type_first_operand_type_p): New function.
4313 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4314 (ipa_value_from_jfunc): Adjust declaration.
4315 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4316 Use it as result type for arithmetics, unless it is NULL in which case
4317 be more conservative.
4318 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4319 ipa_get_jf_pass_through_result.
4320 (propagate_vals_across_pass_through): Likewise.
4321 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4322 is to propagate_vals_across_pass_through.
4323 (propagate_constants_across_call): Pass PARM_TYPE to
4324 propagate_scalar_across_jump_function.
4325 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4326 ipa_value_from_jfunc.
4327 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4328 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4329 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4330 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4331 target_type, pass it to ipa_value_from_jfunc.
4332 (update_indirect_edges_after_inlining): Pass parameter type to
4333 try_make_edge_direct_simple_call.
4335 2017-11-28 Jeff Law <law@redhat.com>
4337 * gimple-ssa-evrp-analyze.c
4338 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4339 refine ranges if scev_initialized_p returns true.
4340 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4342 2017-11-28 Julia Koval <julia.koval@intel.com>
4344 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4345 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4346 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4347 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4348 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4349 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4350 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4351 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4352 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4353 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4354 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4355 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4356 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4357 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4358 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4359 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4360 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4361 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4362 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4363 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4364 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4365 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4367 2017-11-28 Julia Koval <julia.koval@intel.com>
4369 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4370 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4371 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4372 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4373 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4374 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4375 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4376 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4377 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4378 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4379 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4380 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4381 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4382 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4383 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4384 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4385 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4386 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4387 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4388 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4389 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4390 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4391 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4392 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4393 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4394 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4395 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4396 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4397 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4398 __builtin_ia32_vpshld_v2di_mask): New builtins.
4399 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4400 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4402 2017-11-28 Richard Biener <rguenther@suse.de>
4404 PR tree-optimization/80776
4405 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4407 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4409 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4410 If the incoming edge is an effective fallthru because the other
4411 edge only reaches a __builtin_unreachable () then record ranges
4412 derived from the controlling condition in SSA info.
4413 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4414 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4416 2017-11-28 Olivier Hainque <hainque@adacore.com>
4418 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4419 on mingw build hosts.
4421 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4424 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4426 2017-11-28 Richard Biener <rguenther@suse.de>
4429 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4430 copies generated from memcpy use a character array as reference
4433 2017-11-28 Julia Koval <julia.koval@intel.com>
4434 Sebastian Peryt <sebastian.peryt@intel.com>
4436 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
4437 c-family/c-cilkplus.o, c-family/array-notation-common.o,
4438 cilk-common.o, cilk.h, cilk-common.c): Remove.
4440 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
4441 * builtins.c (is_builtin_name): Remove cilkplus condition.
4442 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
4443 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
4444 cilk-builtins.def, cilkplus.def): Remove.
4445 * cif-code.def (CILK_SPAWN): Remove.
4446 * cilk-builtins.def: Delete.
4447 * cilk-common.c: Ditto.
4449 * cilkplus.def: Ditto.
4450 * config/darwin.h (fcilkplus): Delete.
4451 * cppbuiltin.c: Ditto.
4452 * doc/extend.texi: Remove cilkplus doc.
4453 * doc/generic.texi: Ditto.
4454 * doc/invoke.texi: Ditto.
4455 * doc/passes.texi: Ditto.
4456 * gcc.c (fcilkplus): Remove.
4457 * gengtype.c (cilk.h): Remove.
4458 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
4460 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
4462 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
4464 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
4465 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
4466 cilkplus conditions.
4467 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
4468 inline_read_section): Ditto.
4469 * ipa-inline-analysis.c (cilk.h): Remove.
4470 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
4471 * lto-wrapper.c (merge_and_complain, append_compiler_options,
4472 append_linker_options): Remove condition for fcilkplus.
4473 * lto/lto-lang.c (cilk.h): Remove.
4474 (lto_init): Remove condition for fcilkplus.
4475 * omp-expand.c (expand_cilk_for_call): Delete.
4476 (expand_omp_taskreg, expand_omp_for_static_chunk,
4477 expand_omp_for): Remove cilkplus
4479 (expand_cilk_for): Delete.
4480 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
4481 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
4482 execute_lower_omp, diagnose_sb_0): Ditto.
4483 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
4484 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4485 * tree-nested.c: Ditto.
4486 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
4487 (dump_generic_node): Ditto.
4488 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
4489 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
4491 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
4493 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4495 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
4496 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
4497 for vector mode and !TARGET_SIMD.
4499 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4501 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
4503 * tree.h (SWITCH_LABELS): Remove.
4504 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
4505 assert SWITCH_BODY is non-NULL.
4506 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
4508 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
4510 PR tree-optimization/80788
4511 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
4512 has TREE_OVERFLOW set, call drop_tree_overflow.
4514 2017-11-28 Richard Biener <rguenther@suse.de>
4516 PR tree-optimization/83158
4517 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
4519 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
4522 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
4523 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
4525 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
4527 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
4528 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
4530 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
4533 * varasm.c (compare_constant): Take the mode of the constants into
4534 account when comparing floating point constants.
4536 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
4538 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
4539 from explicit instantiation of debug_helper.
4540 * vec.h (DEFINE_DEBUG_VEC): Ditto.
4542 2017-11-27 Richard Biener <rguenther@suse.de>
4544 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
4547 2017-11-27 Richard Biener <rguenther@suse.de>
4549 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
4550 (type_hash_canon): Also clear int_cst_hash_table entry for
4552 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
4554 2017-11-27 Tamar Christina <tamar.christina@arm.com>
4556 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4557 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4558 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
4560 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4562 * hash-map.h (gt_cleare_cache): Avoid UB.
4564 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
4566 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
4567 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
4568 (decide_unroll_runtime_iterations): Likewise.
4569 (decide_unroll_stupid): Likewise.
4571 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4574 * config/i386/i386.md: Add a loop with incssp.
4576 2017-11-27 Martin Jambor <mjambor@suse.cz>
4578 PR tree-optimization/81248
4579 * tree-sra.c (splice_param_accesses): Remove size check.
4580 (decide_one_param_reduction): Fix size check.
4581 * gimple-pretty-print.c (dump_profile): Silence warning.
4582 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
4584 2017-11-27 Jakub Jelinek <jakub@redhat.com>
4587 * dbxout.c (lastlineno): New variable.
4588 (dbx_debug_hooks): Use dbxout_switch_text_section as
4589 switch_text_section debug hook.
4590 (dbxout_function_end): Switch to current_function_section
4591 rather than function_section. If crtl->has_bb_partition,
4592 output just one N_FUN, depending on in_cold_section_p.
4593 (dbxout_source_line): Remember last lineno in lastlineno.
4594 (dbxout_switch_text_section): New function.
4595 (dbxout_function_decl): Adjust dbxout_block caller.
4596 (dbx_block_with_cold_children): New function.
4597 (dbxout_block): Return true if any LBRAC/RBRAC have been
4598 emitted. Use dbx_block_with_cold_children at depth == 0
4599 in second partition. Add PARENT_BLOCKNUM argument, pass
4600 it optionally adjusted to children. Output LBRAC/RBRAC
4601 around recursive call only if the block is in the current
4602 partition, if not and anything was output, emit empty
4604 * final.c (final_scan_insn): Compute cold_function_name
4605 before calling switch_text_section debug hook. Call
4606 that hook even if dwarf2out_do_frame if not emitting
4610 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
4611 TREE_READONLY decls.
4613 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
4615 PR rtl-optimization/82488
4616 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
4618 2017-11-26 Julia Koval <julia.koval@intel.com>
4620 * config/i386/i386.c (processor_target_table): Add skylake_cost for
4622 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
4625 2017-11-26 Julia Koval <julia.koval@intel.com>
4627 * config/i386/driver-i386.c (host_detect_local_cpu):
4628 Detect skylake-avx512.
4630 2017-11-26 Julia Koval <julia.koval@intel.com>
4632 * config.gcc: Add -march=cannonlake.
4633 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
4634 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
4635 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
4636 (PTA_CANNONLAKE): New.
4637 (processor_target_table): Add cannonlake.
4638 (ix86_option_override_internal): Ditto.
4639 (fold_builtin_cpu): Ditto.
4640 (get_builtin_code_for_version): Handle cannonlake.
4641 (M_INTEL_COREI7_CANNONLAKE): New.
4642 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
4643 * doc/invoke.texi: Add -march=cannonlake.
4645 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
4647 * plugin.c (add_new_plugin): Use platform-specific library extensions.
4648 (try_init_one_plugin): Alternative implementation for MinGW.
4649 * Makefile.in (plugin_implib): New.
4650 (gengtype-lex.c): Fix broken AIX workaround.
4651 * configure: Regenerate.
4652 * doc/plugins.texi: Document support for MinGW.
4654 2017-11-25 Jakub Jelinek <jakub@redhat.com>
4656 PR rtl-optimization/81553
4657 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
4658 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
4659 is a shift where C1 has different mode than the whole shift, use C1's
4660 mode for MULT rather than the shift's mode.
4663 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
4664 builtins not enabled in the currently selected ISA.
4666 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
4668 PR tree-optimization/71026
4669 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
4670 (insert_reciprocals): Change to insert reciprocals before a division
4671 by a square and to insert the square of a reciprocal.
4672 (execute_cse_reciprocals_1): Change to consider division by a square.
4673 (register_division_in): Add importance parameter.
4675 2017-11-24 Richard Biener <rguenther@suse.de>
4677 PR tree-optimization/82402
4678 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
4679 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
4681 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
4683 * match.pd (0-ptr): New transformation.
4685 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
4688 * ipa-inline.c (inline_small_functions): Set current badnes correctly
4689 when skipping checking.
4691 2017-11-24 Richard Biener <rguenther@suse.de>
4693 PR tree-optimization/83128
4694 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
4695 (vn_reference_lookup_3): Likewise.
4697 2017-11-24 Jakub Jelinek <jakub@redhat.com>
4700 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
4701 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
4704 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
4705 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
4706 check for the latter with gimple_call_builtin_p. Do not handle
4707 BUILT_IN_STPNCPY_CHK which is not a pass through call.
4709 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
4711 * config/arm/arm_neon.h: Fix pragma GCC push_options before
4714 2017-11-23 Julia Koval <julia.koval@intel.com>
4716 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
4717 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
4718 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
4719 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
4720 _mm512_maskz_expandloadu_epi16): New intrinsics.
4721 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
4722 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
4723 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
4724 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
4725 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
4726 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
4727 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
4728 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
4729 _mm256_maskz_expandloadu_epi8): New intrinsics.
4730 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
4731 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
4732 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
4733 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
4734 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
4735 * config/i386/sse.md (VI248_VLBW): New iterator.
4736 (expand<mode>_mask, expand<mode>_maskz): New patterns.
4738 2017-11-23 Julia Koval <julia.koval@intel.com>
4740 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
4741 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
4742 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
4743 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
4744 _mm512_mask_compressstoreu_epi16): New.
4745 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
4746 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
4747 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
4748 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
4749 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
4750 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
4751 _mm256_mask_compressstoreu_epi8): New.
4752 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
4753 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
4754 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
4755 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
4756 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
4757 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
4758 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
4759 __builtin_ia32_compresshi128_mask,
4760 __builtin_ia32_compressstoreuqi512_mask,
4761 __builtin_ia32_compressstoreuhi512_mask,
4762 __builtin_ia32_compressstoreuqi256_mask,
4763 __builtin_ia32_compressstoreuqi128_mask,
4764 __builtin_ia32_compressstoreuhi256_mask,
4765 __builtin_ia32_compressstoreuhi128_mask): New builtins.
4766 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
4768 (ix86_expand_special_args_builtin): Handle new types.
4769 (s4fma_expand): Handle new builtin array.
4770 * config/i386/immintrin.h: Include new headers.
4771 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4772 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4774 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4776 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4777 fixes. Declare temp and g variables at the top in order to avoid
4778 {} in most of the cases.
4780 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4782 * match.pd (ptr-0): New transformation.
4784 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4786 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4787 (struct addr_mode_cost_table): New.
4788 (struct tune_params): Add field addr_mode_costs.
4789 * config/arm/arm.c (generic_addr_mode_costs): New.
4790 (arm_slowmul_tune): Initialise addr_mode_costs field.
4791 (arm_fastmul_tune): Likewise.
4792 (arm_strongarm_tune): Likewise.
4793 (arm_xscale_tune): Likewise.
4794 (arm_9e_tune): Likewise.
4795 (arm_marvell_pj4_tune): Likewise.
4796 (arm_v6t2_tune): Likewise.
4797 (arm_cortex_tune): Likewise.
4798 (arm_cortex_a8_tune): Likewise.
4799 (arm_cortex_a7_tune): Likewise.
4800 (arm_cortex_a15_tune): Likewise.
4801 (arm_cortex_a35_tune): Likewise.
4802 (arm_cortex_a53_tune): Likewise.
4803 (arm_cortex_a57_tune): Likewise.
4804 (arm_exynosm1_tune): Likewise.
4805 (arm_xgene1_tune): Likewise.
4806 (arm_cortex_a5_tune): Likewise.
4807 (arm_cortex_a9_tune): Likewise.
4808 (arm_cortex_a12_tune): Likewise.
4809 (arm_cortex_a73_tune): Likewise.
4810 (arm_v7m_tune): Likewise.
4811 (arm_cortex_m7_tune): Likewise.
4812 (arm_v6m_tune): Likewise.
4813 (arm_fa726te_tune): Likewise.
4814 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4816 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4818 * config/arm/arm.c (arm_mem_costs): New function.
4819 (arm_rtx_costs_internal): Use arm_mem_costs.
4821 2017-11-23 Mark Wielaard <mark@klomp.org>
4823 * dwarf2out.c (init_sections_and_labels): Use generation to create
4824 unique ranges_section_label and ranges_base_label. Return generation.
4825 (output_rnglists): Add generation argument. Use generation to create
4826 unique ranges labels.
4827 (dwarf2out_finish): Get generation from init_sections_and_labels
4828 and pass generation to output_rnglists.
4830 2017-11-23 Mike Stump <mikestump@comcast.net>
4831 Eric Botcazou <ebotcazou@adacore.com>
4833 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4834 * cfgloop.h (struct loop): Add unroll field.
4835 * function.h (struct function): Add has_unroll bitfield.
4836 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4837 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4838 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4839 (pass_rtl_unroll_loops::gate): Likewise.
4840 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4841 for which loop->unroll==1.
4842 (decide_unroll_constant_iterations): Use note for consistency and
4843 take loop->unroll into account. Return early if loop->unroll is set.
4844 Fix thinko in existing test.
4845 (decide_unroll_runtime_iterations): Use note for consistency and
4846 take loop->unroll into account.
4847 (decide_unroll_stupid): Likewise.
4848 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4849 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4850 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4852 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4853 (print_loop): Print loop->unroll if set.
4854 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4855 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4856 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4858 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4859 loop->unroll is set and smaller than the trip count. Otherwise bypass
4860 entirely the heuristics if loop->unroll is set. Remove dead note.
4861 Fix off-by-one bug in other note.
4862 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4863 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4865 (tree_unroll_loops_completely): Make static.
4866 (pass_complete_unroll::execute): Use correct type for variable.
4867 (pass_complete_unrolli::execute): Fix formatting.
4868 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4870 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4872 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4873 enable when TARGET_PREFER_AVX128 is set.
4875 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4877 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4879 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4880 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4882 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4884 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4885 frequencies. Fix estimation of aggregate parameters.
4887 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4889 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4892 2017-11-23 Tom de Vries <tom@codesourcery.com>
4894 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4896 2017-11-23 Tom de Vries <tom@codesourcery.com>
4898 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4901 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4904 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4905 sibcall_value_pcrel_fdpic): Use local variable instead of
4907 (calli_tbr_rel): Add missing operand 2.
4908 (call_valuei_tbr_rel): Add missing operand 3.
4910 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4913 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4914 bitpos/bitsize covering the whole destination, use store_expr only if
4915 the complex mode is the same. Otherwise, use expand_normal and if
4916 it returns CONCAT, subreg each part separately instead of trying to
4917 subreg the whole result.
4919 2017-11-23 Richard Biener <rguenther@suse.de>
4921 PR tree-optimization/23094
4922 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4923 come here from walking over backedges in the first iteration.
4924 (vn_reference_lookup_3): Skip clobbers that store the same value.
4926 2017-11-23 Richard Biener <rguenther@suse.de>
4928 PR tree-optimization/81403
4929 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4930 a block we need a leader relative to.
4931 (phi_translate_1): For nary processing require a leader from
4932 get_representative_for given we run expression simplification
4933 using match-and-simplify. Remove previous fix.
4935 2017-11-22 Jeff Law <law@redhat.com>
4937 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4938 Use new method allocate_value_range rather than accessing the
4939 vrp_value_range_pool data member directly.
4940 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4941 to use extract_range_from_stmt method to avoid need for
4942 extract_range_from_assignment method.
4943 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4944 method rather than setting values_propgated data member directly.
4945 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4946 and values propagated data members and extract_range_from_assignment
4947 method. Reorder private data members to conform to standards.
4948 Add new methods set_lattice_propagation_complete and
4949 allocate_value_range.
4951 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4953 PR rtl-optimization/83030
4954 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4955 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4956 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4957 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4958 CROSSING_JUMP_P flag is already set before setting it.
4959 * cfgrtl.c (fixup_partition_crossing): Likewise.
4960 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4963 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4965 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4966 Handle the case where both arguments are using gen_const_vec_series.
4968 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4971 * pretty-print.c (pp_format): Move quoting implementation to
4972 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4973 to pass address of "quote" local.
4974 (pp_begin_quote): New function.
4975 (pp_end_quote): New function.
4976 * pretty-print.h (printer_fn): Convert penultimate param from bool
4978 (pp_begin_quote): New decl.
4979 (pp_end_quote): New decl.
4980 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4981 param from bool to bool *.
4982 * tree-diagnostic.h (default_tree_printer): Likewise.
4984 2017-11-22 Jeff Law <law@redhat.com>
4986 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4987 Initialize vr_values.
4988 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4989 vr_values via vr_values class instance rather than delegators.
4990 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4991 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4992 (evrp_range_analyzer::push_value_range): Likewise.
4993 (evrp_range_analyzer::pop_value_range): Likewise.
4994 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4995 most delegators. Those remaining are exposed as public interfaces.
4996 Make vr_values a pointer and private.
4997 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4999 (evrp_range_analyzer::get_vr_value): New method.
5000 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5001 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5002 (class evrp_dom_walker): Attach evrp_folder class, initialize
5003 it in the ctor. Remove temporary delegators.
5004 (evrp_dom_walker::before_dom_children): Call methods in attached
5005 evrp_range_analyzer class via class instance pointer. Use
5006 free value_range_constant_singleton to remove need for
5007 op_with_constant_singleton_value delegator method. Do not
5008 create a vrp_prop class instance for every call! Narrow
5009 scope of a couple variables.
5010 (evrp_dom_walker::cleanup): Call methods in attached
5011 evrp_range_analyzer class via class instance pointer.
5012 * vr-values.h (class vr_values): Privatize many methods and
5015 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5017 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5018 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5019 * expr.c (expand_expr_real_2): Likewise.
5020 * fold-const.c (const_unop): Likewise.
5021 * optabs-tree.c (optab_for_tree_code): Likewise.
5022 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5023 * tree-inline.c (estimate_operator_cost): Likewise.
5024 * tree-pretty-print.c (dump_generic_node): Likewise.
5025 (op_code_prio): Likewise.
5026 (op_symbol_code): Likewise.
5027 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5028 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5029 * internal-fn.c (direct_internal_fn_optab): New function.
5030 (direct_internal_fn_array, direct_internal_fn_supported_p
5031 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5032 * fold-const-call.c (fold_const_reduction): New function.
5033 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5035 * tree-vect-loop.c: Include internal-fn.h.
5036 (reduction_code_for_scalar_code): Rename to...
5037 (reduction_fn_for_scalar_code): ...this and return an internal
5039 (vect_model_reduction_cost): Take an internal_fn rather than
5041 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5043 (vectorizable_reduction): Use internal functions rather than tree
5044 codes for the reduction operation. Update calls to the functions
5046 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5047 Use calls to internal functions rather than REDUC tree codes.
5048 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5050 2017-11-22 Olivier Hainque <hainque@adacore.com>
5052 * config/vxworks.c (vxworks_override_options): Pick default
5053 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5054 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5055 DWARF_GNAT_ENCODINGS_DEFAULT.
5056 * config/vxworksae.h: Likewise.
5058 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5060 PR tree-optimization/83104
5061 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5064 2017-11-22 Marek Polacek <polacek@redhat.com>
5065 H.J. Lu <hongjiu.lu@intel.com>
5066 Jason Merrill <jason@redhat.com>
5071 * calls.c (initialize_argument_information): Call
5072 warn_parameter_passing_abi target hook.
5073 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5074 argument onto stack.
5075 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5076 * common.opt: Update -fabi-version description.
5077 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5078 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5080 (ix86_is_empty_record): New function.
5081 (ix86_warn_parameter_passing_abi): New function.
5082 (TARGET_EMPTY_RECORD_P): Redefine.
5083 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5084 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5085 * doc/tm.texi: Regenerated.
5086 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5087 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5088 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5089 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5090 instead of int_size_in_bytes.
5091 * expr.c (copy_blkmode_to_reg): Likewise.
5092 * function.c (aggregate_value_p): Return 0 for empty types.
5093 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5095 (locate_and_pad_parm): Call arg size_in_bytes instead
5097 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5098 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5099 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5101 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5102 (std_gimplify_va_arg_expr): Skip empty records. Call
5103 arg_size_in_bytes instead size_in_bytes.
5104 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5105 * tree-core.h (tree_type_common): Add empty_flag.
5106 (tree_decl_common): Update comments.
5107 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5109 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5110 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5112 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5113 * tree.c (default_is_empty_type): New function.
5114 (default_is_empty_record): New function.
5115 (arg_int_size_in_bytes): New function.
5116 (arg_size_in_bytes): New function.
5117 (get_ultimate_context): New function.
5118 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5119 TRANSLATION_UNIT_WARN_EMPTY_P.
5120 (default_is_empty_record, arg_int_size_in_bytes,
5121 arg_size_in_bytes, get_ultimate_context): Declare.
5123 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5125 * config/arm/arm.c (cmse_clear_registers): New function.
5126 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5127 code by call to cmse_clear_registers.
5128 (cmse_nonsecure_entry_clear_before_return): Likewise.
5130 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5132 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5133 (vdot_s32, vdotq_s32): New.
5134 (vdot_lane_u32, vdotq_lane_u32): New.
5135 (vdot_lane_s32, vdotq_lane_s32): New.
5138 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5141 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5142 for unsigned values with fewer HWIs than the precision.
5143 (test_overflow): New function.
5144 (wide_int_cc_tests): Call it.
5146 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5148 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5149 has the same mode class as Pmode.
5151 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5153 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5154 padding_bits_to_clear_ptr.
5155 (cmse_nonsecure_entry_clear_before_return): Likewise.
5157 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5159 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5160 auto_sbitap instead of integer bitfield to control register needing
5163 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5165 PR tree-optimization/83044
5166 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5167 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5168 that need the upper bound. Subtract offset from
5169 get_addr_base_and_unit_offset only if positive and subtract it
5170 before division by eltsize rather than after it.
5173 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5174 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5175 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5179 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5180 expand_binop with *mul_widen_optab, make sure at least one of the
5181 operands doesn't have VOIDmode.
5184 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5186 PR rtl-optimization/82044
5187 PR tree-optimization/82042
5188 * dse.c (record_store): Check for overflow.
5189 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5190 add_wild_read instead of clear_rhs_from_active_local_stores on
5191 overflow and log it into dump_file.
5193 2017-11-22 Richard Biener <rguenther@suse.de>
5195 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5196 fake edges to exit when looking for a place to insert.
5197 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5199 (insert_into_preds_of_block): Commit edge insertion immediately,
5200 assert that doesn't require new BBs.
5201 (fini_pre): Release expressions.
5202 (pass_pre::execute): Shuffle things around a bit, if the fn
5203 is too large do not compute AVAIL either as this is really the
5206 2017-11-22 Richard Biener <rguenther@suse.de>
5208 PR tree-optimization/83089
5209 * tree-if-conv.c (pass_if_conversion::execute): If anything
5210 changed reset SCEV and free the number of iteration estimates.
5212 2017-11-21 Martin Sebor <msebor@redhat.com>
5214 PR tree-optimization/82945
5215 * calls.h (warn_nonstring_bound): Remove unused function.
5217 2017-11-21 Martin Sebor <msebor@redhat.com>
5219 PR tree-optimization/82945
5220 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5221 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5222 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5224 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5225 * calls.h (get_attr_nonstring_decl): Declare new function.
5226 * doc/extend.texi (attribute nonstring): Update.
5227 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5228 get_attr_nonstring_decl and handle it.
5229 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5230 detection of nul-termination.
5231 (strlen_to_stridx): Change to a pointer.
5232 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5233 (pass_strlen::execute): Same.
5235 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5237 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5238 for the new option -mprefer-vector-width=[none|128|256|512].
5239 * config/i386/i386.c (ix86_target_string): remove old style options
5240 -mprefer-avx256 and make -mprefer-avx128 as alias.
5241 (ix86_option_override_internal): Apply defaults for the
5242 -mprefer-vector-width=[128|256] option.
5243 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5244 Implement macros to work with -mprefer-vector-width=.
5245 * config/i386/i386.opt: Implemented option
5246 -mprefer-vector-width=[none|128|256|512].
5247 * doc/invoke.texi: Documentation for
5248 -mprefer-vector-width=[none|128|256|512].
5250 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5252 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5254 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5256 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5258 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5260 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5262 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5263 * expr.c (expand_expr_real_2): Likewise.
5264 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5265 fold_binary_loc): Likewise.
5266 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5267 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5268 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5269 MINUS_EXPR transformations.
5270 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5271 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5272 * tree-inline.c (estimate_operator_cost): Likewise.
5273 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5274 op_symbol_code): Likewise.
5275 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5276 * vr-values.c (extract_range_from_binary_expr): Likewise.
5277 * varasm.c (initializer_constant_valid_p_1): Likewise.
5278 * tree.def: New tree code POINTER_DIFF_EXPR.
5280 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5282 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5283 integer suffix to movbe mnemonic.
5284 (*bswaphi2_movbe): Ditto.
5285 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5287 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5290 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5291 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5292 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5294 2017-11-21 Martin Liska <mliska@suse.cz>
5296 * tree-inline.c (expand_call_inline): Remove not needed
5299 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5300 Jakub Jelinek <jakub@redhat.com>
5303 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5304 Remove static keyword from f variable.
5306 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5308 PR tree-optimization/83086
5309 * gimple-ssa-store-merging.c
5310 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5311 rather than n.base_addr.
5313 2017-11-21 Martin Liska <mliska@suse.cz>
5315 PR rtl-optimization/82044
5316 PR tree-optimization/82042
5317 * dse.c (check_mem_read_rtx): Check for overflow.
5319 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5321 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5324 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5326 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5327 and its corresponding call arg location note.
5329 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5330 Andrew Burgess <andrew.burgess@embecosm.com>
5332 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5334 (arc_return_slot_offset): Likewise.
5335 (arc_eh_return_address_location): New declaration.
5336 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5337 (MUST_SAVE_REGISTER): Add exception handler case.
5338 (MUST_SAVE_RETURN_ADDR): Likewise.
5339 (arc_frame_pointer_required): Likewise.
5340 (arc_frame_pointer_needed): New function.
5341 (arc_compute_frame_size): Changed.
5342 (arc_expand_prologue): Likewise.
5343 (arc_expand_epilogue): Likewise.
5344 (arc_initial_elimination_offset): Likewise.
5345 (arc_return_slot_offset): Delete.
5346 (arc_eh_return_address_location): New function.
5347 (arc_builtin_setjmp_frame_value): Likewise.
5348 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5349 (EH_RETURN_STACKADJ_RTX): Define.
5350 (EH_RETURN_HANDLER_RTX): Likewise.
5351 * config/arc/arc.md (eh_return): Delete.
5353 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5355 * print-tree.h (debug_vec_tree): Remove prototype.
5356 * gdbinit.in (pvt): Remove macro.
5358 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5360 PR tree-optimization/83047
5361 * gimple-ssa-store-merging.c
5362 (imm_store_chain_info::output_merged_store): If the loads with the
5363 same vuse are in different basic blocks, for load_gsi pick a load
5364 location that is dominated by the other loads.
5367 * config/i386/i386.c (ix86_memmodel_check): Start
5368 -Winvalid-memory-model diagnostics with lowercase letter.
5371 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5372 set in_cold_section_p to the partition containing loc_list->first.
5373 When seeing loc_list->last_before_switch node, update secname and
5374 perform range_across_switch second partition handling only after that.
5377 * run-rtl-passes.c: Include debug.h.
5378 (run_rtl_passes): Call debug_hooks->assembly_start.
5379 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5383 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5384 OPTAB_DIRECT in calls to expand_simple_binop.
5386 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5389 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5391 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5394 * spellcheck.h (best_match::blithely_get_best_candidate): New
5397 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5399 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5400 with lower case letter.
5402 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5404 * config/i386/i386.md (bswaphi2): New expander.
5405 (*bswaphi2_movbe): New insn pattern.
5406 (bswaphi -> rorhi pepehole2): New peephole pattern.
5408 2017-11-20 Jeff Law <law@redhat.com>
5410 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5411 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5412 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5413 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5414 gimple-ssa-evrp-analyze.h.
5416 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5417 set BB_VISITED here.
5418 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5420 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5422 * tree-predcom.c: Add general comment on Store-Store chains.
5423 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5425 (get_chain_last_ref_at): Rename into...
5426 (get_chain_last_write_at): ...this.
5427 (get_chain_last_write_before_load): New function.
5428 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5429 CT_STORE_STORE when write reference is added.
5430 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5431 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5432 (initialize_root_vars_store_elim_1): Ditto.
5433 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
5434 definition is created.
5435 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
5436 chain by replacing it with dominant stored value.
5438 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5440 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
5442 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
5444 * vr-values.c (extract_range_from_binary_expr): Use a full range
5447 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
5449 * config/arm/arm.md (R4_REGNUM): Define constant.
5450 (nonsecure_call_internal): Remove r4 clobber.
5451 (nonsecure_call_value_internal): Likewise.
5452 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
5453 clobber and resequence match_operands.
5454 (nonsecure_call_value_reg_thumb1_v5): Likewise.
5455 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
5456 (nonsecure_call_value_reg_thumb2): Likewise.
5458 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5460 PR tree-optimization/78821
5461 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
5462 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
5463 to the address of the base rather than the base itself.
5464 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
5465 (find_bswap_or_nop_finalize): New function.
5466 (find_bswap_or_nop): Use it.
5467 (bswap_replace): Return a tree rather than bool, change first
5468 argument from gimple * to gimple_stmt_iterator, allow inserting
5469 into an empty sequence, allow ins_stmt to be NULL - then emit
5470 all stmts into gsi. Fix up MEM_REF address gimplification.
5471 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
5472 (struct store_immediate_info): Add N and INS_STMT non-static
5474 (store_immediate_info::store_immediate_info): Initialize them
5475 from newly added ctor args.
5476 (merged_store_group::apply_stores): Formatting fixes. Sort by
5478 (stmts_may_clobber_ref_p): For stores call also
5479 refs_anti_dependent_p.
5480 (gather_bswap_load_refs): New function.
5481 (imm_store_chain_info::try_coalesce_bswap): New method.
5482 (imm_store_chain_info::coalesce_immediate_stores): Use it.
5483 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
5484 (imm_store_chain_info::output_merged_store): Fail if number of
5485 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
5486 and NOP_EXPR rhs_code.
5487 (pass_store_merging::process_store): Compute n and ins_stmt, if
5488 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
5489 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
5491 (pass_store_merging::execute): Calculate dominators.
5493 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
5494 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
5495 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
5496 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5497 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
5498 class pass_optimize_bswap, bswap_replace,
5499 pass_optimize_bswap::execute): Moved to ...
5500 * gimple-ssa-store-merging.c: ... this file.
5501 Include optabs-tree.h.
5502 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
5503 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
5504 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
5505 anonymous namespace, remove static keywords.
5506 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
5507 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
5509 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
5512 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
5514 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
5516 * vec.h (debug_helper): New function.
5517 (DEFINE_DEBUG_VEC): New macro.
5518 * hash-set.h (debug_helper): New function.
5519 (DEFINE_DEBUG_HASH_SET): New macro.
5520 * cfg.c (debug_slim (edge)): New function.
5521 Call DEFINE_DEBUG_VEC for edges.
5522 Call DEFINE_DEBUG_HASH_SET for edges.
5523 * cfghooks.c (debug_slim (basic_block)): New function.
5524 Call DEFINE_DEBUG_VEC for basic blocks.
5525 Call DEFINE_DEBUG_HASH_SET for basic blocks.
5526 * print-tree.c (debug_slim): New function to handle trees.
5527 Call DEFINE_DEBUG_VEC for trees.
5528 Call DEFINE_DEBUG_HASH_SET for trees.
5529 (debug (vec<tree, va_gc>) &): Remove.
5530 (debug (<vec<tree, va_gc>) *): Remove.
5531 * print-rtl.c (debug_slim): New function to handle const_rtx.
5532 Call DEFINE_DEBUG_VEC for rtx_def.
5533 Call DEFINE_DEBUG_VEC for rtx_insn.
5534 Call DEFINE_DEBUG_HASH_SET for rtx_def.
5535 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
5536 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
5537 (debug (vec<rtx_insn *> *ptr): Remove.
5538 (debug_insn_vector): Remove.
5539 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
5541 2017-11-20 Tom de Vries <tom@codesourcery.com>
5543 PR rtl-optimization/82020
5544 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5545 IF_THEN_ELSE condition.
5547 2017-11-19 Jeff Law <law@redhat.com>
5549 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
5550 of degenerates resulting from ignoring an edge.
5552 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5555 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
5557 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5560 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5561 for uninitialied counts.
5563 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5566 * tree-inline.c (estimate_num_insns): Set to 1 at least.
5568 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
5571 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
5573 2017-11-19 Tom de Vries <tom@codesourcery.com>
5575 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
5578 2017-11-19 Tom de Vries <tom@codesourcery.com>
5580 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
5583 2017-11-19 Tom de Vries <tom@codesourcery.com>
5585 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5588 2017-11-19 Tom de Vries <tom@codesourcery.com>
5590 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
5592 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
5594 2017-11-19 Tom de Vries <tom@codesourcery.com>
5596 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
5598 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
5599 semicolon after MCORE_EXPORT_NAME call.
5601 2017-11-19 Tom de Vries <tom@codesourcery.com>
5604 * vmsdbgout.c (vmsdbgout_early_finish): New function.
5605 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
5607 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
5609 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
5610 (cgraph_edge::clone): Cleanup updating of profile.
5611 * ipa-cp.c (update_profiling_info): Likewise.
5612 * ipa-inline-transform.c (inline_transform): Likewise.
5613 * ipa-inline.c (inline_small_functions): Add missing space to dump.
5614 * ipa-split.c (execute_split_functions): Do not split when function
5616 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
5617 * profile-count.c (profile_count::dump): Add global0.
5618 (profile_count::to_cgraph_frequency): Do not ICE when entry is
5620 (profile_count::to_sreal_scale): Likewise.
5621 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
5622 (profile_count::combine_with_ipa_count): New function.
5623 * profile-count.h (profile_guessed_global0adjusted): New.
5624 (profile_count::adjusted_zero): New.
5625 (profile_count::global0adjusted): New.
5626 (profile_count::combine_with_ipa_count): New.
5627 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
5628 correct profile of return block of split functions.
5629 (copy_cfg_body): Remove unused profile_count.
5630 (copy_body): Likewise.
5631 (expand_call_inline): Update.
5632 (tree_function_versioning): Update.
5634 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
5636 * hash-set.h (hash_set::empty): New.
5637 * tree-ssa-threadbackward.h: Delete.
5638 * tree-ssa-threadbackward.c (class thread_jumps): New.
5639 Move max_threaded_paths into class.
5640 (fsm_find_thread_path): Remove arguments that are now in class.
5641 (profitable_jump_thread_path): Rename to...
5642 (thread_jumps::profitable_jump_thread_path): ...this.
5643 (convert_and_register_jump_thread_path): Rename to...
5644 (thread_jumps::convert_and_register_current_path): ...this.
5645 (check_subpath_and_update_thread_path): Rename to...
5646 (thread_jumps::check_subpath_and_update_thread_path): ...this.
5647 (register_jump_thread_path_if_profitable): Rename to...
5648 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
5649 (handle_phi): Rename to...
5650 (thread_jumps::handle_phi): ...this.
5651 (handle_assignment): Rename to...
5652 (thread_jumps::handle_assignment): ...this.
5653 (fsm_find_control_statement_thread_paths): Rename to...
5654 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
5655 (find_jump_threads_backwards): Rename to...
5656 (thread_jumps::find_jump_threads_backwards): ...this.
5657 Initialize path local data.
5658 (pass_thread_jumps::execute): Call find_jump_threads_backwards
5659 from within thread_jumps class.
5660 (pass_early_thread_jumps::execute): Same.
5662 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5664 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
5666 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5668 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
5669 to canonical location.
5671 2017-11-17 Steve Ellcey <sellcey@cavium.com>
5674 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
5676 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
5678 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
5680 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
5681 rcrt1.o%s/grcrt1.o%s for -static-pie.
5683 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5685 * i386.c (ix86_multiplication_cost, ix86_division_cost,
5686 ix86_shift_rotate_cost): Break out from ...
5687 (ix86_rtx_costs): ... here.
5688 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
5691 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5693 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
5694 when entry block was promoted unlikely.
5695 (estimate_bb_frequencies): Increase frequency scale.
5696 * profile-count.h (profile_count): Export precision info.
5698 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5700 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
5701 disturbin profile of entry block.
5703 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5705 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
5708 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
5710 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
5711 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
5713 2017-11-17 Jeff Law <law@redhat.com>
5715 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
5716 from evrp_dom_walker class. Various methods moved into new class.
5717 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
5718 (evrp_range_analyzer::enter): New method.
5719 (evrp_range_analyzer::leave): New method.
5720 (evrp_dom_walker): Remove delegators no longer needed by this class.
5721 Replace vr_values data member with evrp_range_analyzer
5723 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
5724 method extracted from evrp_dom_walker::before_dom_children.
5725 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
5726 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
5728 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
5729 Add private copy constructor and move assignment operators.
5730 Privatize methods and class data where trivially possible.
5731 (evrp_dom_walker::cleanup): New function, extracted from
5732 execute_early_vrp. Simplify access to class data.
5734 * vr-values.h (get_output_for_vrp): Prototype.
5735 * vr-values.c (get_output_for_vrp): New function extracted from
5736 vrp_visit_assignment_or_call and extract_range_from_stmt.
5737 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
5739 2017-11-17 Luis Machado <luis.machado@linaro.org>
5741 * config/aarch64/aarch64.c
5742 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
5743 (qdf24xx_tunings) <autoprefetcher_model>: Set to
5744 tune_params::AUTOPREFETCHER_WEAK.
5746 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5749 * config/arm/arm.c (arm_valid_target_attribute_rec):
5750 Parse "arch=" and "+<ext>".
5751 (arm_valid_target_attribute_tree): Re-init global options.
5752 (arm_option_override): Make non-static.
5753 (arm_options_perform_arch_sanity_checks): Make errors fatal.
5754 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
5755 (__ARM_FEATURE_CRC32): Support undef.
5756 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
5757 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
5759 2017-11-17 David Malcolm <dmalcolm@redhat.com>
5761 * gdbinit.in (break-on-diagnostic): New command.
5763 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5765 * config/i386/i386.c (ix86_expand_epilogue): Change simple
5766 return to indirect jump for EH return if control-flow
5767 protection is enabled. Change explicit 'false' argument in
5768 pro_epilogue_adjust_stack with a value of
5770 * config/i386/i386.md (simple_return_indirect_internal):
5771 Remove SImode restriction to support 64-bit.
5773 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5775 * combine.c (added_notes_insn): New.
5776 (try_combine): Handle added_notes_insn like added_links_insn.
5777 Rewrite return value code.
5778 (distribute_notes): Set added_notes_insn to the earliest insn we added
5781 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5783 PR rtl-optimization/82621
5784 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5785 dest of one of those SETs is unused.
5787 2017-11-17 Richard Biener <rguenther@suse.de>
5790 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5791 * tree-pretty-print.c (dump_generic_node): Handle
5792 annot_expr_parallel_kind.
5793 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5794 * gimplify.c (gimple_boolify): Likewise.
5796 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5798 * config.gcc (extra_headers): Add cet.h for x86 targets.
5799 * config/i386/cet.h: New file.
5800 * doc/install.texi: Add --enable-cet/--disable-cet.
5802 2017-11-17 Richard Biener <rguenther@suse.de>
5804 PR tree-optimization/83017
5805 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5806 (gen_parallel_loop): Properly count iterations.
5807 (parallelize_loops): Handle loop->can_be_parallel independent
5808 of flag_loop_parallelize_all. Make static profitability test match
5810 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5811 * invoke.texi (parloops-min-per-thread): Document.
5813 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5815 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5816 upstreaming review comments.
5818 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5820 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5821 with fast unaligned access.
5822 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5824 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5826 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5827 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5829 2017-11-17 Richard Biener <rguenther@suse.de>
5831 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5832 folding of references.
5834 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5837 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5838 of replacing call to strncmp with corresponding call to strcmp when
5841 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5843 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5844 option prefer-avx256 for skylake-avx512 configuration.
5845 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5846 (get_builtin_code_for_version): Ditto.
5848 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5849 Monk Chiang <sh.chiang04@gmail.com>
5851 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5852 (FIXED_REGISTERS): Reserve more register numbers.
5853 (CALL_USED_REGISTERS): Likewise.
5854 (REG_ALLOC_ORDER): Likewise.
5855 (REG_CLASS_CONTENTS): Likewise.
5856 (REGISTER_NAMES): Likewise.
5858 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5859 Kito Cheng <kito.cheng@gmail.com>
5861 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5863 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5865 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5867 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5868 parameter list for vec_splats.
5870 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5872 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5874 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5876 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5879 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5880 with the required version of automake 1.11.6.
5882 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5884 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5886 (DU_C2_3_power9): Correct reservation combinations.
5887 (FP_DIV_power9, VEC_DIV_power9): New.
5888 (power9-alu): Split out rotate/shift...
5889 (power9-rot): ...to here, correct dispatch resource.
5890 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5892 (power9-fp): Correct latency.
5893 (power9-sdiv): Add div/sqrt resource.
5894 (power9-ddiv): Correct latency, add div/sqrt resource.
5895 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5896 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5898 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5900 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5902 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5903 switch statement mapping KF built-ins to TF built-ins if we don't
5904 have the proper ISA 3.0 assembler support.
5906 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5908 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5909 (lower_emutls_function_body): Do not compute it.
5911 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5913 * ipa-split.c (split_bb_info): Turn time to sreal.
5914 (split_point): Likewise.
5915 (dump_split_point): Likewise.
5916 (fine_split_points): Likewise.
5917 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5919 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5921 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5924 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5926 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5928 (expensive_function_p): Remove useless assert.
5929 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5931 2017-11-16 Martin Sebor <msebor@redhat.com>
5933 PR tree-optimization/82588
5934 PR tree-optimization/82583
5935 * tree-vrp.c (check_array_ref): Handle flexible array members,
5936 string literals, and inner indices.
5937 (search_for_addr_array): Add detail to diagnostics.
5939 2017-11-16 Nathan Sidwell <nathan@acm.org>
5943 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5944 (SET_DECL_ASSEMBLER_NAME): Forward to
5945 overwrite_decl_assembler_name.
5946 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5947 (overwrite_decl_assembler_name): Declare.
5948 * tree.c (overwrite_decl_assembler_name): New.
5949 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5950 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5951 (LANG_HOOKS_INITIALIZER): Add it.
5952 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5953 * langhooks.c (lhd_set_decl_assembler_name): Use
5954 SET_DECL_ASSEMBLER_NAME.
5955 (lhd_overwrite_decl_assembler_name): Default implementation.
5957 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5958 Jackson Woodruff <jackson.woodruff@arm.com>
5960 PR tree-optimization/71026
5961 * match.pd: Canonicalize constant multiplies in division.
5963 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5965 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5966 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5967 Remove OPT_fomit_frame_pointer entry.
5968 * common/config/arc/arc-common.c: Likewise.
5969 * common/config/arm/arm-common.c: Likewise.
5970 * common/config/avr/avr-common.c: Likewise.
5971 * common/config/c6x/c6x-common.c: Likewise.
5972 * common/config/cr16/cr16-common.c: Likewise.
5973 * common/config/cris/cris-common.c: Likewise.
5974 * common/config/epiphany/epiphany-common.c: Likewise.
5975 * common/config/fr30/fr30-common.c: Likewise.
5976 * common/config/frv/frv-common.c: Likewise.
5977 * common/config/ia64/ia64-common.c: Likewise.
5978 * common/config/iq2000/iq2000-common.c: Likewise.
5979 * common/config/lm32/lm32-common.c: Likewise.
5980 * common/config/m32r/m32r-common.c: Likewise.
5981 * common/config/mcore/mcore-common.c: Likewise.
5982 * common/config/microblaze/microblaze-common.c: Likewise.
5983 * common/config/mips/mips-common.c: Likewise.
5984 * common/config/mmix/mmix-common.c: Likewise.
5985 * common/config/mn10300/mn10300-common.c: Likewise.
5986 * common/config/nios2/nios2-common.c: Likewise.
5987 * common/config/pa/pa-common.c: Likewise.
5988 * common/config/pdp11/pdp11-common.c: Likewise.
5989 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5990 * common/config/riscv/riscv-common.c: Likewise.
5991 * common/config/rs6000/rs6000-common.c: Likewise.
5992 * common/config/rx/rx-common.c: Likewise.
5993 * common/config/s390/s390-common.c: Likewise.
5994 * common/config/sh/sh-common.c: Likewise.
5995 * common/config/sparc/sparc-common.c: Likewise.
5996 * common/config/tilegx/tilegx-common.c: Likewise.
5997 * common/config/tilepro/tilepro-common.c: Likewise.
5998 * common/config/v850/v850-common.c: Likewise.
5999 * common/config/visium/visium-common.c: Likewise.
6000 * common/config/xstormy16/xstormy16-common.c: Likewise.
6001 * common/config/xtensa/xtensa-common.c: Likewise.
6002 * invoke.texi (-fomit-frame-pointer): Update documentation.
6004 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6006 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6008 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6010 * cfg.c (scale_bbs_frequencies_int,
6011 cale_bbs_frequencies_gcov_type): Remove.
6012 * cfg.h (scale_bbs_frequencies_int,
6013 cale_bbs_frequencies_gcov_type): Remove.
6015 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6017 * tree-ssa-loop-manip.c
6018 (scale_dominated_blocks_in_loop): Update to profile counts.
6019 (tree_transform_and_unroll_loop): Likewise.
6021 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6023 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6024 scale_bbs_frequencies_int.
6026 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6028 * final.c (compute_alignments): Use counts rather than frequencies.
6030 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6032 * cfgloopanal.c: Include sreal.h
6033 (average_num_loop_insns): Use counts and sreal for accounting.
6035 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6037 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6040 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6041 Kito Cheng <kito.cheng@gmail.com>
6043 * config/nds32/constraints.md: Provide more constraints.
6044 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6045 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6046 support constraints usage.
6048 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6050 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6052 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6053 Kito Cheng <kito.cheng@gmail.com>
6055 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6056 * config/nds32/nds32.opt: Refine the layout.
6057 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6058 TARGET_EXT_STRING): Support new options.
6059 * config/nds32/nds32.h: Likewise.
6060 * config/nds32/nds32.md: Likewise.
6061 * config/nds32/nds32-predicates.c: Likewise.
6062 * config/nds32/constraints.md: Likewise.
6063 * common/config/nds32/nds32-common.c: Likewise.
6065 2017-11-16 Julia Koval <julia.koval@intel.com>
6068 * config/i386/gfniintrin.h: Add sse check.
6069 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6071 2017-11-16 Julia Koval <julia.koval@intel.com>
6073 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6074 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6075 (ix86_handle_option): Handle -mavx512vbmi2.
6076 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6077 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6078 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6079 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6080 (ix86_valid_target_attribute_inner_p): Ditto.
6081 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6082 * config/i386/i386.opt (mavx512vbmi2): New option.
6083 * doc/invoke.texi: Add new option.
6085 2017-11-16 Julia Koval <julia.koval@intel.com>
6087 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6088 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6089 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6090 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6091 _mm512_gf2p8mul_epi8): New intrinsics.
6092 * config/i386/i386-builtin-types.def
6093 (V64QI_FTYPE_V64QI_V64QI): New type.
6094 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6095 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6096 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6097 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6098 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6099 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6101 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6103 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6104 explicitly as a stream of bytes.
6106 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6108 * config/rs6000/altivec.h (vec_xst_be): New #define.
6109 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6110 and externalize from *altivec_vperm_<mode>_internal.
6111 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6113 (XL_BE_V8HI): Likewise.
6114 (XL_BE_V4SI): Likewise.
6115 (XL_BE_V4SI): Likewise.
6116 (XL_BE_V2DI): Likewise.
6117 (XL_BE_V4SF): Likewise.
6118 (XL_BE_V2DF): Likewise.
6119 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6120 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6121 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6122 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6123 VSX_BUILTIN_VEC_XST_BE.
6124 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6125 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6127 (altivec_init_builtins): Replace conditional calls to def_builtin
6128 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6129 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6130 with unconditional calls. Remove calls to def_builtin for
6131 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6132 __builtin_vec_xst_be.
6133 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6134 to define_expand, and add alternate RTL generation for P8.
6135 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6136 vsx_ld_elemrev_v8hi.
6137 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6138 add alternate RTL generation for P8.
6139 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6140 vsx_ld_elemrev_v16qi.
6141 (vsx_st_elemrev_v8hi): Convert define_insn
6142 to define_expand, and add alternate RTL generation for P8.
6143 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6144 vsx_st_elemrev_v8hi.
6145 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6146 add alternate RTL generation for P8.
6147 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6148 vsx_st_elemrev_v16qi.
6150 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6153 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6154 TARGET_AVX512ER check.
6155 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6156 neither -mzeroupper nor -mno-zeroupper is used and
6157 TARGET_EMIT_VZEROUPPER is set.
6158 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6159 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6161 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6163 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6164 folding of vector compares.
6165 (fold_build_vec_cmp): New helper function.
6166 (fold_compare_helper): New helper function.
6167 (builtin_function_type): Add compare builtins to the list of functions
6168 having unsigned arguments. Cosmetic updates to comment indentation.
6169 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6170 the not+eq combination.
6172 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6174 PR tree-optimization/82726
6175 PR tree-optimization/70754
6176 * tree-predcom.c (order_drefs_by_pos): New function.
6177 (combine_chains): Move code setting has_max_use_after to...
6178 (try_combine_chains): ...here. New parameter. Sort combined chains
6179 according to position information.
6180 (tree_predictive_commoning_loop): Update call to above function.
6181 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6183 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6185 PR tree-optimization/82726
6187 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6189 PR tree-optimization/70754
6190 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6191 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6192 combined stmt before it if not NULL.
6193 (combine_chains): Process refs reversely and compute dominance point
6197 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6199 PR tree-optimization/79663
6200 * tree-predcom.c (combine_chains): Process refs in reverse order
6201 only for ZERO length chains, and add explaining comment.
6203 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6205 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6206 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6207 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6209 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6211 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6213 2017-11-15 Martin Liska <mliska@suse.cz>
6215 * tree-cfg.c (pass_warn_function_return::execute):
6216 Compare warn_return_type for greater than zero.
6218 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6222 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6223 to return true on Xeon and not on Xeon Phi.
6224 (ix86_check_avx256_register): Changed to ...
6225 (ix86_check_avx_upper_register): ... this. Add extra check for
6226 VALID_AVX512F_REG_OR_XI_MODE.
6227 (ix86_avx_u128_mode_needed): Changed
6228 ix86_check_avx256_register to ix86_check_avx_upper_register.
6229 (ix86_check_avx256_stores): Changed to ...
6230 (ix86_check_avx_upper_stores): ... this. Changed
6231 ix86_check_avx256_register to ix86_check_avx_upper_register.
6232 (ix86_avx_u128_mode_after): Changed
6233 avx_reg256_found to avx_upper_reg_found. Changed
6234 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6235 (ix86_avx_u128_mode_entry): Changed
6236 ix86_check_avx256_register to ix86_check_avx_upper_register.
6237 (ix86_avx_u128_mode_exit): Ditto.
6238 * config/i386/i386.h: (host_detect_local_cpu): New define.
6240 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6242 * config/arm/xgene1.md (xgene1): Split into automatons
6243 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6244 (xgene1_f_load): Adjust reservations and/or types.
6245 (xgene1_f_store): Likewise.
6246 (xgene1_load_pair): Likewise.
6247 (xgene1_store_pair): Likewise.
6248 (xgene1_fp_load1): Likewise.
6249 (xgene1_load1): Likewise.
6250 (xgene1_store1): Likewise.
6251 (xgene1_move): Likewise.
6252 (xgene1_alu): Likewise.
6253 (xgene1_simd): Likewise.
6254 (xgene1_bfm): Likewise.
6255 (xgene1_neon_load1): Likewise.
6256 (xgene1_neon_store1): Likewise.
6257 (xgene1_neon_logic): Likewise.
6258 (xgene1_neon_st1): Likewise.
6259 (xgene1_neon_ld1r): Likewise.
6260 (xgene1_alu_cond): Added.
6261 (xgene1_shift_reg): Likwise.
6262 (xgene1_bfx): Likewise.
6263 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6265 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6268 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6270 (can_widen_mult_without_libcall): New function.
6271 (expand_mul_overflow): If only checking unsigned mul overflow,
6272 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6273 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6274 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6275 (expand_DIVMOD): Formatting fix.
6276 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6277 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6278 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6280 PR tree-optimization/82977
6281 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6282 constructed temporary to strlen_to_stridx.put.
6284 2017-11-15 Martin Liska <mliska@suse.cz>
6286 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6287 * configure: Regenerate.
6289 2017-11-15 Martin Liska <mliska@suse.cz>
6292 * config/sh/sh-mem.cc: Use proper probability for
6295 2017-11-14 Jeff Law <law@redhat.com>
6297 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6298 the red zone for stack_clash_protection_final_dynamic_probe targets
6299 when the total dynamic stack size is zero bytes.
6301 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6302 blocks is post order.
6304 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6306 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6307 * final.c (rest_of_clean_state): Set it for the
6308 -fcompare-debug dump.
6309 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6310 class when TDF_COMPARE_DEBUG is set.
6312 * dwarf2out.c (gen_producer_string): Discard
6315 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6318 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6319 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6320 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6321 (__TGMATH_REAL_2_3): Remove macros.
6322 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6323 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6325 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6327 (remquo): Define using __TGMATH_REAL_3.
6329 2017-11-14 Jeff Law <law@redhat.com>
6331 * vr-values.c: New file with contents extracted from tree-vrp.c.
6332 * Makefile.in (OBJS): Add vr-values.o
6333 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6334 (set_value_range, set_and_canonicalize_value_range): Likewise.
6335 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6336 (value_range_constant_singleton, symbolic_range_p): Likewise.
6337 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6338 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6339 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6340 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6341 (find_case_label_range, find_case_label_index): Likewise.
6342 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6343 (range_int_cst_singleton_p, value_inside_range): Likewise.
6344 (get_single_symbol): Likewise.
6345 (switch_update): Move structure definition here.
6346 (to_remove_edges, to_update_switch_stmts): Provide externs.
6347 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6348 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6349 (vrp_val_is_min, set_value_range): Likewise.
6350 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6351 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6352 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6353 (range_is_nonnull, range_int_cst_p): Likewwise.
6354 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6355 (get_single_symbol, operand_less_p): Likewise
6356 (compare_values_warnv, compare_values): Likewise.
6357 (value_inside_range, value_range_constant_singleton): Likewise.
6358 (zero_nonzero_bitgs_from_vr): Likewise.
6359 (extract_range_from_binary_expr_1): Likewise.
6360 (overflow_comparison_p): Likewise.
6361 (to_remove_edges, to_update_switch_stmts): Likewise.
6362 (find_case_label-index, find_case_label_range): Likewise.
6363 (switch_update, set_value_range_to_nonnegative): Remove.
6364 (set_value_range_to_truthvalue): Likewise.
6365 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6366 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6367 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6368 (find_case_label_ranges, test_for_singularity): Likewise.
6369 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6370 (x_vr_values): Move to its remaining use site.
6372 2017-11-10 Jeff Law <law@redhat.com>
6374 * vr-values.h (VR_INITIALIZER): Move #define here.
6375 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6376 * Makefile.in (OBJS): Add tree-evrp.o
6377 * tree-vrp.h (assert_info): Move structure definition here.
6378 (set_value_range_to_varying): Prototype.
6379 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6380 (infer_value_range, register_edge_assert_for): Likewise.
6381 (stmt_interesting_for_vrp): Likewise.
6382 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6383 (set_value_range_to_varying): No longer static.
6384 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6385 (infer_value_range, register_edge_assert_for): Likewise.
6387 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6389 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6390 generate the XXBRD instruction.
6392 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6393 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6394 as being compatible if -mabi=ieeelongdouble.
6395 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6396 to setup float128 built-ins with hardware support.
6397 (BU_FLOAT128_HW_2): Likewise.
6398 (BU_FLOAT128_HW_3): Likewise.
6399 (BU_FLOAT128_HW_VSX_1): Likewise.
6400 (BU_FLOAT128_HW_VSX_2): Likewise.
6401 (scalar_extract_expq): Change float128 built-in functions to
6402 accommodate having both KFmode and TFmode functions. Use the
6403 KFmode variant as the default.
6404 (scalar_extract_sigq): Likewise.
6405 (scalar_test_neg_qp): Likewise.
6406 (scalar_insert_exp_q): Likewise.
6407 (scalar_insert_exp_qp): Likewise.
6408 (scalar_test_data_class_qp): Likewise.
6409 (sqrtf128_round_to_odd): Delete processing the round to odd
6410 built-in functions as special built-in functions, and define them
6411 as float128 built-ins. Use the KFmode variant as the default.
6412 (truncf128_round_to_odd): Likewise.
6413 (addf128_round_to_odd): Likewise.
6414 (subf128_round_to_odd): Likewise.
6415 (mulf128_round_to_odd): Likewise.
6416 (divf128_round_to_odd): Likewise.
6417 (fmaf128_round_to_odd): Likewise.
6418 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6419 support for KFmode and TFmode xststdcqp calls.
6420 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6421 point, switch the built-in handlers for the get/set float128
6422 exponent, get float128 mantissa, float128 test built-ins, and the
6423 float128 round to odd built-in functions. Eliminate creating the
6424 float128 round to odd built-in functions as special built-ins.
6425 (rs6000_init_builtins): Eliminate special creation of the float128
6426 round to odd built-in functions.
6427 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6428 function insns to support both TFmode and KFmode variants.
6429 (xsxsigqp_<mode>): Likewise.
6430 (xsiexpqpf_<mode>): Likewise.
6431 (xsiexpqp_<mode>): Likewise.
6432 (xststdcqp_<mode>): Likewise.
6433 (xststdcnegqp_<mode>): Likewise.
6434 (xststdcqp_<mode>): Likewise.
6436 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6438 * tree-ssa-threadupdate.c (compute_path_counts): Remove
6439 unused path_in_freq_ptr parameter.
6440 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
6442 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6444 * ipa-inline.c (edge_badness): Dump sreal frequency.
6445 (compute_inlined_call_time): Match natural implementaiton ...
6446 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
6447 forgotten division by CGRAPH_FREQ_BASE.
6449 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6451 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
6452 Solaris 11. Update comment.
6453 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
6455 * config/sol2.h (STARTFILE_SPEC): Likewise.
6456 * configure: Regenerate.
6458 2017-11-14 Carl Love <cel@us.ibm.com>
6460 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
6461 le_ and be_ prefixes to swap* variables. Remove
6462 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
6465 2017-11-14 Jason Merrill <jason@redhat.com>
6467 Support GTY((cache)) on hash_map.
6468 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
6469 (ggc_cache_remove): Override it instead of ggc_mx.
6470 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
6471 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
6472 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
6473 (simple_cache_map_traits): Override maybe_mx.
6474 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
6475 (hash_map): Friend gt_cleare_cache.
6476 (gt_cleare_cache): New.
6477 * tree.h (tree_cache_traits): New hash_map traits class.
6478 (tree_cache_map): New typedef.
6480 2017-11-14 Richard Biener <rguenther@suse.de>
6482 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
6483 paramter and handling.
6484 (cleanup_control_flow_bb): Likewise.
6485 (cleanup_control_flow_pre): New helper performing a DFS walk
6486 to call cleanup_control_flow_bb in PRE order.
6487 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
6488 via cleanup_control_flow_pre.
6490 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
6492 * config/aarch64/aarch64-simd.md
6493 (aarch64_simd_bsl<mode>_internal): Remove DImode.
6494 (*aarch64_simd_bsl<mode>_alt): Likewise.
6495 (aarch64_simd_bsldi_internal): New.
6496 (aarch64_simd_bsldi_alt): Likewise.
6498 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6500 * tracer.c (better_p): Do not compare frequencies.
6501 * reg-stack.c (better_edge): Likewise.
6502 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
6505 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6507 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
6508 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
6509 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
6510 * lto-streamer-in.c (input_function): Use update_max_bb_count.
6511 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
6512 * predict.c (maybe_hot_frequency_p): Inline to ...
6513 (maybe_hot_count_p): ... here; rewrite to counts.
6514 (counts_to_freqs): Rename to ...
6515 (update_max_bb_count): ... this one.
6516 (expensive_function_p): Use counts.
6517 (estimate_bb_frequencies): Update.
6518 (rebuild_frequencies): Update.
6519 * predict.h (counts_to_freqs): Rename to ...
6520 (update_max_bb_count): ... this one.
6521 * profile.c (compute_branch_probabilities): Add debug info
6522 * tree-inline.c (expand_call_inline): Update debug info.
6523 (optimize_inline_calls): Use update_max_bb_count..
6524 (tree_function_versioning): Use update_max_bb_count..
6525 * value-prof.c (gimple_value_profile_transformations):
6526 Do not use update_max_bb_count.
6528 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6530 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
6531 always use frequencies.
6533 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6535 * bb-reorder.c: Remove frequencies from comments.
6536 (better_edge_p): Use profile counts.
6537 (find_traces): Dump profile counts.
6538 (rotate_loop): Use profile counts.
6539 (find_traces_1_round): Likewise.
6540 (connect_better_edge_p): Use counts instead of probabilities for
6542 (copy_bb_p): Drop early check for non-0 frequency.
6543 (sanitize_hot_paths): Update comments.
6545 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6547 * ipa-split.c (struct split_point): Add count.
6548 (consider_split): Do not compute incoming frequency; compute incoming
6549 count and store it to split_point.
6550 (split_function): Set count of the call to split part correctly.
6552 2017-11-13 Carl Love <cel@us.ibm.com>
6554 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
6556 2017-11-13 Tom Tromey <tom@tromey.com>
6558 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
6560 2017-11-13 Carl Love <cel@us.ibm.com>
6562 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
6563 Add support for builtins:
6564 unsigned int vec_first_{,miss}_match_{,or_eos}index,
6565 vector {un,}signed {char,int,short},
6566 vector {un,}signed {char,int,short}) arguments.
6567 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
6568 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
6569 Add BU_P9V_AV_2 expansions for the builtins.
6570 * config/rs6000/altivec.h (vec_first_match_index,
6571 vec_first_mismatch_index, vec_first_match_or_eos_index,
6572 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
6573 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
6574 new extern declaration.
6575 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
6576 * config/rs6000/vsx.md (first_match_index_<mode>,
6577 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
6578 first_mismatch_or_eos_index_<mode>): Add define expand.
6579 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
6580 * doc/extend.texi: Update the built-in documenation file for the new
6583 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6585 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
6586 into the min/max operations for _Float<N> and _Float<N>X types.
6588 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6591 * dwarf2out.c (do_eh_frame): New static variable.
6592 (dwarf2out_begin_prologue): Set it.
6593 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
6595 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6597 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
6599 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
6601 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
6604 * debug.h (dwarf2out_do_eh_frame): Declare.
6605 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
6606 (dwarf2out_do_frame): Use it.
6607 (dwarf2out_do_cfi_asm): Likewise.
6608 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
6609 (dwarf2out_assembly_start): Likewise.
6610 (dwarf2out_begin_prologue): Fix comment.
6611 * toplev.c (compile_file): Always call dwarf2out_frame_finish
6612 if the target needs either debug or unwind DWARF2 info.
6613 * lto-opts.c (lto_write_options): Do not save -fexceptions,
6614 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
6615 -ftrapping-math, -ftrapv and -fwrapv.
6617 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
6619 * cgraph.c (cgraph_edge::sreal_frequency): New function.
6620 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
6621 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
6622 (estimate_edge_size_and_time): Likewise.
6623 (ipa_merge_fn_summary_after_inlining): Likewise.
6624 * ipa-inline.c (cgraph_freq_base_rec): Remove.
6625 (compute_uninlined_call_time): Use sreal_frequency.
6626 (compute_inlined_call_time): Likewise.
6627 (ipa_inline): Do not initialize cgraph_freq_base_rec.
6628 * profile-count.c: Include sreal.h.
6629 (profile_count::to_sreal_scale): New.
6630 * profile-count.h: Forward declare sreal.
6631 (profile_count::to_sreal_scale): Declare.
6633 2017-11-13 Nathan Sidwell <nathan@acm.org>
6635 * diagnostic.c (maybe_line_and_column): New.
6636 (diagnostic_get_location_text): Use it.
6637 (diagnostic_report_current_module): Likewise.
6638 (test_diagnostic_get_location_text): Add tests.
6640 2017-11-13 Luis Machado <luis.machado@linaro.org>
6642 * doc/md.texi (Specifying processor pipeline description): Fix
6643 incorrect latency for the div instruction example.
6645 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6647 PR tree-optimization/78821
6648 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
6649 that bit_not_p is the same.
6650 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6651 (split_group): Count precisely bit_not_p bits in each statement.
6652 (invert_op): New function.
6653 (imm_store_chain_info::output_merged_store): Use invert_op to
6654 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
6655 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
6657 2017-11-13 Martin Liska <mliska@suse.cz>
6659 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
6660 (struct source_info): Likewise.
6661 (add_branch_counts): Likewise.
6662 (add_line_counts): Likewise.
6663 (function_summary): Likewise.
6664 (output_intermediate_line): Likewise.
6665 (generate_results): Likewise.
6667 2017-11-13 Martin Liska <mliska@suse.cz>
6669 * gcov.c (struct block_info): Remove typedef for block_t.
6670 (struct line_info): Likewise.
6671 (line_info::has_block): Likewise.
6672 (EXIT_BLOCK): Likewise.
6673 (unblock): Likewise.
6674 (circuit): Likewise.
6675 (get_cycles_count): Likewise.
6676 (process_file): Likewise.
6677 (read_graph_file): Likewise.
6678 (solve_flow_graph): Likewise.
6679 (find_exception_blocks): Likewise.
6680 (add_line_counts): Likewise.
6681 (accumulate_line_info): Likewise.
6682 (output_line_details): Likewise.
6684 2017-11-13 Martin Liska <mliska@suse.cz>
6686 * gcov.c (struct arc_info): Remove typedef for arc_t.
6687 (struct line_info): Likewise.
6688 (add_branch_counts): Likewise.
6689 (output_branch_count): Likewise.
6690 (function_info::~function_info): Likewise.
6691 (circuit): Likewise.
6692 (output_intermediate_line): Likewise.
6693 (read_graph_file): Likewise.
6694 (solve_flow_graph): Likewise.
6695 (find_exception_blocks): Likewise.
6696 (add_line_counts): Likewise.
6697 (accumulate_line_info): Likewise.
6698 (output_line_details): Likewise.
6699 (output_function_details): Likewise.
6701 2017-11-13 Martin Liska <mliska@suse.cz>
6703 * gcov.c (struct function_info): Remove typedef for function_t.
6704 (struct source_info): Likewise.
6705 (source_info::get_functions_at_location): Likewise.
6706 (solve_flow_graph): Likewise.
6707 (find_exception_blocks): Likewise.
6708 (add_line_counts): Likewise.
6709 (output_intermediate_file): Likewise.
6710 (process_file): Likewise.
6711 (generate_results): Likewise.
6712 (release_structures): Likewise.
6713 (read_graph_file): Likewise.
6714 (read_count_file): Likewise.
6715 (accumulate_line_counts): Likewise.
6716 (output_lines): Likewise.
6718 2017-11-13 Martin Liska <mliska@suse.cz>
6720 * gcov.c (function_info::function_info): Remove num_counts
6721 and add vector<gcov_type>.
6722 (function_info::~function_info): Use the vector.
6723 (process_file): Likewise.
6724 (read_graph_file): Likewise.
6725 (read_count_file): Likewise.
6726 (solve_flow_graph): Likewise.
6728 2017-11-13 Martin Liska <mliska@suse.cz>
6730 * gcov.c (function_info::is_artificial): New function.
6731 (process_file): Erase all artificial early.
6732 (generate_results): Skip as all artificial are already
6735 2017-11-13 Martin Liska <mliska@suse.cz>
6737 * gcov.c (read_graph_file): Store to global vector of functions.
6738 (read_count_file): Iterate the vector.
6739 (process_file): Likewise.
6740 (generate_results): Likewise.
6741 (release_structures): Likewise.
6743 2017-11-13 Jakub Jelinek <jakub@redhat.com>
6745 PR tree-optimization/82954
6746 * gimple-ssa-store-merging.c
6747 (imm_store_chain_info::coalesce_immediate_stores): If
6748 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
6750 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
6752 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
6753 Upddate call to ENDIAN_LANE_N.
6754 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
6755 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
6756 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
6757 and use aarch64_endian_lane_rtx.
6758 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
6760 2017-11-12 Tom de Vries <tom@codesourcery.com>
6762 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
6764 2017-11-12 Tom de Vries <tom@codesourcery.com>
6766 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
6768 2017-11-12 Tom de Vries <tom@codesourcery.com>
6770 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6772 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6773 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6774 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6776 2017-11-11 Martin Sebor <msebor@redhat.com>
6779 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6782 * prefic.c (translate_name): Replace strncpy with memcpy to
6783 avoid -Wstringop-truncation.
6785 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6787 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6789 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6791 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6793 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6795 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6798 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6800 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6801 enable generating XXBRH if the value is in a vector register.
6802 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6803 value is in a vector register.
6804 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6805 register to register bswap64's instead of doing the GPR sequence
6806 used on previous machines.
6807 (bswapdi2_xxbrd): New insn.
6808 (bswapdi2_reg): Disallow on ISA 3.0.
6809 (register to register bswap64 splitter): Do not split the insn on
6810 ISA 3.0 systems that use XXBRD.
6812 2017-11-10 Martin Sebor <msebor@redhat.com>
6815 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6817 (find_subframework_file): Same.
6819 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6821 * auto-profile.c (afdo_indirect_call): Drop frequency.
6822 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6823 (cgraph_node::create_edge): Drop frequency argument.
6824 (cgraph_node::create_indirect_edge): Drop frequency argument.
6825 (cgraph_edge::make_speculative): Drop frequency arguments.
6826 (cgraph_edge::resolve_speculation): Do not update frequencies
6827 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6828 (cgraph_node::dump): Check consistency in IPA mode.
6829 (cgraph_edge::maybe_hot_p): Use IPA counter.
6830 (cgraph_edge::verify_count_and_frequency): Rename to ...
6831 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6832 (cgraph_node::verify_node): Update.
6833 * cgraph.h (struct cgraph_edge): Drop frequency.
6834 (cgraph_edge::frequency): New function.
6835 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6837 (cgraph_edge::rebuild_edges): Likewise.
6838 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6839 (duplicate_thunk_for_node): Do not pass frequency.
6840 (cgraph_node::create_clone): Scale only counts.
6841 (cgraph_node::create_virtual_clone): Do not pass frequency.
6842 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6843 (cgraph_node::create_version_clone): Do not pass frequency.
6844 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6845 (cgraph_node::expand_thunk): Do not pass frequency.
6846 (cgraph_node::create_wrapper): Do not pass frequency.
6847 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6849 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6850 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6851 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6852 (gather_caller_stats): Use frequency function.
6853 (ipcp_cloning_candidate_p): Use frequency function.
6854 (ipcp_propagate_stage): Use frequency function.
6855 (get_info_about_necessary_edges): Use frequency function.
6856 (update_profiling_info): Update only IPA profile.
6857 (update_specialized_profile): Use frequency functoin.
6858 (perhaps_add_new_callers): Update only IPA profile.
6859 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6860 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6861 (dump_ipa_call_summary): Use frequency function.
6862 (estimate_edge_size_and_time): Use frequency function.
6863 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6864 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6865 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6866 (update_noncloned_counts): ... ths one; scale counts only.
6867 (clone_inlined_nodes): Do not scale frequency.
6868 (inline_call): Do not pass frequency.
6869 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6870 (compute_inlined_call_time): Use IPA profile.
6871 (want_inline_small_function_p): Use IPA profile.
6872 (want_inline_self_recursive_call_p): Use IPA profile.
6873 (edge_badness): Use IPA profile.
6874 (lookup_recursive_calls): Use IPA profile.
6875 (recursive_inlining): Do not pass frequency.
6876 (resolve_noninline_speculation): Do not update frequency.
6877 (inline_small_functions): Collect max of IPA profile.
6878 (dump_overall_stats): Dump IPA porfile.
6879 (dump_inline_stats): Dump IPA porfile.
6880 (ipa_inline): Collect IPA stats.
6881 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6882 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6883 (ipa_propagate_frequency): Use frequency function.
6884 (ipa_profile): Cleanup.
6885 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6886 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6887 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6888 (input_node): Do not stream frequency.
6889 (input_edge): Do not stream frequency.
6890 (merge_profile_summaries): Scale only IPA profiles.
6891 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6892 * predict.c (drop_profile): Do not recompute frequency.
6893 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6894 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6895 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6896 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6897 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6898 * tree-inline.c (copy_bb): Do not scale frequency.
6899 (expand_call_inline): Do not scale frequency.
6900 (tree_function_versioning): Do not scale frequency.
6901 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6903 2017-11-10 Julia Koval <julia.koval@intel.com>
6905 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6906 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6907 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6908 (_mm256_maskz_gf2p8affine_epi64_epi8)
6909 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6910 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6911 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6912 (__builtin_ia32_vgf2p8affineqb_v32qi)
6913 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6914 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6916 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6920 (arm_option_override): Refactor.
6921 (arm_option_reconfigure_globals): New.
6922 (arm_options_perform_arch_sanity_checks): New.
6923 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6925 (arm_options_perform_arch_sanity_checks): Likewise
6927 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6929 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6930 (power9-qpmul): New.
6931 * rs6000/rs6000.md ("type" attr): Add qmul.
6932 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6933 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6934 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6936 2017-11-10 Martin Sebor <msebor@redhat.com>
6939 * builtins.c (compute_objsize): Handle arrays that
6940 compute_builtin_object_size likes to fail for. Make extern.
6941 * builtins.h (compute_objsize): Declare.
6942 (check_strncpy_sizes): New function.
6943 (expand_builtin_strncpy): Call check_strncpy_sizes.
6944 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6945 -Wstringop-truncation.
6946 (gimple_fold_builtin_strncat): Same.
6947 * gimple.c (gimple_build_call_from_tree): Set call location.
6948 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6949 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6950 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6951 (handle_builtin_strlen): Use strlen_to_stridx.
6952 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6954 Use strlen_to_stridx.
6955 (pass_strlen::execute): Release strlen_to_stridx.
6956 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6957 (-Wstringop-truncation): Document new option.
6959 2017-11-10 Martin Liska <mliska@suse.cz>
6961 PR gcov-profile/82702
6962 * gcov.c (main): Handle intermediate files in a different
6964 (get_gcov_intermediate_filename): New function.
6965 (output_gcov_file): Remove support of intermediate files.
6966 (generate_results): Allocate intermediate file.
6967 (release_structures): Clean-up properly fn_end.
6968 (output_intermediate_file): Start iterating with line 1.
6970 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6972 PR tree-optimization/82929
6973 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6974 ops_swapped_p non-static data member.
6975 (store_immediate_info::store_immediate_info): Clear it.
6976 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6977 ops set ops_swapped_p.
6978 (count_multiple_uses): Handle ops_swapped_p.
6980 2017-11-10 Martin Liska <mliska@suse.cz>
6982 * coverage.c (coverage_init): Stream information about
6983 support of has_unexecuted_blocks.
6984 * doc/gcov.texi: Document that.
6985 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6986 * gcov.c (read_graph_file): Likewise.
6987 (output_line_beginning): Fix a small issue with
6990 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6992 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6993 reference of trivial component.
6995 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6998 * gimple-ssa-store-merging.c
6999 (pass_store_merging::terminate_all_aliasing_chains): For
7000 gimple_store_p stmts also call refs_output_dependent_p.
7002 PR rtl-optimization/82913
7003 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7006 2017-11-09 Jeff Law <law@redhat.com>
7008 * vr-values.h: New file with vr_values class.
7009 * tree-vrp.c: Include vr-values.h
7010 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7011 data objects into the vr_values class.
7012 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7013 (get_value_range): Make it a member function within vr_values class.
7014 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7015 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7016 (op_with_constant_singleton_value_range): Likewise.
7017 (extract_range_for_var_from_comparison_expr): Likewise.
7018 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7019 (extract_range_from_binary_expr): Likewise.
7020 (extract_range_from_unary_expr): Likewise.
7021 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7022 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7023 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7024 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7025 (compare_name_with_value, compare_names): Likewise.
7026 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7027 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7028 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7029 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7030 (extract_range_from_phi_node): Likewise.
7031 (simplify_truth_ops_using_ranges): Likewise.
7032 (simplify_div_or_mod_using_ranges): Likewise.
7033 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7034 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7035 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7036 (simplify_float_conversion_using_ranges): Likewise.
7037 (simplify_internal_call_using_ranges): Likewise.
7038 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7039 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7040 poitner via x_vr_values for calls into gimple folder.
7041 (vrp_initialize_lattice): Make this the vr_values ctor.
7042 (vrp_free_lattice): Make this the vr_values dtor.
7043 (set_vr_value): New function.
7044 (class vrp_prop): Add vr_values data member. Add various member
7045 functions as well as member functions that delegate to vr_values.
7046 (check_array_ref): Make a member function within vrp_prop class.
7047 (search_for_addr_array, vrp_initialize): Likewise.
7048 (vrp_finalize): Likewise. Revamp to avoid direct access to
7049 vr_value, values_propagated, etc.
7050 (check_array_bounds): Extract vrp_prop class instance pointer from
7051 walk info structure. Use it to call member functions.
7052 (check_all_array_refs): Make a member function within vrp_prop class.
7053 Smuggle class instance pointer via walk info structure.
7054 (x_vr_values): New local static.
7055 (vrp_valueize): Use x_vr_values to get class instance.
7056 (vr_valueize_1): Likewise.
7057 (class vrp_folder): Add vr_values data member. Add various member
7058 functions as well as member functions that delegate to vr_values.
7059 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7060 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7061 class instance from vr_values. Use it to call member functions.
7062 (vrp_dom_walker): Add vr_values data member.
7063 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7064 instance via x_vr_values.
7065 (identify_jump_threads): Accept vr_values as argument. Store
7066 it into the walker structure.
7067 (evrp_dom_walker): Add vr_values class data member. Add various
7069 (evrp_dom_walker::try_find_new_range): Use vr_values data
7070 member to access the memory allocator.
7071 (evrp_dom_walker::before_dom_children): Store vr_values class
7072 instance into the vrp_folder class.
7073 (evrp_dom_walker::push_value_range): Rework to avoid direct
7074 access to num_vr_values and vr_value.
7075 (evrp_dom_walker::pop_value_range): Likewise.
7076 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7077 Use vr_values to get to dump_all_value_ranges member function.
7078 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7079 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7080 Pass vr_values class instance down to identify_jump_threads.
7081 Remove call to vrp_free_lattice.
7082 (debug_all_value_ranges): Remove.
7084 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7085 (vrp_folder): Likewise.
7087 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7088 Get it from the existing bitmap instead.
7089 (vrp_intersect_ranges_1): Likewise.
7091 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7093 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7095 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7096 set bit_not_p to it.
7097 (imm_store_chain_info::coalesce_immediate_stores): Break group
7098 if bit_not_p is different.
7099 (count_multiple_uses, split_group,
7100 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7101 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7102 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7103 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7104 to store_immediate_info ctor.
7106 2017-11-09 Jim Wilson <jimw@sifive.com>
7108 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7109 (scan_prog_file): Likewise.
7111 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7113 * bb-reorder.c (max_entry_frequency): Remove.
7114 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7115 connect_traces, push_to_next_round_p): Remove prototypes.
7116 (find_traces_1_round): Use counts only.
7117 (push_to_next_round_p): Likewise.
7118 (find_traces): Likewise.
7119 (rotate_loop): Likewise.
7120 (find_traces_1_round): Likewise.
7121 (connect_traces): Likewise.
7122 (edge_order): Likewise.
7124 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7126 * config/arm/arm.c (output_return_instruction): Add comments to
7127 indicate requirement for cmse_nonsecure_entry return to account
7128 for the size of clearing instruction output here.
7129 (thumb_exit): Likewise.
7130 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7131 return in hardfloat mode.
7133 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7135 * config/rs6000/rs6000.c (machine_function): Add a bool,
7136 "toc_is_wrapped_separately".
7137 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7138 if it wasn't explicitly set or unset, we are optimizing for speed, and
7139 doing separate shrink-wrapping.
7140 (rs6000_get_separate_components): Enable the TOC component if
7141 saving the TOC register in the prologue.
7142 (rs6000_components_for_bb): Handle the TOC component.
7143 (rs6000_emit_prologue_components): Store the TOC register where needed.
7144 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7145 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7147 2017-11-09 Martin Jambor <mjambor@suse.cz>
7149 * ipa-param-manipulation.c: New file.
7150 * ipa-param-manipulation.h: Likewise.
7151 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7152 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7153 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7154 (ipa_parm_adjustment): Likewise.
7155 (ipa_parm_adjustment_vec): Likewise.
7156 (ipa_get_vector_of_formal_parms): Moved declaration to
7157 ipa-param-manipulation.h.
7158 (ipa_get_vector_of_formal_parm_types): Likewise.
7159 (ipa_modify_formal_parameters): Likewise.
7160 (ipa_modify_call_arguments): Likewise.
7161 (ipa_combine_adjustments): Likewise.
7162 (ipa_dump_param_adjustments): Likewise.
7163 (ipa_modify_expr): Likewise.
7164 (ipa_get_adjustment_candidate): Likewise.
7165 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7166 ipa-param-manipulation.c.
7167 (ipa_get_vector_of_formal_parm_types): Likewise.
7168 (ipa_modify_formal_parameters): Likewise.
7169 (ipa_modify_call_arguments): Likewise.
7170 (ipa_modify_expr): Likewise.
7171 (get_ssa_base_param): Likewise.
7172 (ipa_get_adjustment_candidate): Likewise.
7173 (index_in_adjustments_multiple_times_p): Likewise.
7174 (ipa_combine_adjustments): Likewise.
7175 (ipa_dump_param_adjustments): Likewise.
7176 * tree-sra.c: Also include ipa-param-manipulation.h
7177 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7180 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7181 Alan Hayward <alan.hayward@arm.com>
7182 David Sherwood <david.sherwood@arm.com>
7184 * doc/sourcebuild.texi (vect_masked_store): Document.
7186 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7187 Alan Hayward <alan.hayward@arm.com>
7188 David Sherwood <david.sherwood@arm.com>
7190 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7192 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7193 Alan Hayward <alan.hayward@arm.com>
7194 David Sherwood <david.sherwood@arm.com>
7196 * doc/sourcebuild.texi (vect_variable_length): Document.
7198 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7199 Alan Hayward <alan.hayward@arm.com>
7200 David Sherwood <david.sherwood@arm.com>
7202 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7204 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7205 Alan Hayward <alan.hayward@arm.com>
7206 David Sherwood <david.sherwood@arm.com>
7208 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7210 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7211 Alan Hayward <alan.hayward@arm.com>
7212 David Sherwood <david.sherwood@arm.com>
7214 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7215 previously undocumented selectors.
7216 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7218 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7220 * doc/rtl.texi (const_vector): Say that elements can be
7221 const_wide_ints too.
7222 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7223 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7224 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7225 * optabs.c (expand_vector_broadcast): Likewise.
7227 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7228 Alan Hayward <alan.hayward@arm.com>
7229 David Sherwood <david.sherwood@arm.com>
7231 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7232 scaled index even if the unscaled address was invalid.
7233 Don't increase the complexity of using a scale in that case.
7235 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7236 Alan Hayward <alan.hayward@arm.com>
7237 David Sherwood <david.sherwood@arm.com>
7239 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7240 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7241 UNITS_PER_WORD bytes.
7242 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7243 based on REGMODE_NATURAL_SIZE of the inner mode.
7244 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7245 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7246 * expmed.c (lowpart_bit_field_p): Divide the value up into
7247 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7248 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7249 whether something is likely to occupy more than one register.
7251 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7254 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7255 profile_count::adjust_for_ipa_scaling.
7256 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7257 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7259 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7261 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7263 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7264 (split_group): Add total_orig and total_new arguments, estimate the
7265 number of statements related to the store group without store merging
7266 and with store merging.
7267 (imm_store_chain_info::output_merged_store): Adjust split_group
7268 callers, punt if estimated number of statements with store merging
7269 is not smaller than estimated number of statements without it.
7271 (handled_load): Remove has_single_use checks.
7272 (pass_store_merging::process_store): Likewise.
7274 2017-11-09 Richard Biener <rguenther@suse.de>
7276 PR tree-optimization/82902
7277 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7279 2017-11-09 Martin Liska <mliska@suse.cz>
7282 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7285 2017-11-09 Martin Liska <mliska@suse.cz>
7287 PR tree-optimization/82669
7288 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7290 2017-11-09 Martin Liska <mliska@suse.cz>
7292 PR gcov-profile/48463
7293 * coverage.c (coverage_begin_function): Output also end locus
7294 of a function and information whether the function is
7296 * gcov-dump.c (tag_function): Parse and print the information.
7297 * gcov.c (INCLUDE_MAP): Add include.
7298 (INCLUDE_SET): Likewise.
7299 (struct line_info): Move earlier in the source file because
7300 of vector<line_info> in function_info structure.
7301 (line_info::line_info): Likewise.
7302 (line_info::has_block): Likewise.
7303 (struct source_info): Add new member index.
7304 (source_info::get_functions_at_location): New function.
7305 (function_info::group_line_p): New function.
7306 (output_intermediate_line): New function.
7307 (output_intermediate_file): Use the mentioned function.
7308 (struct function_start): New.
7309 (struct function_start_pair_hash): Likewise.
7310 (process_file): Add code that identifies group functions.
7311 Assign lines either to global or function scope.
7312 (generate_results): Skip artificial functions.
7313 (find_source): Assign index for each source file.
7314 (read_graph_file): Read new flag artificial and end_line.
7315 (add_line_counts): Assign it either to global of function scope.
7316 (accumulate_line_counts): Isolate core of the function to
7317 accumulate_line_info and call it for both function and global
7319 (accumulate_line_info): New function.
7320 (output_line_beginning): Fix GNU coding style.
7321 (print_source_line): New function.
7322 (output_line_details): Likewise.
7323 (output_function_details): Likewise.
7324 (output_lines): Iterate both source (global) scope and function
7326 (struct function_line_start_cmp): New class.
7327 * doc/gcov.texi: Reflect changes in documentation.
7329 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7332 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7333 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7334 and similarly for not instead of neg.
7336 2017-11-08 Andi Kleen <ak@linux.intel.com>
7338 * config/i386/i386.opt: Add -mforce-indirect-call.
7339 * config/i386/predicates.md: Check for flag_force_indirect_call.
7340 * doc/invoke.texi: Document -mforce-indirect-call
7342 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7344 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7346 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7347 config/riscv/riscv.c (predict.h): New include.
7348 (riscv_slow_unaligned_access_p): No longer static.
7349 (riscv_block_move_straight): Add require.
7350 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7352 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7355 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7356 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7357 nonimmediate_operand predicate for operand 1 instead of
7360 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7362 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7364 * config/aarch64/constraints.md (Uml): New constraint.
7365 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7368 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7370 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7371 of two vec_duplicates into a vec_concat.
7373 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7375 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7376 Simplify vec_merge of vec_duplicate and vec_concat.
7377 * config/aarch64/constraints.md (Utq): New constraint.
7378 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7381 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7383 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7384 Simplify vec_merge of vec_duplicate and const_vector.
7385 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7387 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7388 mode iterator. Update predicate on operand 1 to
7389 handle non-const_vec constants. Delete constraints.
7390 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7392 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7394 PR tree-optimization/78821
7395 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7397 (store_operand_info::store_operand_info): Initialize it to false.
7398 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7399 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7400 store in the group, and if chain_info is non-NULL, to ignore altogether
7402 (compatible_load_p): Fail if bit_not_p does not match.
7403 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7404 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7405 (pass_store_merging::process_store): Adjust
7406 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7407 call terminate_all_aliasing_chains newly when adding a store into
7408 a chain with non-NULL chain_info.
7410 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7412 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7414 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7416 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7418 (aarch64_layout_frame): Initialise emit_frame_chain.
7419 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7420 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7422 2017-11-08 Martin Liska <mliska@suse.cz>
7424 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7425 of gimple_call_internal_p.
7427 2017-11-07 Tom de Vries <tom@codesourcery.com>
7429 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7431 2017-11-07 Tom de Vries <tom@codesourcery.com>
7433 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
7436 2017-11-08 Martin Liska <mliska@suse.cz>
7439 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
7441 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7443 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
7445 2017-11-07 Jakub Jelinek <jakub@redhat.com>
7448 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
7449 (*cmp<mode>_ccz_1): New insn with $k alternative.
7452 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
7453 RTX_COMM_COMPARE as commutative as well.
7454 (ix86_binary_operator_ok): Formatting fix.
7455 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
7456 *<code><mode>3<mask_name><round_saeonly_name>,
7457 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
7458 *<s>mul<mode>3_highpart<mask_name>,
7459 *vec_widen_umult_even_v16si<mask_name>,
7460 *vec_widen_umult_even_v8si<mask_name>,
7461 *vec_widen_umult_even_v4si<mask_name>,
7462 *vec_widen_smult_even_v16si<mask_name>,
7463 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
7464 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
7465 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
7466 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
7467 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
7468 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
7469 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
7470 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
7471 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
7472 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
7473 ix86_binary_operator_ok. Formatting fixes.
7474 (*<plusminus_insn><mode>3<mask_name><round_name>,
7475 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
7478 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
7480 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
7481 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
7482 eq and ne if TARGET_ISEL.
7483 (cmp): New code_iterator.
7484 (UNS, UNSU_, UNSIK): New code_attrs.
7485 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
7486 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
7488 ("eq<mode>3"): ... this.
7489 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
7491 ("ne<mode>3"): ... this.
7493 2017-11-07 Julia Koval <julia.koval@intel.com>
7496 * common/config/i386/i386-common.c
7497 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
7498 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
7499 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
7500 * config/i386/i386.opt: Ditto.
7501 * config/i386/i386.c (ix86_target_string): Ditto.
7502 (ix86_option_override_internal): Ditto.
7503 (ix86_init_mpx_builtins): Move MPX to args2.
7504 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
7505 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7506 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7507 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7508 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7509 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7510 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
7512 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
7515 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
7516 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
7517 (zero-extendsidi peephole2): Remove peephole.
7519 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
7522 * stor-layout.c: Include attribs.h.
7523 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
7524 explicit lookup of "aligned" attribute.
7526 2017-11-07 Andrew Waterman <andrew@sifive.com>
7528 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
7529 (riscv_expand_block_move): Likewise.
7530 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
7532 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
7533 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
7534 * config/riscv/riscv.c (riscv_block_move_straight): New function.
7535 (riscv_adjust_block_mem): Likewise.
7536 (riscv_block_move_loop): Likewise.
7537 (riscv_expand_block_move): Likewise.
7538 * config/riscv/riscv.md (movmemsi): New pattern.
7540 2017-11-07 Michael Clark <michaeljclark@mac.com>
7542 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
7543 (MUSL_DYNAMIC_LINKER): Likewise.
7545 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
7547 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
7548 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
7551 2017-11-07 Richard Biener <rguenther@suse.de>
7553 * match.pd: Fix build.
7555 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
7556 Jackson Woodruff <jackson.woodruff@arm.com>
7558 PR tree-optimization/71026
7559 * match.pd: Canonicalize negate in division.
7561 2017-11-07 Sudakshina Das <sudi.das@arm.com>
7564 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
7566 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7568 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
7569 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
7571 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
7573 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
7574 non-scalar integral types.
7575 * match.pd (negate_expr_p): Handle MINUS_EXPR.
7576 (-(A-B), -(~A)): New transformations.
7578 2017-11-07 Tom de Vries <tom@codesourcery.com>
7580 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
7581 semicolon after "do {} while (0)".
7582 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7583 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7584 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7585 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7586 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7588 2017-11-07 Tom de Vries <tom@codesourcery.com>
7590 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
7591 after "do {} while (0)".
7592 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7593 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7594 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7595 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7596 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
7598 2017-11-07 Tom de Vries <tom@codesourcery.com>
7601 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
7603 (arm_rtx_costs_internal): Add missing semicolon after
7604 HANDLE_NARROW_SHIFT_ARITH call.
7606 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7608 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7609 disable isel if it was not set explicitly.
7611 2017-11-06 James Bowman <james.bowman@ftdichip.com>
7613 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
7614 (add_type_attribute) likewise.
7616 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
7618 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
7619 of struct ix86_frame.
7620 (ix86_initial_elimination_offset): Likewise.
7621 (ix86_expand_split_stack_prologue): Likewise.
7623 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
7625 * tree-vrp.h (enum value_range_type): Update stale comment.
7627 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7628 Alan Hayward <alan.hayward@arm.com>
7629 David Sherwood <david.sherwood@arm.com>
7631 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
7632 (aarch64_expand_vec_perm_const): Take the number of units too.
7633 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
7634 (aarch64_expand_vec_perm_const): Likewise.
7635 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
7636 (vec_perm<mode>): Update accordingly.
7638 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7639 Alan Hayward <alan.hayward@arm.com>
7640 David Sherwood <david.sherwood@arm.com>
7642 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
7643 Take the number of units too.
7644 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
7645 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
7646 but check for a vector mode before rather than after the call.
7647 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7648 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
7649 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
7650 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
7651 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
7652 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
7653 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
7654 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
7655 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
7656 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
7657 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
7658 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
7659 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
7660 (aarch64_sqdmull2_n<mode>): Update accordingly.
7662 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7663 Alan Hayward <alan.hayward@arm.com>
7664 David Sherwood <david.sherwood@arm.com>
7666 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
7667 the number of units too.
7668 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
7669 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
7670 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
7671 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
7672 (vec_store_lanesxi<mode>): Update accordingly.
7674 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7675 Alan Hayward <alan.hayward@arm.com>
7676 David Sherwood <david.sherwood@arm.com>
7678 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
7679 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
7680 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
7681 of units rather than the mode.
7682 * config/aarch64/iterators.md (nunits): New mode attribute.
7683 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7684 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
7685 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
7686 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
7687 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
7688 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
7689 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
7690 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
7691 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
7692 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
7693 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
7694 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
7695 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
7696 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
7697 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
7698 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
7699 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
7700 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
7701 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
7702 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
7703 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
7704 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
7705 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
7706 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
7707 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
7708 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
7709 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
7710 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
7711 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
7712 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
7713 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
7714 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
7715 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
7716 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
7717 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
7718 (aarch64_simd_vec_setv2di): Likewise.
7720 2017-11-06 Carl Love <cel@us.ibm.com>
7722 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
7724 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
7725 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
7727 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
7729 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
7730 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
7731 Add power 8 macro expansions.
7732 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
7733 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
7734 power 8 instructions. (VSX_XXBR): Add iterator.
7736 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
7738 * config/arm/arm.md (predicable_short_it): Change default to "no",
7739 improve documentation, remove uses that are identical to the default.
7740 (enabled_for_depr_it): Rename to enabled_for_short_it.
7741 * gcc/config/arm/arm-fixed.md (predicable_short_it):
7742 Remove default uses.
7743 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
7744 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
7745 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
7746 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
7748 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7751 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
7752 float128 helper macros, which are no longer used after deleting
7753 the old 'q' built-in functions, and moving the round to odd
7754 built-in functions to being special built-in functions.
7755 (BU_FLOAT128_2): Likewise.
7756 (BU_FLOAT128_1_HW): Likewise.
7757 (BU_FLOAT128_2_HW): Likewise.
7758 (BU_FLOAT128_3_HW): Likewise.
7759 (FABSQ): Delete old 'q' built-in functions.
7760 (COPYSIGNQ): Likewise.
7761 (SQRTF128_ODD): Move round to odd built-in functions to be
7762 special built-in functions, so that we can handle
7763 -mabi=ieeelongdouble.
7764 (TRUNCF128_ODD): Likewise.
7765 (ADDF128_ODD): Likewise.
7766 (SUBF128_ODD): Likewise.
7767 (MULF128_ODD): Likewise.
7768 (DIVF128_ODD): Likewise.
7769 (FMAF128_ODD): Likewise.
7770 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7771 built-in names to 'f128'.
7772 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7773 old 'q' built-in functions, as the machine independent code for
7774 'f128' built-in functions handles this.
7775 (rs6000_expand_builtin): Add expansion for float128 round to odd
7776 functions, keying off on -mabi=ieeelongdouble of whether to use
7777 the KFmode or TFmode variant.
7778 (rs6000_init_builtins): Initialize the _Float128 round to odd
7780 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7781 _Float128 'q' built-in functions are now mapped into the new
7782 'f128' built-in functions.
7784 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7786 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7787 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7789 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7792 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7793 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7794 * toplev.c: Include "ipa-fnsummary.h".
7795 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7797 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7799 PR tree-optimization/82838
7800 * gimple-ssa-store-merging.c
7801 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7802 on a separate gimple_seq which is then appended to seq.
7804 2017-11-06 Jeff Law <law@redhat.com>
7807 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7808 (get_probe_interval): New functions.
7809 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7810 (ix86_adjust_stack_and_probe): Likewise.
7811 (output_adjust_stack_and_probe): Likewise.
7812 (ix86_emit_probe_stack_range): Likewise.
7813 (ix86_expand_prologue): Likewise.
7815 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7817 PR tree-optimization/82816
7818 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7819 if the modes of the two types are the same.
7820 (convert_plusminus_to_widen): Likewise.
7822 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7824 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7826 (usadv16qi): New define_expand.
7827 (usadv8hi): New define_expand.
7829 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7832 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7834 (inline_transform): Likewise.
7835 * predict.c (counts_to_freqs): Remove useless conditional.
7836 * profile-count.h (profile_count::apply_scale): Move sanity check.
7837 * tree-inline.c (copy_bb): Always scale.
7838 (copy_cfg_body): Likewise.
7840 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7843 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7845 (*cmp_ite0): Add enabled_for_depr_it attribute.
7846 (*cmp_ite1): Likewise.
7848 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7850 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7853 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7855 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7856 return the value by pointer.
7857 (extract_range_from_multiplicative_op_1): Update accordingly.
7858 Return as soon as an operation fails.
7860 2017-11-05 Tom de Vries <tom@codesourcery.com>
7863 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7864 (DEF_SANITIZER_BUILTIN): ... here.
7865 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7866 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7868 2017-11-05 Tom de Vries <tom@codesourcery.com>
7871 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7873 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7874 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7875 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7877 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7878 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7879 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7881 2017-11-05 Tom de Vries <tom@codesourcery.com>
7884 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7887 2017-11-04 Michael Clark <michaeljclark@mac.com>
7889 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7890 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7892 (*addsi3_extended): Likewise.
7893 (*addsi3_extended2): Likewise.
7894 (<optab>si3): Likewise.
7895 (<optab>di3): Likewise.
7896 (<optab><mode>3): Likewise.
7897 (<*optabe>si3_internal): Likewise.
7898 (zero_extendqi<SUPERQI:mode>2): Likewise.
7899 (*add<mode>hi3): Likewise.
7900 (*xor<mode>hi3): Likewise.
7901 (<optab>di3): Likewise.
7902 (*<optab>si3_extend): Likewise.
7903 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7904 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7905 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7907 2017-11-04 Andrew Waterman <andrew@sifive.com>
7909 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7910 TARGET_STRICT_ALIGN based upon -mtune argument.
7912 2017-11-04 Andrew Waterman <andrew@sifive.com>
7914 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7916 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7918 * config/i386/i386.c (choose_basereg): Use optional scratch
7919 register and add assertion.
7920 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7921 needed, and don't allocate stack.
7922 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7923 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7924 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7926 2017-11-03 Jeff Law <law@redhat.com>
7928 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7929 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7930 to probe at the start of a noreturn function.
7932 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7934 PR tree-optimization/78821
7935 * gimple-ssa-store-merging.c: Update the file comment.
7936 (MAX_STORE_ALIAS_CHECKS): Define.
7937 (struct store_operand_info): New type.
7938 (store_operand_info::store_operand_info): New constructor.
7939 (struct store_immediate_info): Add rhs_code and ops data members.
7940 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7941 and op1r arguments to the ctor, initialize corresponding data members.
7942 (struct merged_store_group): Add load_align_base and load_align
7944 (merged_store_group::merged_store_group): Initialize them.
7945 (merged_store_group::do_merge): Update them.
7946 (merged_store_group::apply_stores): Pick the constant for
7947 encode_tree_to_bitpos from one of the two operands, or skip
7948 encode_tree_to_bitpos if neither operand is a constant.
7949 (class pass_store_merging): Add process_store method decl. Remove
7950 bool argument from terminate_all_aliasing_chains method decl.
7951 (pass_store_merging::terminate_all_aliasing_chains): Remove
7952 var_offset_p argument and corresponding handling.
7953 (stmts_may_clobber_ref_p): New function.
7954 (compatible_load_p): New function.
7955 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7956 if there is overlap and rhs_code is not INTEGER_CST. For
7957 non-overlapping stores terminate group if rhs is not mergeable.
7958 (get_alias_type_for_stmts): Change first argument from
7959 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7960 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7961 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7963 (get_location_for_stmts): Change first argument from
7964 auto_vec<gimple *> & to vec<gimple *> &.
7965 (struct split_store): Remove orig_stmts data member, add orig_stores.
7966 (split_store::split_store): Create orig_stores rather than orig_stmts.
7967 (find_constituent_stmts): Renamed to ...
7968 (find_constituent_stores): ... this. Change second argument from
7969 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7970 to info structures rather than the statements.
7971 (split_group): Rename ALLOW_UNALIGNED argument to
7972 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7973 it. Adjust find_constituent_stores caller.
7974 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7975 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7976 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7977 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7978 (mem_valid_for_store_merging): New function.
7979 (handled_load): New function.
7980 (pass_store_merging::process_store): New method.
7981 (pass_store_merging::execute): Use process_store method. Adjust
7982 terminate_all_aliasing_chains caller.
7984 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7986 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7987 Return true for more constants, symbols and label references.
7988 (aarch64_valid_floating_const): Remove unused function.
7990 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7993 * config/aarch64/aarch64.c (aarch64_layout_frame):
7994 Undo forcing of LR at bottom of frame.
7996 2017-11-03 Jeff Law <law@redhat.com>
7999 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8000 for int_registers_saved.
8002 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8003 extracted from tree-ssa-dom.c.
8004 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8005 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8006 (record_equivalences_from_incoming_edge): Add additional argument
8007 to single_pred_edge_ignoring_loop_edges call.
8008 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8009 (uncprop_dom_walker::before_dom_children): Add additional argument
8010 to single_pred_edge_ignoring_loop_edges call.
8011 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8012 single_pred_edge_ignoring_loop_edges rather than open coding.
8013 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8015 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8017 * match.pd (-(-A)): Rewrite.
8019 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8021 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8022 (rs6000_emit_int_cmove): New declaration.
8023 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8024 (rs6000_emit_sISEL): Delete.
8025 (rs6000_emit_int_cmove): Make non-static.
8026 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8027 instead of rs6000_emit_sISEL.
8029 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8031 * asan.c (create_cond_insert_point): Maintain profile.
8032 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8034 * basic-block.h (struct basic_block_def): Remove frequency.
8035 (EDGE_FREQUENCY): Use to_frequency
8036 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8038 (find_traces): Update to use to_frequency.
8039 (find_traces_1_round): Likewise; use only IPA counts.
8040 (bb_to_key): Likewise.
8041 (connect_traces): Use IPA counts only.
8042 (copy_bb_p): Update to use to_frequency.
8043 (fix_up_crossing_landing_pad): Likewise.
8044 (sanitize_hot_paths): Likewise.
8045 * bt-load.c (basic_block_freq): Likewise.
8046 * cfg.c (init_flow): Set count_max to uninitialized.
8047 (check_bb_profile): Remove frequencies; check counts.
8048 (dump_bb_info): Do not dump frequencies.
8049 (update_bb_profile_for_threading): Update counts only.
8050 (scale_bbs_frequencies_int): Likewise.
8051 (MAX_SAFE_MULTIPLIER): Remove.
8052 (scale_bbs_frequencies_gcov_type): Update counts only.
8053 (scale_bbs_frequencies_profile_count): Update counts only.
8054 (scale_bbs_frequencies): Update counts only.
8055 * cfg.h (struct control_flow_graph): Add count-max.
8056 (update_bb_profile_for_threading): Update prototype.
8057 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8058 (find_many_sub_basic_blocks): Likewise.
8059 * cfgcleanup.c (try_forward_edges): Likewise.
8060 (try_crossjump_to_edge): Likewise.
8061 * cfgexpand.c (expand_gimple_cond): Likewise.
8062 (expand_gimple_tailcall): Likewise.
8063 (construct_init_block): Likewise.
8064 (construct_exit_block): Likewise.
8065 * cfghooks.c (verify_flow_info): Check consistency of counts.
8066 (dump_bb_for_graph): Do not dump frequencies.
8067 (split_block_1): Do not update frequencies.
8068 (split_edge): Do not update frequencies.
8069 (make_forwarder_block): Do not update frequencies.
8070 (duplicate_block): Do not update frequencies.
8071 (account_profile_record): Do not update frequencies.
8072 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8073 for global heuristics.
8074 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8075 (expected_loop_iterations_unbounded): Use counts only.
8076 * cfgloopmanip.c (scale_loop_profile): Simplify.
8077 (create_empty_loop_on_edge): Simplify
8079 (duplicate_loop_to_header_edge): Simplify
8080 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8081 (update_br_prob_note): Take care of removing note when profile
8083 (relink_block_chain): Do not dump frequency.
8084 (rtl_account_profile_record): Use to_frequency.
8085 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8086 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8087 (cgraph_update_edges_for_call_stmt_node): Likewise.
8088 (cgraph_edge::verify_count_and_frequency): Update.
8089 (cgraph_node::verify_node): Temporarily disable frequency verification.
8090 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8091 to_cgraph_frequency.
8092 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8093 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8095 (cgraph_node::expand_thunk): Update profile.
8096 * except.c (dw2_build_landing_pads): Do not update frequency.
8097 * final.c (compute_alignments): Use to_frequency.
8098 (dump_basic_block_info): Do not dump frequency.
8099 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8100 (dump_gimple_bb_header): Do not dump frequency.
8101 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8103 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8104 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8105 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8106 (init_before_recovery): Do not update frequency.
8107 (sched_create_recovery_edges): Do not update frequency.
8108 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8109 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8110 (ipa_cp_c_finalize): Set max_count to uninitialized.
8111 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8112 (param_change_prob): Use counts.
8113 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8115 * ipa-split.c (consider_split): Use to_frequency.
8116 (split_function): Use to_frequency.
8117 * ira-build.c (loop_compare_func): Likewise.
8118 (mark_loops_for_removal): Likewise.
8119 (mark_all_loops_for_removal): Likewise.
8120 * loop-doloop.c (doloop_modify): Do not update frequency.
8121 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8123 * lto-streamer-in.c (input_function): Update count_max.
8124 * omp-expand.c (expand_omp_taskreg): Update count_max.
8125 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8126 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8127 (maybe_hot_count_p): Use ipa counts only.
8128 (maybe_hot_bb_p): Simplify.
8129 (maybe_hot_edge_p): Simplify.
8130 (probably_never_executed): Do not take frequency argument.
8131 (probably_never_executed_bb_p): Do not pass frequency.
8132 (probably_never_executed_edge_p): Likewise.
8133 (combine_predictions_for_bb): Check that profile is nonzero.
8134 (propagate_freq): Do not set frequency.
8135 (drop_profile): Simplify.
8136 (counts_to_freqs): Simplify.
8137 (expensive_function_p): Use to_frequency.
8138 (propagate_unlikely_bbs_forward): Simplify.
8139 (determine_unlikely_bbs): Simplify.
8140 (estimate_bb_frequencies): Add hack to silence graphite issues.
8141 (compute_function_frequency): Use ipa counts.
8142 (pass_profile::execute): Update.
8143 (rebuild_frequencies): Use counts only.
8144 (force_edge_cold): Use counts only.
8145 * profile-count.c (profile_count::dump): Dump new count types.
8146 (profile_count::differs_from_p): Check compatiblity.
8147 (profile_count::to_frequency): New function.
8148 (profile_count::to_cgraph_frequency): New function.
8149 * profile-count.h (struct function): Declare.
8150 (enum profile_quality): Add profile_guessed_local and
8151 profile_guessed_global0.
8152 (class profile_proability): Decrease number of bits to 29;
8153 update from_reg_br_prob_note and to_reg_br_prob_note.
8154 (class profile_count: Update comment; decrease number of bits
8155 to 61. Check compatibility.
8156 (profile_count::compatible_p): New private member function.
8157 (profile_count::ipa_p): New member function.
8158 (profile_count::operator<): Handle global zero correctly.
8159 (profile_count::operator>): Handle global zero correctly.
8160 (profile_count::operator<=): Handle global zero correctly.
8161 (profile_count::operator>=): Handle global zero correctly.
8162 (profile_count::nonzero_p): New member function.
8163 (profile_count::force_nonzero): New member function.
8164 (profile_count::max): New member function.
8165 (profile_count::apply_scale): Handle IPA scalling.
8166 (profile_count::guessed_local): New member function.
8167 (profile_count::global0): New member function.
8168 (profile_count::ipa): New member function.
8169 (profile_count::to_frequency): Declare.
8170 (profile_count::to_cgraph_frequency): Declare.
8171 * profile.c (OVERLAP_BASE): Delete.
8172 (compute_frequency_overlap): Delete.
8173 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8174 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8175 * sched-ebb.c (rank): Use counts only.
8176 * shrink-wrap.c (handle_simple_exit): Use counts only.
8177 (try_shrink_wrapping): Use counts only.
8178 (place_prologue_for_one_component): Use counts only.
8179 * tracer.c (find_best_predecessor): Use to_frequency.
8180 (find_trace): Use to_frequency.
8181 (tail_duplicate): Use to_frequency.
8182 * trans-mem.c (expand_transaction): Do not update frequency.
8183 * tree-call-cdce.c: Do not update frequency.
8184 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8185 (gimple_merge_blocks): Likewise.
8186 (gimple_split_edge): Likewise.
8187 (gimple_duplicate_sese_region): Likewise.
8188 (gimple_duplicate_sese_tail): Likewise.
8189 (move_sese_region_to_fn): Likewise.
8190 (gimple_account_profile_record): Likewise.
8191 (insert_cond_bb): Likewise.
8192 * tree-complex.c (expand_complex_div_wide): Likewise.
8193 * tree-eh.c (lower_resx): Update profile.
8194 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8196 (initialize_cfun): Do not initialize frequencies
8197 (freqs_to_counts): Delete.
8198 (copy_cfg_body): Ignore count parameter.
8199 (copy_body): Update.
8200 (expand_call_inline): Update count_max.
8201 (optimize_inline_calls): Update count_max.
8202 (tree_function_versioning): Update count_max.
8203 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8204 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8206 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8207 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8208 (try_peel_loop): Likewise.
8209 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8211 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8212 (tree_transform_and_unroll_loop): Do not use frequencies
8213 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8214 Use reliable prediction only.
8215 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8216 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8217 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8218 probability scaling.
8219 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8220 not update frequency
8221 (any_remaining_duplicated_blocks): Likewise.
8222 (update_profile): Likewise.
8223 (estimated_freqs_path): Delete.
8224 (freqs_to_counts_path): Delete.
8225 (clear_counts_path): Delete.
8226 (ssa_fix_duplicate_block_edges): Likewise.
8227 (duplicate_thread_path): Likewise.
8228 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8229 * tree-tailcall.c (decrease_profile): Do not update frequency.
8230 (eliminate_tail_call): Likewise.
8231 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8232 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8233 (optimize_mask_stores): Likewise.
8234 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8235 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8236 (ubsan_expand_ptr_ifn): Update profile.
8237 * value-prof.c (gimple_ic): Simplify.
8238 * value-prof.h (gimple_ic): Update prototype.
8239 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8240 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8242 (want_inline_self_recursive_call_p): Likewise.
8243 (resolve_noninline_speculation): Only cummulate defined counts.
8244 (inline_small_functions): Use nonzero_p.
8245 (ipa_inline): Do not access freed node.
8247 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8249 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8250 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8252 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8254 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8255 non-legitimate address.
8257 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8259 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8260 (*lt0_<mode>di, *lt0_<mode>si): New.
8262 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8264 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8265 TARGET_PAIRED_FLOAT.
8267 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8268 Jim Wilson <jim.wilson@linaro.org>
8270 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8271 * config/aarch64/aarch64-tune.md: Regenerated.
8272 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8273 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8275 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8277 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8278 (arc_expand_prologue): Restore blink for millicode.
8279 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8281 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8284 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8285 gen_vec_duplicate after forcing the scalar into a register.
8287 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8289 * combine (try_combine): Print the insns input to try_combine to the
8292 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8295 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8296 Remove second argument from aarch64_process_target_attr call.
8297 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8299 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8301 (aarch64_handle_attr_arch): Remove second argument.
8302 (aarch64_handle_attr_cpu): Ditto.
8303 (aarch64_handle_attr_tune): Ditto.
8304 (aarch64_handle_attr_isa_flags): Ditto.
8305 (aarch64_process_one_target_attr): Ditto.
8306 (aarch64_process_target_attr): Ditto.
8307 (aarch64_option_valid_attribute_p): Remove second argument.
8308 on aarch64_process_target_attr call.
8310 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8312 * diagnostic.c: Include "selftest-diagnostic.h".
8313 (selftest::assert_location_text): New function.
8314 (selftest::test_diagnostic_get_location_text): New function.
8315 (selftest::diagnostic_c_tests): Call it.
8317 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8319 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8320 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8321 (class selftest::test_diagnostic_context): Move to...
8322 * selftest-diagnostic.c: New file.
8323 * selftest-diagnostic.h: New file.
8325 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8327 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8328 offset range for FT32B.
8329 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8330 * config/ft32/ft32.md: Add TARGET_NOPM.
8331 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8332 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8334 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8336 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8338 2017-11-02 Jeff Law <law@redhat.com>
8340 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8341 virtual keyword on FINAL OVERRIDE members.
8343 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8344 virtuals together. Add virtual destructor.
8345 (substitute_and_fold_engine): Similarly.
8347 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8349 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8351 2017-11-02 Richard Biener <rguenther@suse.de>
8353 PR tree-optimization/82795
8354 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8356 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8358 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8360 (gcc_GAS_CHECK_FEATURE): Remove.
8361 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8362 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8363 * configure: Regenerate.
8365 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8367 * config/arc/arc.c (hwloop_optimize): Account for empty
8370 2017-11-02 Richard Biener <rguenther@suse.de>
8373 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8374 Truncate ARRAY_REF index and element size.
8376 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8378 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8380 2017-11-01 Jeff Law <law@redhat.com>
8382 * tree-ssa-ccp.c (ccp_folder): New class derived from
8383 substitute_and_fold_engine.
8384 (ccp_folder::get_value): New member function.
8385 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8386 (ccp_fold_stmt): Remove prototype.
8387 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8388 * tree-ssa-copy.c (copy_folder): New class derived from
8389 substitute_and_fold_engine.
8390 (copy_folder::get_value): Renamed from get_value.
8391 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8392 * tree-vrp.c (vrp_folder): New class derived from
8393 substitute_and_fold_engine.
8394 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8395 (vrp_folder::get_value): New member function.
8396 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8397 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8398 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8399 provide a class interface to folder/substitute routines.
8400 (ssa_prop_fold_stmt_fn): Remove typedef.
8401 (ssa_prop_get_value_fn): Likewise.
8402 (subsitute_and_fold): Remove prototype.
8403 (replace_uses_in): Likewise.
8404 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8405 Renamed from replace_uses_in. Call the virtual member function
8406 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8407 (substitute_and_fold_dom_walker): Remove initialization of
8408 data member entries for calbacks. Add substitute_and_fold_engine
8409 member and initialize it.
8410 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8411 member functions for get_value, replace_phi_args_in c
8412 replace_uses_in, and fold_stmt calls.
8413 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8414 substitute_and_fold. Remove assert. Update ctor call.
8416 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8417 (ssa_prop_visit_phi_fn): Likewise.
8418 (class ssa_propagation_engine): New class to provide an interface
8420 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8422 (ssa_prop_visit_phi): Likewise.
8423 (ssa_propagation_engine::simulate_stmt): Moved into class.
8424 Call visit_phi/visit_stmt from the class rather than via
8425 file scoped static variables.
8426 (ssa_propagation_engine::simulate_block): Moved into class.
8427 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8428 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8429 set file scoped statics for the visit_stmt/visit_phi callbacks.
8430 * tree-complex.c (complex_propagate): New class derived from
8431 ssa_propagation_engine.
8432 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8433 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
8434 (tree_lower_complex): Call ssa_propagate via the complex_propagate
8436 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
8437 ssa_propagation_engine.
8438 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
8439 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
8440 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
8441 * tree-ssa-copy.c (copy_prop): New class derived from
8442 ssa_propagation_engine.
8443 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
8444 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
8445 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
8446 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
8447 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
8448 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
8449 (execute_vrp): Call ssa_propagate from the vrp_prop class.
8451 2017-11-01 Jakub Jelinek <jakub@redhat.com>
8453 PR rtl-optimization/82778
8454 PR rtl-optimization/82597
8455 * compare-elim.c (struct comparison): Add in_a_setter field.
8456 (find_comparison_dom_walker::before_dom_children): Remove killed
8457 bitmap and df_simulate_find_defs call, instead walk the defs.
8458 Compute last_setter and initialize in_a_setter. Merge definitions
8459 with first initialization for a few variables.
8460 (try_validate_parallel): Use insn_invalid_p instead of
8461 recog_memoized. Return insn rather than just the pattern.
8462 (try_merge_compare): Fix up comment. Don't uselessly test if
8463 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
8465 (execute_compare_elim_after_reload): Remove df_chain_add_problem
8468 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8469 Alan Hayward <alan.hayward@arm.com>
8470 David Sherwood <david.sherwood@arm.com>
8472 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
8473 aarch64_hard_regno_nregs to get the number of registers
8476 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8477 Alan Hayward <alan.hayward@arm.com>
8478 David Sherwood <david.sherwood@arm.com>
8480 * config/aarch64/constraints.md (Upl): Rename to...
8482 * config/aarch64/aarch64.md
8483 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
8486 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8487 Alan Hayward <alan.hayward@arm.com>
8488 David Sherwood <david.sherwood@arm.com>
8490 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
8491 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
8494 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8495 Alan Hayward <alan.hayward@arm.com>
8496 David Sherwood <david.sherwood@arm.com>
8498 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
8499 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
8500 (aarch64_evpc_dup): Generate rtl direcly, rather than using
8502 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
8503 of a single element.
8504 * config/aarch64/iterators.md: Add a comment above the permute
8505 unspecs to say that they are generated directly by
8506 aarch64_expand_vec_perm_const.
8507 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
8509 2017-11-01 Nathan Sidwell <nathan@acm.org>
8511 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
8513 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8515 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
8516 and medany code models, and describe what they do.
8518 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8520 Revert accidental duplicate:
8522 * combine.c (can_change_dest_mode): Reject changes in
8523 REGMODE_NATURAL_SIZE.
8525 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
8527 PR rtl-optimization/64682
8528 PR rtl-optimization/69567
8529 PR rtl-optimization/69737
8530 PR rtl-optimization/82683
8531 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
8532 register mentioned in the note, drop the note, unless it came from I3,
8533 in which case it should go to I3 again.
8535 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8537 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
8538 and return false if not.
8539 (clear_bytes_written_by, live_bytes_read): Update accordingly.
8541 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8543 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
8544 range is known to be empty.
8546 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8547 Alan Hayward <alan.hayward@arm.com>
8548 David Sherwood <david.sherwood@arm.com>
8550 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
8551 and CONST_VECTOR_NUNITS instead of computing the number of units from
8552 the byte sizes of the vector and element.
8553 (simplify_binary_operation_1): Likewise.
8554 (simplify_const_binary_operation): Likewise.
8555 (simplify_ternary_operation): Likewise.
8557 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8558 Alan Hayward <alan.hayward@arm.com>
8559 David Sherwood <david.sherwood@arm.com>
8561 * var-tracking.c (INT_MEM_OFFSET): Replace with...
8562 (int_mem_offset): ...this new function.
8563 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
8564 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
8565 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
8568 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8569 Alan Hayward <alan.hayward@arm.com>
8570 David Sherwood <david.sherwood@arm.com>
8572 * lower-subreg.c (interesting_mode_p): New function.
8573 (compute_costs, find_decomposable_subregs, decompose_register)
8574 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
8575 (resolve_clobber, dump_choices): Use it.
8577 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8578 Alan Hayward <alan.hayward@arm.com>
8579 David Sherwood <david.sherwood@arm.com>
8581 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
8583 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8584 Alan Hayward <alan.hayward@arm.com>
8585 David Sherwood <david.sherwood@arm.com>
8587 * alias.c (find_base_value, find_base_term): Only process integer
8588 truncations. Check the precision rather than the size.
8590 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8591 Alan Hayward <alan.hayward@arm.com>
8592 David Sherwood <david.sherwood@arm.com>
8594 * machmode.h (is_narrower_int_mode): New function
8595 * optabs.c (expand_float, expand_fix): Use it.
8596 * dwarf2out.c (rotate_loc_descriptor): Likewise.
8598 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8599 Alan Hayward <alan.hayward@arm.com>
8600 David Sherwood <david.sherwood@arm.com>
8602 * rtl.h (narrower_subreg_mode): New function.
8603 * ira-color.c (update_costs_from_allocno): Use it.
8605 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8606 Alan Hayward <alan.hayward@arm.com>
8607 David Sherwood <david.sherwood@arm.com>
8609 * optabs-query.h (convert_optab_p): New function, split out from...
8610 (convert_optab_handler): ...here.
8611 (widening_optab_handler): Delete.
8612 (find_widening_optab_handler): Remove permit_non_widening parameter.
8613 (find_widening_optab_handler_and_mode): Likewise. Provide an
8614 override that operates on mode class wrappers.
8615 * optabs-query.c (widening_optab_handler): Delete.
8616 (find_widening_optab_handler_and_mode): Remove permit_non_widening
8617 parameter. Assert that the two modes are the same class and that
8618 the "from" mode is narrower than the "to" mode. Use
8619 convert_optab_handler instead of widening_optab_handler.
8620 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
8621 instead of widening_optab_handler.
8622 * expr.c (expand_expr_real_2): Update calls to
8623 find_widening_optab_handler.
8624 * optabs.c (expand_widen_pattern_expr): Likewise.
8625 (expand_binop_directly): Take the insn_code as a parameter.
8626 (expand_binop): Only call find_widening_optab_handler for
8627 conversion optabs; use optab_handler otherwise. Update calls
8628 to find_widening_optab_handler and expand_binop_directly.
8629 Use convert_optab_handler instead of widening_optab_handler.
8630 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
8631 find_widening_optab_handler and use scalar_mode rather than
8633 (convert_plusminus_to_widen): Likewise.
8635 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8636 Alan Hayward <alan.hayward@arm.com>
8637 David Sherwood <david.sherwood@arm.com>
8639 * machmode.h (fixed_size_mode): New class.
8640 * rtl.h (get_pool_mode): Return fixed_size_mode.
8641 * gengtype.c (main): Add fixed_size_mode.
8642 * target.def (get_raw_result_mode): Return a fixed_size_mode.
8643 (get_raw_arg_mode): Likewise.
8644 * doc/tm.texi: Regenerate.
8645 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
8646 * targhooks.c (default_get_reg_raw_mode): Likewise.
8647 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
8648 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
8649 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
8650 (msp430_get_raw_result_mode): Likewise.
8651 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
8652 * dbxout.c (dbxout_parms): Require fixed-size modes.
8653 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
8654 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8655 * omp-low.c (lower_oacc_reductions): Likewise.
8656 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
8657 (simplify_subreg): Update accordingly.
8658 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
8659 (force_const_mem): Update accordingly. Return NULL_RTX for modes
8660 that aren't fixed-size.
8661 (get_pool_mode): Return a fixed_size_mode.
8662 (output_constant_pool_2): Take a fixed_size_mode.
8664 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8665 Alan Hayward <alan.hayward@arm.com>
8666 David Sherwood <david.sherwood@arm.com>
8668 * doc/rtl.texi (vec_series): Document.
8669 (const): Say that the operand can be a vec_series.
8670 * rtl.def (VEC_SERIES): New rtx code.
8671 * rtl.h (const_vec_series_p_1): Declare.
8672 (const_vec_series_p): New function.
8673 * emit-rtl.h (gen_const_vec_series): Declare.
8674 (gen_vec_series): Likewise.
8675 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
8676 (gen_vec_series): Likewise.
8677 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
8678 * simplify-rtx.c (simplify_unary_operation): Handle negations
8680 (simplify_binary_operation_series): New function.
8681 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
8682 (test_vector_ops_series): New function.
8683 (test_vector_ops): Call it.
8684 * config/powerpcspe/altivec.md (altivec_lvsl): Use
8685 gen_const_vec_series.
8686 (altivec_lvsr): Likewise.
8687 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
8689 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8690 Alan Hayward <alan.hayward@arm.com>
8691 David Sherwood <david.sherwood@arm.com>
8693 * doc/rtl.texi (const): Update description of address constants.
8694 Say that vector constants are allowed too.
8695 * common.md (E, F): Use CONSTANT_P instead of checking for
8697 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
8698 checking for CONST_VECTOR.
8699 * expmed.c (make_tree): Use build_vector_from_val for a CONST
8701 * expr.c (expand_expr_real_2): Check for vector modes instead
8702 of checking for CONST_VECTOR.
8703 * rtl.h (const_vec_p): New function.
8704 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
8705 (unwrap_const_vec_duplicate): Handle them here too.
8707 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8708 David Malcolm <dmalcolm@redhat.com>
8709 Alan Hayward <alan.hayward@arm.com>
8710 David Sherwood <david.sherwood@arm.com>
8712 * rtl.h (vec_duplicate_p): New function.
8713 * selftest-rtl.c (assert_rtx_eq_at): New function.
8714 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
8715 (assert_rtx_eq_at): Declare.
8716 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
8717 * selftest-run-tests.c (selftest::run_tests): Call it.
8718 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
8719 (simplify_unary_operation_1): Recursively handle vector duplicates.
8720 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
8722 (simplify_subreg): Handle subregs of vector duplicates.
8723 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
8724 (selftest::simplify_rtx_c_tests): New functions.
8726 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8727 Alan Hayward <alan.hayward@arm.com>
8728 David Sherwood <david.sherwood@arm.com>
8730 * emit-rtl.h (gen_const_vec_duplicate): Declare.
8731 (gen_vec_duplicate): Likewise.
8732 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
8734 (gen_const_vector): ...here.
8735 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
8736 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
8737 whose elements are all equal.
8738 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
8739 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
8740 (simplify_relational_operation): Likewise.
8741 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8743 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
8744 (aarch64_expand_vector_init): Likewise.
8745 * config/arm/arm.c (neon_vdup_constant): Likewise.
8746 (neon_expand_vector_init): Likewise.
8747 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
8748 (arm_block_set_unaligned_vect): Likewise.
8749 (arm_block_set_aligned_vect): Likewise.
8750 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
8751 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
8752 (expand_vec_perm_even_odd_pack): Likewise.
8753 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
8754 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
8755 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
8756 gen_const_vec_duplicate.
8757 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
8758 * config/mips/mips.c (mips_gen_const_int_vector): Use
8759 gen_const_vec_duplicate.
8760 (mips_expand_vector_init): Use CONST0_RTX.
8761 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
8762 (define_split): Use gen_const_vec_duplicate.
8763 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
8764 (define_split): Use gen_const_vec_duplicate.
8765 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
8766 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
8767 * config/spu/spu.c (spu_const): Likewise.
8769 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8770 Alan Hayward <alan.hayward@arm.com>
8771 David Sherwood <david.sherwood@arm.com>
8773 * combine.c (can_change_dest_mode): Reject changes in
8774 REGMODE_NATURAL_SIZE.
8776 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8778 * configure.ac (--enable-libssp): New.
8779 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8780 trying to determine target-specific default. Adjust indentation.
8781 * configure: Regenerated.
8782 * doc/install.texi (Configuration): Expand --disable-libssp
8785 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8787 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8790 2017-10-31 Martin Jambor <mjambor@suse.cz>
8793 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8795 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8797 * auto-profile.c (autofdo_source_profile::read): Use
8798 UNKNOWN_LOCATION rather than 0.
8799 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8800 (warning_at): ...this overload.
8801 (warning_at_rich_loc_n): Rename to...
8802 (warning_n): ...this overload.
8803 (error_at_rich_loc): Rename to...
8804 (error_at): ...this overload.
8805 (pedwarn_at_rich_loc): Rename to...
8806 (pedwarn): ...this overload.
8807 (permerror_at_rich_loc): Rename to...
8808 (permerror): ...this overload.
8809 (inform_at_rich_loc): Rename to...
8810 (inform): ...this overload.
8811 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8812 (diagnostic_n_impl_richloc): Rename to...
8813 (diagnostic_n_impl): ...this rich_location *-based decl.
8814 (inform_at_rich_loc): Rename to...
8815 (inform): ...this, and add an assertion.
8816 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8817 (warning_at_rich_loc): Rename to...
8818 (warning_at): ...this, and add an assertion.
8819 (warning_at_rich_loc_n): Rename to...
8820 (warning_n): ...this, and add an assertion.
8821 (warning_n): Update location_t-based implementation for removal of
8822 location_t-based diagnostic_n_impl.
8823 (pedwarn_at_rich_loc): Rename to...
8824 (pedwarn): ...this, and add an assertion.
8825 (permerror_at_rich_loc): Rename to...
8826 (permerror): ...this, and add an assertion.
8827 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8828 (error_at_rich_loc): Rename to...
8829 (error_at): ...this, and add an assertion.
8830 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8831 (driver::do_spec_on_infiles): Likewise.
8832 * substring-locations.c (format_warning_va): Update for renaming
8833 of inform_at_rich_loc.
8835 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8837 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8838 _Float<N>X built-in functions so that the variant without the
8839 "__builtin_" prefix is only enabled for the GNU C and Objective C
8840 languages when they are in non-strict ANSI/ISO mode.
8841 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8842 * target.def (floatn_builtin_p): Add a target hook to control
8843 whether _Float<N> and _Float<N>X built-in functions without the
8844 "__builtin_" prefix are enabled, and return true for C and
8845 Objective C in the default hook. Include langhooks.h in
8847 * targhooks.h (default_floatn_builtin_p): Likewise.
8848 * targhooks.c (default_floatn_builtin_p): Likewise.
8849 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8850 floatn_builtin_p target hook.
8851 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8853 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8854 Eric Botcazou <ebotcazou@adacore.com>
8856 PR rtl-optimization/81803
8857 * lra-constraints.c (curr_insn_transform): Also reload the whole
8858 register for a strict subreg no wider than a word if this is for
8859 a WORD_REGISTER_OPERATIONS target.
8861 2017-10-31 Jason Merrill <jason@redhat.com>
8863 * gdbinit.in: Skip over inlines from timevar.h.
8865 2017-10-31 Martin Liska <mliska@suse.cz>
8867 * doc/gcov.texi: Document new option.
8868 * gcov.c (print_usage): Likewise print it.
8869 (process_args): Support the argument.
8870 (format_count): New function.
8871 (format_gcov): Use the function.
8873 2017-10-31 Martin Liska <mliska@suse.cz>
8875 * gcov.c (struct name_map): do not use typedef.
8876 Define operator== and operator<.
8877 (name_search): Remove.
8878 (name_sort): Remove.
8879 (main): Do not allocate names.
8880 (process_file): Add vertical space.
8881 (generate_results): Use std::find.
8882 (release_structures): Do not release memory.
8883 (find_source): Use std::find.
8885 2017-10-31 Martin Liska <mliska@suse.cz>
8887 * gcov.c (struct line_info): Remove it's typedef.
8888 (line_info::line_info): Add proper ctor.
8889 (line_info::has_block): Do not use a typedef.
8890 (struct source_info): Do not use typedef.
8891 (circuit): Likewise.
8892 (get_cycles_count): Likewise.
8893 (output_intermediate_file): Iterate via vector iterator.
8894 (add_line_counts): Use std::vector methods.
8895 (accumulate_line_counts): Likewise.
8896 (output_lines): Likewise.
8898 2017-10-31 Martin Liska <mliska@suse.cz>
8900 * gcov.c (struct source_info): Remove typedef.
8901 (source_info::source_info): Add proper ctor.
8902 (accumulate_line_counts): Use struct, not it's typedef.
8903 (output_gcov_file): Likewise.
8904 (output_lines): Likewise.
8905 (main): Do not allocate an array.
8906 (output_intermediate_file): Use size of vector container.
8907 (process_file): Resize the vector.
8908 (generate_results): Do not preallocate, use newly added vector
8910 (release_structures): Do not release sources.
8911 (find_source): Use vector methods.
8912 (add_line_counts): Do not use typedef.
8914 2017-10-31 Martin Liska <mliska@suse.cz>
8916 * doc/gcov.texi: Document that.
8917 * gcov.c (add_line_counts): Mark lines with a non-executed
8919 (output_line_beginning): Handle such lines.
8920 (output_lines): Pass new argument.
8921 (output_intermediate_file): Print it in intermediate format.
8923 2017-10-31 Martin Liska <mliska@suse.cz>
8925 * color-macros.h: New file.
8926 * diagnostic-color.c: Factor out color related to macros to
8928 * doc/gcov.texi: Document -k option.
8929 * gcov.c (INCLUDE_STRING): Include string.h.
8930 (print_usage): Add -k option.
8931 (process_args): Parse it.
8932 (pad_count_string): New function.
8933 (output_line_beginning): Likewise.
8934 (DEFAULT_LINE_START): New macro.
8935 (output_lines): Support color output.
8937 2017-10-31 Martin Liska <mliska@suse.cz>
8939 PR gcov-profile/82633
8940 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8941 their interaction with GCOV infrastructure.
8942 * configure.ac: Add -fkeep-{inline,static}-functions to
8944 * configure: Regenerate.
8946 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8949 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8951 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8954 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8955 into a register if it does not fit into an immediate offset field.
8957 2017-10-31 Olivier Hainque <hainque@adacore.com>
8959 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8961 2017-10-31 Julia Koval <julia.koval@intel.com>
8963 * config.gcc: Add gfniintrin.h.
8964 * config/i386/gfniintrin.h: New.
8965 * config/i386/i386-builtin-types.def
8966 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8967 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8968 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8969 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8970 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8971 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8972 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8973 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8974 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8975 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8976 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8977 * config/i386/immintrin.h: Include gfniintrin.h.
8978 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8980 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8982 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8984 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8986 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8987 (arm_ashldi3_1bit): Remove pattern.
8988 (ashrdi3): Remove shift by 1 expansion.
8989 (arm_ashrdi3_1bit): Remove pattern.
8990 (lshrdi3): Remove shift by 1 expansion.
8991 (arm_lshrdi3_1bit): Remove pattern.
8992 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8993 cost of ashldi3 by 1.
8994 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8995 (<shift>di3_neon): Likewise.
8997 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8999 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9000 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9001 and (*aarch64_simd_mov<VQ:mode>).
9002 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9003 pattern alternative.
9004 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9005 attributes to match pattern alternative.
9007 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9009 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9010 * config/rs6000/emmintrin.h: New file.
9011 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9013 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9015 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9016 * (movdi_vfp_cortexa8): Remove pattern.
9018 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9020 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9022 (Specific, alpha*-dec-osf5.1): Remove.
9023 (Specific, mips-sgi-irix5): Remove.
9024 (Specific, mips-sgi-irix6): Remove.
9026 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9029 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9030 arguments to clear_bit_region_be.
9032 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9034 * gimplify.c: Include memmodel.h.
9036 2017-10-30 Martin Jambor <mjambor@suse.cz>
9038 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9039 condition whether loop should be executed at all.
9041 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9043 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9044 gimple folding of vec_madd() intrinsics.
9045 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9046 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9047 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9049 2017-10-30 Richard Biener <rguenther@suse.de>
9051 PR tree-optimization/82762
9053 2017-10-23 Richard Biener <rguenther@suse.de>
9055 PR tree-optimization/82129
9057 2017-08-01 Richard Biener <rguenther@suse.de>
9059 PR tree-optimization/81181
9060 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9061 (compute_antic): ... end of iteration here.
9063 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9065 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9066 -std=iso9899:2017 and -std=gnu17.
9067 * doc/standards.texi (C Language): Document C17 support.
9068 * doc/cpp.texi (Overview): Mention -std=c17.
9069 (Standard Predefined Macros): Document C11 and C17 values of
9070 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9071 * doc/extend.texi (Inline): Do not list individual options for
9072 standards newer than C99.
9073 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9075 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9078 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9080 * asan.c (asan_finish_file): Align asan globals array by shadow
9083 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9086 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9087 (struct store_immediate_info): Add bitregion_start and bitregion_end
9089 (store_immediate_info::store_immediate_info): Add brs and bre
9090 arguments and initialize bitregion_{start,end} from those.
9091 (struct merged_store_group): Add bitregion_start, bitregion_end,
9092 align_base and mask fields. Drop unnecessary struct keyword from
9093 struct store_immediate_info. Add do_merge method.
9094 (clear_bit_region_be): Use memset instead of loop storing zeros.
9095 (merged_store_group::do_merge): New method.
9096 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9097 stores as long as the surrounding bitregions have no gaps.
9098 (merged_store_group::merge_overlapping): Use do_merge.
9099 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9100 is byte aligned, rather than requiring that start and width are
9101 byte aligned. Drop unnecessary struct keyword from
9102 struct store_immediate_info. Allocate and populate also mask array.
9103 Make start of the arrays relative to bitregion_start rather than
9104 start and size them according to bitregion_{end,start} difference.
9105 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9106 struct store_immediate_info.
9107 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9108 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9109 struct keyword from struct store_immediate_info.
9110 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9111 between stores as long as the surrounding bitregions have no gaps.
9113 (struct split_store): Add orig non-static data member.
9114 (split_store::split_store): Initialize orig to false.
9115 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9116 if there is exactly a single original stmt. Change stmts argument
9117 to pointer from reference, if NULL, don't push anything to it. Add
9118 first argument, use it to optimize skipping over orig stmts that
9119 are known to be before bitpos already. Simplify.
9120 (split_group): Return unsigned int count how many stores are or
9121 would be needed rather than a bool. Add allow_unaligned argument.
9122 Change split_stores argument from reference to pointer, if NULL,
9123 only do a dry run computing how many stores would be produced.
9124 Rewritten algorithm to use both alignment and misalign if
9125 !allow_unaligned and handle bitfield stores with gaps.
9126 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9127 from bitregion_start instead of start. Compute allow_unaligned
9128 here, if true, do 2 split_group dry runs to compute which one
9129 produces fewer stores and prefer aligned if equal. Punt if
9130 new count is bigger or equal than original before emitting any
9131 statements, rather than during that. Remove no longer needed
9132 new_ssa_names tracking. Replace num_stmts with
9133 split_stores.length (). Use 32-bit stack allocated entries
9134 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9135 if possible. Handle bitfields with gaps.
9136 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9137 Compute bitregion_{start,end} for the stores and construct
9138 store_immediate_info with that. Formatting fixes.
9140 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9143 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9144 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9146 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9148 * gimplify.c: Include tm_p.h.
9150 * common.opt (gcoff): Re-add as ignored option.
9151 (gcoff1, gcoff2, gcoff3): Likewise.
9153 * Makefile.in (OBJS): Delete sdbout.o.
9154 (GTFILES): Delete $(srcdir)/sdbout.c.
9155 * debug.h: Delete sdb_debug_hooks.
9156 * final.c: Delete sdbout.h include.
9157 (final_scan_insn): Delete SDB_DEBUG check.
9158 (rest_of_clean_state): Likewise.
9159 * output.h: Delete sdb_begin_function_line.
9162 * toplev.c: Delete sdbout.h include.
9163 (process_options): Delete SDB_DEBUG check.
9164 * tree-core.h (tree_type_common): Delete pointer field of
9166 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9167 TYPE_SYMTAB_POINTER.
9168 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9169 (TYPE_SYMTAB_IS_POINTER): Delete.
9170 (TYPE_SYMTAB_IS_DIE): Renumber.
9171 * xcoffout.c: Refer to former sdbout.c file.
9172 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9174 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9175 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9176 (Debugging Options): Delete -gcoff.
9177 (-gstabs): Delete SDB reference.
9179 (-gcoff@var{level}): Delete.
9180 * doc/passes.texi (Debugging information output): Delete SDB and
9181 sdbout.c references.
9182 * doc/tm.texi: Regenerate.
9183 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9184 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9186 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9187 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9188 (SDB_DEBUGGING_INFO): Delete.
9189 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9190 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9191 * target.def (output_source_filename): Delete COFF reference.
9193 * common.opt (gcoff): Delete.
9194 (gxcoff+): Update Negative chain.
9195 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9197 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9198 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9199 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9201 (expand_function_start): Change sdb reference to past tense.
9202 (expand_function_end): Change sdb reference to past tense.
9203 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9204 * opts.c (debug_type_names): Delete coff entry.
9205 (common_handle_option): Delete OPT_gcoff case.
9206 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9208 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9209 * config/cris/cris.h: Delete SDB reference in comment.
9210 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9211 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9212 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9213 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9215 (ix86_expand_prologue): Likewise.
9216 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9217 * config/ia64/ia64.h: Likewise.
9218 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9219 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9221 * config/mmix/mmix.h: Likewise.
9222 * config/nds32/nds32.c: Likewise.
9223 * config/stormy/storym16.h: Likewise.
9224 * config/visium/visium.h: Likewise.
9225 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9227 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9229 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9230 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9231 FRAME_POINTER_REGNUM point at high end of local var area.
9233 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9235 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9236 Move comment around. Do not reset best_edge for a copiable
9237 destination if the copy would cause a partition change.
9238 (better_edge_p): Remove redundant check.
9240 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9242 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9244 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9246 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9247 for math functions that have _Float<N> and _Float<N>X variants.
9248 (mathfn_built_in_2): Add support for math functions that have
9249 _Float<N> and _Float<N>X variants.
9250 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9251 (expand_builtin_mathfn_ternary): Add support for fma with
9252 _Float<N> and _Float<N>X variants.
9253 (expand_builtin): Likewise.
9254 (fold_builtin_3): Likewise.
9255 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9256 create math function _Float<N> and _Float<N>X variants as external
9258 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9259 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9260 the __builtin_ prefix and if not strict ansi, without the prefix.
9261 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9262 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9263 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9264 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9265 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9266 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9267 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9268 function signatures for fma _Float<N> and _Float<N>X variants.
9269 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9270 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9271 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9272 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9273 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9274 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9275 * gencfn-macros.c (print_case_cfn): Add support for math functions
9276 that have _Float<N> and _Float<N>X variants.
9277 (print_define_operator_list): Likewise.
9278 (fltfn_suffixes): Likewise.
9280 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9281 for math functions that have _Float<N> and _Float<N>X variants.
9282 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9283 and _Float<N>X variants.
9284 (COPYSIGN): Likewise.
9287 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9288 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9290 (integer_valued_read_call_p): Likewise.
9291 * fold-const-call.c (fold_const_call_ss): Likewise.
9292 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9293 _Float<N> and _Float<N>X variants.
9294 (fold_const_call_ssss): Add support for fma _Float<N> and
9295 _Float<N>X variants.
9296 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9297 support for copysign and fma _Float<N> and _Float<N>X variants.
9298 (backprop::process_builtin_call_use): Likewise.
9299 * tree-call-cdce.c (can_test_argument_range); Add support for
9300 sqrt _Float<N> and _Float<N>X variants.
9301 (edom_only_function): Likewise.
9302 (get_no_error_domain): Likewise.
9303 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9304 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9305 copysign _Float<N> and _Float<N>X variants.
9306 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9307 handled by machine independent code.
9308 (FMAF128): Likewise.
9309 * doc/cpp.texi (Common Predefined Macros): Document defining
9310 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9311 fma _Float<N> and _Float<N>X variants.
9313 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9316 * config/i386/i386-modes.def (CCFPU): Remove definition.
9317 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9318 (ix86_cc_modes_compatible): Ditto.
9319 (ix86_expand_carry_flag_compare): Ditto.
9320 (ix86_expand_int_movcc): Ditto.
9321 (ix86_expand_int_addcc): Ditto.
9322 (ix86_reverse_condition): Ditto.
9323 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9324 Return true/false for unordered/ordered fp comparisons.
9325 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9326 (ix86_prepare_fp_compare_args): Update for rename.
9327 (ix86_expand_fp_compare): Update for rename. Generate unordered
9328 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9329 (ix86_expand_sse_compare_and_jump): Ditto.
9330 * config/i386/predicates.md (fcmov_comparison_operator):
9331 Remove CCFPU mode handling.
9332 (ix86_comparison_operator): Ditto.
9333 (ix86_carry_flag_operator): Ditto.
9334 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9335 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9336 (*cmpu<mode>_cc_i387): Ditto.
9337 (FPCMP): Remove mode iterator.
9338 (unord): Remove mode attribute.
9339 (unord_subst): New define_subst transformation
9340 (unord): New define_subst attribute.
9342 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9343 (*cmpi<unord>xf_i387): Ditto.
9344 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9345 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9346 using unord_subst transformation.
9347 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9348 (round_saeonly): Also handle CCFP mode.
9349 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9350 Remove UNSPEC_SAHF unspec handling.
9352 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9354 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9356 2017-10-27 Jeff Law <law@redhat.com>
9358 * gimple-ssa-sprintf.c: Include domwalk.h.
9359 (class sprintf_dom_walker): New class, derived from dom_walker.
9360 (sprintf_dom_walker::before_dom_children): New function.
9361 (struct call_info): Moved into sprintf_dom_walker class
9362 (compute_formath_length, handle_gimple_call): Likewise.
9363 (sprintf_length::execute): Call the dominator walker rather
9364 than walking the statements.
9366 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9367 gimple statement locations.
9368 (check_array_bounds): Corresponding changes. Get the statement's
9369 location directly from wi->stmt.
9371 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9374 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9376 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9378 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9379 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9381 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9384 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9385 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9386 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9387 maybe_get_pool_constant.
9388 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9391 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9393 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9394 2.26 caveat. Update gas and gld versions.
9395 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9398 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9400 * cgraph.h (set_malloc_flag): Declare.
9401 * cgraph.c (set_malloc_flag_1): New function.
9402 (set_malloc_flag): Likewise.
9403 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9404 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9406 (read_ipa_call_summary): Add support for reading is_return_callee.
9407 (write_ipa_call_summary): Stream is_return_callee.
9408 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9409 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9410 ipa-prop.h, ipa-fnsummary.h.
9411 (pure_const_names): Change to static.
9412 (malloc_state_e): Define.
9413 (malloc_state_names): Define.
9414 (funct_state_d): Add field malloc_state.
9415 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9416 (check_retval_uses): New function.
9417 (malloc_candidate_p): Likewise.
9418 (analyze_function): Add support for malloc attribute.
9419 (pure_const_write_summary): Stream malloc_state.
9420 (pure_const_read_summary): Add support for reading malloc_state.
9421 (dump_malloc_lattice): New function.
9422 (propagate_malloc): New function.
9423 (warn_function_malloc): New function.
9424 (ipa_pure_const::execute): Call propagate_malloc and
9425 ipa_free_fn_summary.
9426 (pass_local_pure_const::execute): Add support for malloc attribute.
9427 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9428 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9430 2017-10-27 Martin Liska <mliska@suse.cz>
9432 PR gcov-profile/82457
9433 * doc/invoke.texi: Document that one needs a non-strict ISO mode
9434 for fork-like functions to be properly instrumented.
9436 2017-10-27 Richard Biener <rguenther@suse.de>
9439 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
9440 info when we redirected EH.
9442 2017-10-26 Michael Collison <michael.collison@arm.com>
9444 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
9446 (<optab>_trunchf<GPI:mode>2: New pattern.
9447 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
9448 * config/aarch64/iterators.md (wv): New mode attribute.
9449 (vf, VF): New mode attributes.
9450 (vgp, VGP): New mode attributes.
9451 (s): Update attribute with SImode and DImode prefixes.
9453 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9455 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
9456 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
9457 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
9458 (nios2_option_overide): Initialize it. Don't allow R0-relative
9459 addressing with PIC.
9460 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
9461 (nios2_symbolic_constant_p): Likewise.
9462 (nios2_legitimate_address_p): Likewise.
9463 (nios2_r0rel_section_name_p): New.
9464 (nios2_symbol_ref_in_r0rel_data_p): New.
9465 (nios2_emit_move_sequence): Handle r0rel_constant_p.
9466 (r0rel_constant_p): New.
9467 (nios2_print_operand_address): Handle r0rel_constant_p.
9468 (nios2_cdx_narrow_form_p): Likewise.
9469 * config/nios2/nios2.opt (mr0rel-sec=): New option.
9470 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
9471 (Nios II Options): Document -mr0rel-sec.
9473 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
9475 * config/nios2/nios2.c: Include xregex.h.
9476 (nios2_gprel_sec_regex): New.
9477 (nios2_option_overide): Initialize it. Don't allow GP-relative
9478 addressing with PIC.
9479 (nios2_small_section_name_p): Check for regex match.
9480 * config/nios2/nios2.opt (mgprel-sec=): New option.
9481 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
9482 (Nios II Options): Document -mgprel-sec.
9484 2017-10-26 Jim Wilson <wilson@tuliptree.org>
9486 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
9488 2017-10-26 Tom de Vries <tom@codesourcery.com>
9490 PR tree-optimization/82707
9491 * gimple.c (gimple_copy): Fix unsharing of
9492 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
9494 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
9496 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
9497 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
9498 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
9499 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
9500 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
9501 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
9502 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
9503 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
9504 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
9505 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
9506 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
9507 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
9508 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
9509 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
9510 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
9511 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
9512 _mm512_mask_cmpunord_ps_mask): New intrinsics.
9514 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9516 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
9518 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
9519 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
9520 warning to rs6000.c. Remove the Undocumented flag, since it has
9522 (-mabi=ibmlongdouble): Likewise.
9523 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
9524 already set, set the default format for long double.
9525 (rs6000_debug_reg_global): Print whether long double is IBM or
9527 (rs6000_option_override_internal): Rework setting long double
9528 format. Only warn if the user is changing the long double default
9529 and they did not use -Wno-psabi.
9530 * doc/invoke.texi (PowerPC options): Update the documentation for
9531 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
9533 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9534 Alan Hayward <alan.hayward@arm.com>
9535 David Sherwood <david.sherwood@arm.com>
9537 * rtl.h (wider_subreg_mode): New function.
9538 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
9539 rather than an unsigned int *.
9540 * ira-color.c (regno_max_ref_width): Replace with...
9541 (regno_max_ref_mode): ...this new variable.
9542 (coalesced_pseudo_reg_slot_compare): Update accordingly.
9543 Use wider_subreg_mode.
9544 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
9545 rather than an unsigned int *.
9546 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
9547 (process_alt_operands): Likewise.
9548 (invariant_p): Likewise.
9549 * lra-spills.c (assign_mem_slot): Likewise.
9550 (add_pseudo_to_slot): Likewise.
9551 * lra.c (collect_non_operand_hard_regs): Likewise.
9552 (add_regs_to_insn_regno_info): Likewise.
9553 * reload1.c (regno_max_ref_width): Replace with...
9554 (regno_max_ref_mode): ...this new variable.
9555 (reload): Update accordingly. Update call to
9556 ira_sort_regnos_for_alter_reg.
9557 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
9558 (init_eliminable_invariants): Update to use regno_max_ref_mode.
9559 (scan_paradoxical_subregs): Likewise.
9561 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9563 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
9564 (aarch64_frame): Add emit_frame_chain boolean.
9565 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9566 Move eh_return case to aarch64_layout_frame.
9567 (aarch64_layout_frame): Initialize emit_frame_chain.
9568 (aarch64_expand_prologue): Use emit_frame_chain.
9570 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9572 * config/aarch64/aarch64.c (aarch64_layout_frame):
9573 Ensure LR is always stored at the bottom of the callee-saves.
9574 Remove rarely used frame layout which saves callee-saves at top of
9575 frame, so the store of LR can be used as a valid probe in all cases.
9577 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
9579 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
9580 Improve unaligned TImode/TFmode base/offset split.
9582 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9583 Alan Hayward <alan.hayward@arm.com>
9584 David Sherwood <david.sherwood@arm.com>
9586 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
9587 * combine.c (find_single_use_1): Likewise.
9588 (expand_field_assignment): Likewise.
9589 (move_deaths): Likewise.
9590 * lra-constraints.c (simplify_operand_subreg): Likewise.
9591 (curr_insn_transform): Likewise.
9592 * lra.c (collect_non_operand_hard_regs): Likewise.
9593 (add_regs_to_insn_regno_info): Likewise.
9594 * rtlanal.c (reg_referenced_p): Likewise.
9595 (covers_regno_no_parallel_p): Likewise.
9597 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9599 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
9600 Don't print any bits outside the precision of the value.
9601 * wide-int.cc (test_printing): Add some new tests.
9603 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9605 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
9606 supports -xbrace_comment option.
9607 * configure: Regenerate.
9608 * config.in: Regenerate.
9609 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
9610 (ASM_CPU_SPEC): Use it.
9612 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
9614 * target.def (static_rtx_alignment): New hook.
9615 * targhooks.h (default_static_rtx_alignment): Declare.
9616 * targhooks.c (default_static_rtx_alignment): New function.
9617 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
9618 * doc/tm.texi: Regenerate.
9619 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
9620 instead of targetm.constant_alignment. Remove call to
9622 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9623 (cris_preferred_mininum_alignment): New function, split out from...
9624 (cris_constant_alignment): ...here.
9625 (cris_static_rtx_alignment): New function.
9626 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
9628 (ix86_constant_alignment): ...here.
9629 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9630 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9631 (mmix_static_rtx_alignment): New function.
9632 * config/spu/spu.c (spu_static_rtx_alignment): New function.
9633 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
9635 2017-10-26 Tamar Christina <tamar.christina@arm.com>
9638 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
9639 Add flag_trapping_math and flag_fp_int_builtin_inexact.
9641 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
9643 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
9644 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
9645 mark as a sign-extending load.
9646 (local_pic_load_u): Define.
9648 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
9651 * fold-const.c (operand_equal_for_comparison_p): Also return true
9652 if ARG0 is a simple variant of ARG1 with narrower precision.
9653 (fold_ternary_loc): Always pass unstripped operands to the predicate.
9655 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
9657 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
9659 * i386.h (processor_costs): Add gather_static, gather_per_elt,
9660 scatter_static, scatter_per_elt.
9661 * x86-tune-costs.h: Add new cost entries.
9663 2017-10-25 Richard Biener <rguenther@suse.de>
9665 * tree-ssa-sccvn.h (vn_eliminate): Declare.
9666 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
9667 class pass_fre): Move to ...
9668 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
9669 class pass_fre): ... here and adjust for statistics.
9671 2017-10-25 Jakub Jelinek <jakub@redhat.com>
9674 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
9675 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
9676 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
9677 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
9680 2017-10-25 Richard Biener <rguenther@suse.de>
9682 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
9683 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
9684 eliminate_push_avail, eliminate_insert): Move inside...
9685 (class eliminate_dom_walker): ... this class in preparation
9687 (fini_eliminate): Remove by merging with ...
9688 (eliminate): ... this function. Adjust for class changes.
9689 (pass_pre::execute): Remove fini_eliminate call.
9690 (pass_fre::execute): Likewise.
9692 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9695 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
9696 (VPERMI2, VPERMI2I): New mode iterators.
9697 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
9698 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
9700 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
9701 mode iterator. Remove 3 old define_insn patterns.
9702 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
9703 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
9704 VPERMI2 mode iterator, remove the other two expanders.
9705 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
9706 to use VPERMI2 mode iterator, add another alternative for vpermi2*
9707 instructions, remove the other two patterns.
9708 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
9709 mode iterator, remove the other two patterns.
9710 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
9711 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
9712 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
9713 and adjust argument order accordingly.
9714 (ix86_expand_vec_perm): Adjust caller.
9715 (expand_vec_perm_1): Likewise.
9716 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
9717 (expand_vec_perm_vpermt2_vpshub2): ... this.
9718 (ix86_expand_vec_perm_const_1): Adjust caller.
9719 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
9722 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
9723 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
9724 (vec_shl_<mode>): Remove unused expander.
9725 (avx512bw_<shift_insn><mode>3): New define_insn.
9726 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
9727 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
9729 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
9732 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
9735 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9737 PR rtl-optimization/82396
9738 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
9739 (autopref_multipass_init): Simplify initialization.
9740 (autopref_rank_data): Simplify sort order.
9741 * gcc/sched-int.h (autopref_multipass_data_): Remove
9742 multi_mem_insn_p, min_offset and max_offset.
9744 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
9747 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
9748 Check special value of flag_omit_frame_pointer.
9749 (aarch64_can_eliminate): Likewise.
9750 (aarch64_override_options_after_change_1): Simplify handling of
9751 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
9753 2017-10-24 Richard Biener <rguenther@suse.de>
9755 PR tree-optimization/82697
9756 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
9757 zero for conditional load and unconditional store.
9759 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9761 * doc/install.texi: Document bootstrap-cet.
9763 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
9766 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
9767 ENDBR instruction at function entrance if function is only
9770 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9773 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9774 patterns to better describe from which operation the CF is computed.
9775 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9776 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9777 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9778 is 0, use _0 suffixed expanders instead of emitting a comparison
9781 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9783 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9784 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9786 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9789 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9790 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9791 * loop-iv.c (iv_get_reaching_def): Likewise.
9792 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9793 variable is promoted and the partition contains undefined values.
9795 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9797 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9799 (nios2_address_cost): Define.
9800 (nios2_legitimize_address): Recognize (exp + constant) directly.
9801 (TARGET_ADDRESS_COST): Define.
9803 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9805 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9806 (nios2_symbolic_memory_operand_p): Declare.
9807 (nios2_split_large_constant): Declare.
9808 (nios2_split_symbolic_memory_operand): Declare.
9809 * config/nios2/nios2.c: Adjust includes.
9810 (nios2_symbolic_constant_allowed): New.
9811 (nios2_symbolic_constant_p): New.
9812 (nios2_plus_symbolic_constant_p): New.
9813 (nios2_valid_addr_expr_p): Recognize addresses involving
9815 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9816 (nios2_symbolic_memory_operand_p): New.
9817 (nios2_large_constant_p): New.
9818 (nios2_split_large_constant): New.
9819 (nios2_split_plus_large_constant): New.
9820 (nios2_split_symbolic_memory_operand): New.
9821 (nios2_legitimize_address): Code refactoring. Handle addresses
9822 involving symbolic constants.
9823 (nios2_emit_move_sequence): Likewise.
9824 (nios2_print_operand): Improve error output.
9825 (nios2_print_operand_address): Handle LO_SUM.
9826 (nios2_cdx_narrow_form_p): Likewise.
9827 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9828 operands involving symbolic constants.
9829 (movhi_internal, movsi_internal): Likewise.
9830 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9831 (extendhisi2, extendqi<mode>2): Likewise.
9833 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9835 * tree-pass.h (PROP_rtl_split_insns): Define.
9836 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9838 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9840 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9842 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9845 * target.def (const_not_ok_for_debug_p): Default to
9846 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9847 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9848 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9849 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9850 which targetm.const_not_ok_for_debug_p returned true.
9851 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9853 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9855 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9856 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9857 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9859 (ix86_const_not_ok_for_debug_p): New function.
9860 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9861 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9863 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9866 * system.h: Conditionally include "unique-ptr.h" if
9867 INCLUDE_UNIQUE_PTR is defined.
9868 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9869 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9871 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9873 * config/rl78/rl78.md: New define_expand "subdi3".
9875 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9878 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9879 DF_REF_INSN if DF_REF_INSN_INFO is false.
9881 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9883 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9884 xmm_move instead of sse_move.
9885 (sse_store_index): New function.
9886 (ix86_register_move_cost): Be more sensible about mismatch stall;
9887 model AVX moves correctly; make difference between sse->integer and
9889 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9890 moves; make difference between SSE and AVX.
9891 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9892 and zmm_move. Increase size of sse load and store tables;
9893 add unaligned load and store tables; add ssemmx_to_integer.
9894 * x86-tune-costs.h: Update all entries according to real
9895 move latencies from Agner Fog's manual and chip documentation.
9897 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9900 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9901 * config/i386/constraints.md (Wf): New constraint.
9902 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9903 (cmp<dwi>_doubleword): Removed.
9904 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9905 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9906 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9907 expand with cmp<dwi>_doubleword. For LTU and GEU use
9908 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9910 * common.opt (gcolumn-info): Enable by default.
9911 * doc/invoke.texi (gcolumn-info): Document new default.
9913 2017-10-23 Richard Biener <rguenther@suse.de>
9915 PR tree-optimization/82672
9916 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9917 Fold the stmt if we propagated into it.
9919 2017-10-23 Richard Biener <rguenther@suse.de>
9921 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9922 (bitmap_remove_expr_from_set): ... this. All callers call this
9923 for non-constant values.
9924 (bitmap_set_subtract): Rename to...
9925 (bitmap_set_subtract_expressions): ... this. Adjust and
9927 (bitmap_set_contains_value): Remove superfluous check.
9928 (bitmap_set_replace_value): Inline into single caller ...
9929 (bitmap_value_replace_in_set): ... here and simplify.
9930 (dependent_clean): Merge into ...
9931 (clean): ... this using an overload. Adjust.
9932 (prune_clobbered_mems): Adjust.
9933 (compute_antic_aux): Likewise.
9934 (compute_partial_antic_aux): Likewise.
9936 2017-10-23 Richard Biener <rguenther@suse.de>
9938 PR tree-optimization/82129
9940 2017-08-01 Richard Biener <rguenther@suse.de>
9942 PR tree-optimization/81181
9943 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9944 (compute_antic): ... end of iteration here.
9946 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9948 * target.def (starting_frame_offset): New hook.
9949 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9950 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9951 * doc/tm.texi.in: Regenerate.
9952 * hooks.h (hook_hwi_void_0): Declare.
9953 * hooks.c (hook_hwi_void_0): New function.
9954 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9955 STARTING_FRAME_OFFSET.
9956 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9957 * reload1.c (reload): Likewise.
9958 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9959 instead of STARTING_FRAME_OFFSET.
9960 * function.c (try_fit_stack_local): Likewise.
9961 (assign_stack_local_1): Likewise
9962 (instantiate_virtual_regs): Likewise.
9963 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9964 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9965 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9966 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9967 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9968 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9969 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9970 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9971 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9972 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9973 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9974 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9975 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9976 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9977 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9978 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9979 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9980 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9981 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9982 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9983 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9984 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9985 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9986 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9987 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9988 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9989 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9990 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9991 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9992 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9993 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9994 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9995 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9996 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9997 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9998 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9999 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10000 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10001 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10002 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10003 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10004 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10005 return a HOST_WIDE_INT.
10006 (avr_builtin_setjmp_frame_value): Use it instead of
10007 STARTING_FRAME_OFFSET.
10008 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10009 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10010 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10012 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10013 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10014 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10015 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10016 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10017 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10018 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10019 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10020 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10021 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10022 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10023 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10025 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10026 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10027 * config/mips/mips.c (mips_compute_frame_info): Refer to
10028 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10029 (mips_starting_frame_offset): New function.
10030 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10031 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10032 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10033 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10034 and return a HOST_WIDE_INT.
10035 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10036 (mmix_initial_elimination_offset): Refer to
10037 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10038 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10039 * config/pa/pa.c (pa_starting_frame_offset): New function.
10040 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10041 (pa_expand_prologue): Likewise.
10042 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10043 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10044 !FRAME_GROWS_DOWNWARD handling to...
10045 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10046 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10047 !FRAME_GROWS_DOWNWARD handling to...
10048 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10049 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10050 !FRAME_GROWS_DOWNWARD handling to...
10051 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10052 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10054 (rs6000_starting_frame_offset): New function.
10055 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10056 !FRAME_GROWS_DOWNWARD handling to...
10057 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10058 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10059 !FRAME_GROWS_DOWNWARD handling to...
10060 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10061 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10062 !FRAME_GROWS_DOWNWARD handling to...
10063 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10064 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10065 (rs6000_starting_frame_offset): New function.
10066 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10067 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10068 * config/vax/vax.c (vax_starting_frame_offset): New function.
10069 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10070 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10071 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10072 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10073 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10074 * system.h (STARTING_FRAME_OFFSET): Poison.
10076 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10078 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10079 SCALAR_TYPE_MODE instead of TYPE_MODE.
10081 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10082 Alan Hayward <alan.hayward@arm.com>
10083 David Sherwood <david.sherwood@arm.com>
10085 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10087 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10088 Alan Hayward <alan.hayward@arm.com>
10089 David Sherwood <david.sherwood@arm.com>
10091 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10093 2017-10-23 Richard Biener <rguenther@suse.de>
10095 PR tree-optimization/82129
10096 * tree-ssa-pre.c (bitmap_set_and): Remove.
10097 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10098 canonicalizing expressions in the set to those with lowest
10099 ID rather than taking that from the first edge.
10101 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10103 * combine.c (rtx_equal_for_field_assignment_p): Use
10104 byte_lowpart_offset.
10106 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10107 Alan Hayward <alan.hayward@arm.com>
10108 David Sherwood <david.sherwood@arm.com>
10110 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10111 to a SUBREG_PROMOTED_VAR.
10113 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10114 Alan Hayward <alan.hayward@arm.com>
10115 David Sherwood <david.sherwood@arm.com>
10117 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10118 (expand_debug_source_expr): Likewise.
10119 * combine.c (combine_simplify_rtx): Likewise.
10120 * cse.c (fold_rtx): Likewise.
10121 * optabs.c (expand_float): Likewise.
10122 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10123 (simplify_binary_operation_1): Likewise.
10125 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10126 Alan Hayward <alan.hayward@arm.com>
10127 David Sherwood <david.sherwood@arm.com>
10129 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10130 (record_promoted_value): Likewise.
10131 * expr.c (expand_expr_real_2): Likewise.
10132 * ree.c (update_reg_equal_equiv_notes): Likewise.
10133 (combine_set_extension): Likewise.
10134 * rtlanal.c (low_bitmask_len): Likewise.
10135 * simplify-rtx.c (neg_const_int): Likewise.
10136 (simplify_binary_operation_1): Likewise.
10138 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10139 Alan Hayward <alan.hayward@arm.com>
10140 David Sherwood <david.sherwood@arm.com>
10142 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10143 * regcprop.c (maybe_mode_change): Likewise.
10144 * reload1.c (alter_reg): Likewise.
10146 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10148 * inchash.h (inchash::hash::add_wide_int): New function.
10149 * lto-streamer-out.c (hash_tree): Use it.
10151 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10153 * inchash.h (inchash::hash::add_wide_int): Rename to...
10154 (inchash::hash::add_hwi): ...this.
10155 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10156 (polymorphic_call_target_hasher::hash): Likewise.
10157 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10158 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10159 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10160 * lto-streamer-out.c (hash_tree): Likewise.
10161 * optc-save-gen.awk: Likewise.
10162 * tree.c (add_expr): Likewise.
10164 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10167 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10168 for ordered inequality comparisons even with TARGET_IEEE_FP.
10170 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10173 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10174 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10175 Expand with cmp<dwi>_doubleword.
10177 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10179 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10180 List CET intrinsics.
10181 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10182 specific to -fcf-protection option.
10184 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10186 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10187 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10188 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10189 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10190 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10191 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10192 (extra_objs): Add cet.o for Linux/x86 targets.
10193 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10194 * config/i386/cet.c: New file.
10195 * config/i386/cetintrin.h: Likewise.
10196 * config/i386/t-cet: Likewise.
10197 * config/i386/cpuid.h (bit_SHSTK): New.
10198 (bit_IBT): Likewise.
10199 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10200 pass IBT and SHSTK bits.
10201 * config/i386/i386-builtin-types.def
10202 (VOID_FTYPE_UNSIGNED_PVOID): New.
10203 (VOID_FTYPE_UINT64_PVOID): Likewise.
10204 * config/i386/i386-builtin.def: Add CET intrinsics.
10205 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10206 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10207 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10208 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10210 * config/i386/i386.c (rest_of_insert_endbranch): New.
10211 (pass_data_insert_endbranch): Likewise.
10212 (pass_insert_endbranch): Likewise.
10213 (make_pass_insert_endbranch): Likewise.
10214 (ix86_notrack_prefixed_insn_p): Likewise.
10215 (ix86_target_string): Add -mibt, -mshstk flags.
10216 (ix86_option_override_internal): Add flag_cf_protection
10218 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10219 (ix86_print_operand): Add 'notrack' prefix output.
10220 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10221 (ix86_expand_builtin): Expand CET intrinsics.
10222 (x86_output_mi_thunk): Add 'endbranch' instruction.
10223 * config/i386/i386.h (TARGET_IBT): New.
10224 (TARGET_IBT_P): Likewise.
10225 (TARGET_SHSTK): Likewise.
10226 (TARGET_SHSTK_P): Likewise.
10227 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10228 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10229 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10230 (builtin_setjmp_setup): New pattern.
10231 (builtin_longjmp): Likewise.
10232 (rdssp<mode>): Likewise.
10233 (incssp<mode>): Likewise.
10234 (saveprevssp): Likewise.
10235 (rstorssp): Likewise.
10236 (wrss<mode>): Likewise.
10237 (wruss<mode>): Likewise.
10238 (setssbsy): Likewise.
10239 (clrssbsy): Likewise.
10240 (nop_endbr): Likewise.
10241 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10243 * config/i386/immintrin.h: Include <cetintrin.h>.
10244 * config/i386/linux-common.h
10245 (file_end_indicate_exec_stack_and_cet): New prototype.
10246 (TARGET_ASM_FILE_END): New.
10248 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10250 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10251 latencies instead of having separate table; make difference between
10252 integer and float costs.
10253 * i386.h (processor_costs): Remove scalar_stmt_cost,
10254 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10255 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10257 * x86-tune-costs.h: Remove entries which has been removed in
10258 procesor_costs from all tables; make cond_taken_branch_cost
10259 and cond_not_taken_branch_cost COST_N_INSNS based.
10261 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10263 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10265 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10267 * config/i386/i386.md (isa): Remove fma_avx512f.
10268 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10269 <avx512>_fmadd_<mode>_mask3<round_name>,
10270 <avx512>_fmsub_<mode>_mask<round_name>,
10271 <avx512>_fmsub_<mode>_mask3<round_name>,
10272 <avx512>_fnmadd_<mode>_mask<round_name>,
10273 <avx512>_fnmadd_<mode>_mask3<round_name>,
10274 <avx512>_fnmsub_<mode>_mask<round_name>,
10275 <avx512>_fnmsub_<mode>_mask3<round_name>,
10276 <avx512>_fmaddsub_<mode>_mask<round_name>,
10277 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10278 <avx512>_fmsubadd_<mode>_mask<round_name>,
10279 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10280 (*vec_widen_umult_even_v16si<mask_name>,
10281 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10282 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10284 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10286 * extend.texi: Add 'nocf_check' documentation.
10287 * gimple.texi: Add second parameter to
10288 gimple_build_call_from_tree.
10289 * invoke.texi: Add -fcf-protection documentation.
10290 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10292 2017-10-20 Richard Biener <rguenther@suse.de>
10294 PR tree-optimization/82473
10295 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10296 the largest input type.
10298 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10300 * c-attribs.c (handle_nocf_check_attribute): New function.
10301 (c_common_attribute_table): Add 'nocf_check' handling.
10302 * gimple-parser.c: Add second argument NULL to
10303 gimple_build_call_from_tree.
10304 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10305 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10307 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10308 * common.opt: Add fcf-protection flag.
10309 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10310 * flag-types.h: Add enum cf_protection_level.
10311 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10312 Add 'nocf_check' attribute propagation to gimple call.
10313 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10314 (gimple_build_call_from_tree): Update prototype.
10315 (gimple_call_nocf_check_p): New function.
10316 (gimple_call_set_nocf_check): Likewise.
10317 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10318 * ipa-icf.c: Add nocf_check attribute in statement hash.
10319 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10320 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10321 * toplev.c (process_options): Add flag_cf_protection handling.
10323 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10325 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10327 2017-10-20 Richard Biener <rguenther@suse.de>
10329 PR tree-optimization/82603
10330 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10331 remove false predicated stores.
10333 2017-10-20 Richard Biener <rguenther@suse.de>
10335 * graphite-isl-ast-to-gimple.c
10336 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10337 Remove return value and simplify, dump copied stmt after lhs
10339 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10340 Reduce dump verbosity.
10341 (gsi_insert_earliest): Likewise.
10342 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10343 * graphite.c (print_global_statistics): Adjust dumping.
10344 (print_graphite_scop_statistics): Likewise.
10345 (print_graphite_statistics): Do not dump loops here.
10346 (graphite_transform_loops): But here.
10348 2017-10-20 Nicolas Roche <roche@adacore.com>
10350 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10351 * configure: Regenerate.
10353 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10356 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10357 functions when optimizing replace GIMPLE_RETURN stmts with
10358 calls to __builtin_unreachable ().
10361 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10362 for -fsanitize=thread link of executables.
10363 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10364 link of executables.
10367 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10368 New mode iterators.
10369 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10370 define_insns for logical vector shifts to use VI248_AVX512BW
10371 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10372 condition, useless isa and prefix attributes. Change the first
10373 2 of these define_insns to ...
10374 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10375 define_insn for avx512vl.
10376 (<shift_insn><mode>3): ... and this, new define_insn without
10377 masking for non-avx512vl.
10380 * config/i386/sse.md (*andnot<mode>3,
10381 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10382 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10383 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10384 not applied use empty suffix even for TARGET_AVX512VL.
10385 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10386 is applied, supply evex,evex or evex,evex,evex instead of just
10389 2017-10-20 Julia Koval <julia.koval@intel.com>
10391 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10392 (OPTION_MASK_ISA_GFNI_UNSET): New.
10393 (ix86_handle_option): Handle OPT_mgfni.
10394 * config/i386/cpuid.h (bit_GFNI): New.
10395 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10396 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10397 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10398 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10399 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10400 * config/i386/i386.opt: Add mgfni.
10402 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10404 * config/msp430/msp430.c (msp430_option_override): Disable
10405 -fdelete-null-pointer-checks.
10406 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10408 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10410 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10411 of x87 and SSE instructions.
10413 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10415 * asan.c (create_cond_insert_point): Do not update edge count.
10416 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10417 (afdo_propagate_circuit): Likewise.
10418 (afdo_calculate_branch_prob): Likewise.
10419 (afdo_annotate_cfg): Likewise.
10420 * basic-block.h (struct edge_def): Remove count.
10421 (edge_def::count): New accessor.
10422 * bb-reorder.c (rotate_loop): Update.
10423 (find_traces_1_round): Update.
10424 (connect_traces): Update.
10425 (sanitize_hot_paths): Update.
10426 * cfg.c (unchecked_make_edge): Update.
10427 (make_single_succ_edge): Update.
10428 (check_bb_profile): Update.
10429 (dump_edge_info): Update.
10430 (update_bb_profile_for_threading): Update.
10431 (scale_bbs_frequencies_int): Update.
10432 (scale_bbs_frequencies_gcov_type): Update.
10433 (scale_bbs_frequencies_profile_count): Update.
10434 (scale_bbs_frequencies): Update.
10435 * cfganal.c (connect_infinite_loops_to_exit): Update.
10436 * cfgbuild.c (compute_outgoing_frequencies): Update.
10437 (find_many_sub_basic_blocks): Update.
10438 * cfgcleanup.c (try_forward_edges): Update.
10439 (try_crossjump_to_edge): Update
10440 * cfgexpand.c (expand_gimple_cond): Update
10441 (expand_gimple_tailcall): Update
10442 (construct_exit_block): Update
10443 * cfghooks.c (verify_flow_info): Update
10444 (redirect_edge_succ_nodup): Update
10445 (split_edge): Update
10446 (make_forwarder_block): Update
10447 (duplicate_block): Update
10448 (account_profile_record): Update
10449 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10450 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10451 * cfgloopmanip.c (scale_loop_profile): Update.
10453 (lv_adjust_loop_entry_edge): Update.
10454 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10455 (force_nonfallthru_and_redirect): Update.
10456 (purge_dead_edges): Update.
10457 (rtl_flow_call_edges_add): Update.
10458 * cgraphunit.c (init_lowered_empty_function): Update.
10459 (cgraph_node::expand_thunk): Update.
10460 * gimple-pretty-print.c (dump_probability): Update.
10461 (dump_edge_probability): Update.
10462 * gimple-ssa-isolate-paths.c (isolate_path): Update.
10463 * haifa-sched.c (sched_create_recovery_edges): Update.
10464 * hsa-gen.c (convert_switch_statements): Update.
10465 * ifcvt.c (dead_or_predicable): Update.
10466 * ipa-inline-transform.c (inline_transform): Update.
10467 * ipa-split.c (split_function): Update.
10468 * ipa-utils.c (ipa_merge_profiles): Update.
10469 * loop-doloop.c (add_test): Update.
10470 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
10471 * lto-streamer-in.c (input_cfg): Update.
10472 (input_function): Update.
10473 * lto-streamer-out.c (output_cfg): Update.
10474 * modulo-sched.c (sms_schedule): Update.
10475 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10476 * predict.c (maybe_hot_edge_p): Update.
10477 (unlikely_executed_edge_p): Update.
10478 (probably_never_executed_edge_p): Update.
10479 (dump_prediction): Update.
10480 (drop_profile): Update.
10481 (propagate_unlikely_bbs_forward): Update.
10482 (determine_unlikely_bbs): Update.
10483 (force_edge_cold): Update.
10484 * profile.c (compute_branch_probabilities): Update.
10485 * reg-stack.c (better_edge): Update.
10486 * shrink-wrap.c (handle_simple_exit): Update.
10487 * tracer.c (better_p): Update.
10488 * trans-mem.c (expand_transaction): Update.
10489 (split_bb_make_tm_edge): Update.
10490 * tree-call-cdce.c: Update.
10491 * tree-cfg.c (gimple_find_sub_bbs): Update.
10492 (gimple_split_edge): Update.
10493 (gimple_duplicate_sese_region): Update.
10494 (gimple_duplicate_sese_tail): Update.
10495 (gimple_flow_call_edges_add): Update.
10496 (insert_cond_bb): Update.
10497 (execute_fixup_cfg): Update.
10498 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
10499 * tree-complex.c (expand_complex_div_wide): Update.
10500 * tree-eh.c (lower_resx): Update.
10501 (unsplit_eh): Update.
10502 (cleanup_empty_eh_move_lp): Update.
10503 * tree-inline.c (copy_edges_for_bb): Update.
10504 (freqs_to_counts): Update.
10505 (copy_cfg_body): Update.
10506 * tree-ssa-dce.c (remove_dead_stmt): Update.
10507 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10508 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10509 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10510 (unloop_loops): Update.
10511 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10512 * tree-ssa-loop-split.c (connect_loops): Update.
10513 (split_loop): Update.
10514 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10515 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10516 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10517 * tree-ssa-reassoc.c (branch_fixup): Update.
10518 * tree-ssa-tail-merge.c (replace_block_by): Update.
10519 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10520 (compute_path_counts): Update.
10521 (update_profile): Update.
10522 (recompute_probabilities): Update.
10523 (update_joiner_offpath_counts): Update.
10524 (estimated_freqs_path): Update.
10525 (freqs_to_counts_path): Update.
10526 (clear_counts_path): Update.
10527 (ssa_fix_duplicate_block_edges): Update.
10528 (duplicate_thread_path): Update.
10529 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10530 (case_bit_test_cmp): Update.
10531 (collect_switch_conv_info): Update.
10532 (gen_inbound_check): Update.
10533 (do_jump_if_equal): Update.
10534 (emit_cmp_and_jump_insns): Update.
10535 * tree-tailcall.c (decrease_profile): Update.
10536 (eliminate_tail_call): Update.
10537 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10538 (vect_do_peeling): Update.
10539 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10540 * ubsan.c (ubsan_expand_null_ifn): Update.
10541 (ubsan_expand_ptr_ifn): Update.
10542 * value-prof.c (gimple_divmod_fixed_value): Update.
10543 (gimple_mod_pow2): Update.
10544 (gimple_mod_subtract): Update.
10545 (gimple_ic): Update.
10546 (gimple_stringop_fixed_value): Update.
10548 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10551 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
10553 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
10555 PR rtl-optimization/82395
10556 * ira-color.c (allocno_priority_compare_func): Fix comparison step
10557 based on non_spilled_static_chain_regno_p.
10559 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
10561 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
10562 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
10563 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
10565 2017-10-19 Martin Sebor <msebor@redhat.com>
10567 PR tree-optimization/82596
10568 * tree.c (array_at_struct_end_p): Handle STRING_CST.
10570 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10572 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
10573 (get_mem_refs_of_builtin_call): Likewise.
10574 * builtins.c (expand_builtin_apply): Adjust call to
10575 allocate_dynamic_stack_space.
10576 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
10577 the third argument to allocate_dynamic_stack_space, otherwise -1.
10578 (expand_builtin): Deal with all alloca variants.
10579 (is_inexpensive_builtin): Likewise.
10580 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
10581 * calls.c (special_function_p): Deal with all alloca variants.
10582 (initialize_argument_information): Adjust call to
10583 allocate_dynamic_stack_space.
10584 (expand_call): Likewise.
10585 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
10586 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
10587 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
10588 use it for the stack usage computation.
10589 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
10590 * function.c (gimplify_parameters): Call build_alloca_call_expr.
10591 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
10592 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
10593 (in_loop_p): Remove first argument and useless check.
10594 (pass_walloca::execute): Remove useless test and adjust call to above.
10595 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
10596 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
10597 (gimplify_call_expr): Deal with all alloca variants.
10598 * hsa-gen.c (gen_hsa_alloca): Likewise.
10599 (gen_hsa_insns_for_call): Likewise.
10600 * ipa-pure-const.c (special_builtin_state): Likewise.
10601 * tree-chkp.c (chkp_build_returned_bound): Likewise.
10602 * tree-object-size.c (alloc_object_size): Likewise.
10603 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
10604 (call_may_clobber_ref_p_1): Likewise.
10605 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
10606 (ccp_fold_stmt): Likewise.
10607 (optimize_stack_restore): Likewise.
10608 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10609 (mark_all_reaching_defs_necessary_1): Likewise.
10610 (propagate_necessity): Likewise.
10611 (eliminate_unnecessary_stmts): Likewise.
10612 * tree.c (build_common_builtin_nodes): Build
10613 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
10614 (build_alloca_call_expr): New function.
10615 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
10616 (CASE_BUILT_IN_ALLOCA): Likewise.
10617 (build_alloca_call_expr): Declare.
10618 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
10620 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
10623 * dwarf2out.c (new_die_raw): New static inline function.
10624 (new_die): Use it to create the DIE.
10625 (add_AT_external_die_ref): Likewise.
10626 (clone_die): Likewise.
10627 (clone_as_declaration): Likewise.
10628 (dwarf2out_vms_debug_main_pointer): Likewise.
10629 (base_type_die): Likewise. Remove early return for corner cases.
10630 Do not call add_pubtype on the DIE here.
10631 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
10632 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
10633 typedefs for base types with DW_AT_endianity. Make sure a DIE with
10634 native order exists for base types, attach the DIE manually and call
10635 add_pubtype on it. Do not equate a reverse order DIE to the type.
10637 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
10639 * config/arm/arm.c (align_ok_ldrd_strd): New function.
10640 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
10642 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
10644 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10646 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
10647 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
10648 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
10649 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
10650 * opts.c (sanitizer_opts): Add builtin.
10651 * ubsan.c (instrument_builtin): New function.
10652 (pass_ubsan::execute): Call it.
10653 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
10654 * doc/invoke.texi: Document -fsanitize=builtin.
10656 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
10657 builtins, store max (log2 (align), 0) into uchar field instead of
10658 align into uptr field.
10659 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
10660 store uchar 0 field instead of uptr 0 field.
10661 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
10662 instead of passing one address of struct with 2 locations pass
10663 two addresses of structs with 1 location each.
10664 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
10665 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
10666 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
10667 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
10668 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
10669 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
10670 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
10671 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
10673 2017-10-19 Martin Liska <mliska@suse.cz>
10676 * file-find.c (remove_prefix): Remove.
10677 * file-find.h (remove_prefix): Likewise.
10678 * gcc-ar.c: Remove smartness of lookup.
10680 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
10682 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
10683 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
10684 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
10686 2017-10-19 Jakub Jelinek <jakub@redhat.com>
10689 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
10690 (setcc + and to xor + setcc): New peephole2.
10692 2017-10-19 Tom de Vries <tom@codesourcery.com>
10694 * doc/sourcebuild.texi (Test Directives, Variants of
10695 dg-require-support): Add dg-require-stack-size.
10697 2017-10-19 Martin Liska <mliska@suse.cz>
10700 * gimplify.c (gimplify_decl_expr): Do not instrument variables
10701 that have a large alignment.
10702 (gimplify_target_expr): Likewise.
10704 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
10706 PR rtl-optimization/82602
10707 * ira.c (rtx_moveable_p): Return false for volatile asm.
10709 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
10712 * config/i386/i386-modes.def (CCGZ): New CC mode.
10713 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
10714 * config/i386/predicates.md (ix86_comparison_operator):
10716 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10717 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
10718 with double-word subtraction.
10719 (put_condition_code): Handle CCGZmode.
10721 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
10723 * wide-int.cc (debug (const wide_int &)): New.
10724 (debug (const wide_int *)): New.
10725 (debug (const widest_int &)): New.
10726 (debug (const widest_int *)): New.
10728 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
10730 PR middle-end/82556
10731 * lra-constraints.c (curr_insn_transform): Use non-input operand
10732 instead of output one for matched reload.
10734 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10736 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
10737 (tree-ssa-loop-ivopts.h): New header file.
10738 (struct builtin_info): New fields.
10739 (classify_builtin_1): Compute and record base and offset parts for
10740 memset builtin partition by calling strip_offset.
10741 (offset_cmp, fuse_memset_builtins): New functions.
10742 (finalize_partitions): Fuse adjacent memset partitions by calling
10744 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
10745 Expose the interface.
10746 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
10748 2017-10-18 Bin Cheng <bin.cheng@arm.com>
10750 PR tree-optimization/82574
10751 * tree-loop-distribution.c (find_single_drs): New parameter. Check
10752 that data reference must be executed exactly once per iteration
10753 against the outermost loop in nest.
10754 (classify_partition): Update call to above function.
10756 2017-10-18 Richard Biener <rguenther@suse.de>
10758 PR tree-optimization/82591
10759 * graphite.c (graphite_transform_loops): Move code gen message
10761 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10762 Here. Handle scop_to_isl_ast failing.
10763 (scop_to_isl_ast): Limit the number of ISL operations.
10765 2017-10-18 Richard Biener <rguenther@suse.de>
10767 * graphite-isl-ast-to-gimple.c
10768 (translate_isl_ast_to_gimple::set_rename): Simplify.
10769 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10770 (graphite_copy_stmts_from_block): ... here.
10771 (copy_bb_and_scalar_dependences): Simplify.
10772 (add_parameters_to_ivs_params): Canonicalize.
10773 (generate_entry_out_of_ssa_copies): Simplify.
10774 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10775 by passing in ISL dimension.
10776 (parameter_index_in_region_1): Rename to ...
10777 (parameter_index_in_region): ... this.
10778 (extract_affine): Adjust assert, pass down parameter index.
10779 (add_param_constraints): Use range-info when available.
10780 (build_scop_context): Adjust.
10781 * sese.c (new_sese_info): Adjust.
10782 (free_sese_info): Likewise.
10783 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10784 Remove unused typedefs.
10785 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10787 2017-10-18 Martin Liska <mliska@suse.cz>
10789 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10791 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10793 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10794 (s390_sched_init): Do not reset s390_sched_state if we entered the
10795 current basic block via a fallthru edge and all others are unlikely.
10797 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10799 * config/s390/s390.c (NUM_SIDES): New variable.
10800 (LONGRUNNING_THRESHOLD): New variable.
10801 (LATENCY_FACTOR): New variable.
10802 (s390_sched_score): Decrease score for long-running instructions on
10804 (s390_sched_variable_issue): Perform bookkeeping for long-running
10807 2017-10-18 Richard Biener <rguenther@suse.de>
10809 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10810 Simplify with removal of the parameter rename map.
10811 (set_rename): Likewise.
10812 (should_copy_to_new_region): Likewise.
10813 (graphite_copy_stmts_from_block): Likewise.
10814 (copy_bb_and_scalar_dependences): Remove initialization of
10815 unused copied_bb_map.
10816 (copy_def): Remove.
10817 (copy_internal_parameters): Likewise.
10818 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10819 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10820 Use INTEGRAL_TYPE_P.
10821 (parameter_index_in_region_1): Rename to ...
10822 (assign_parameter_index_in_region): ... this. Assert we have
10823 a parameter we handle.
10824 (scan_tree_for_params): Adjust.
10825 * sese.h (parameter_rename_map_t): Remove.
10826 (struct sese_info_t): Remove unused parameter_rename_map and
10827 copied_bb_map members.
10828 * sese.c (new_sese_info): Adjust.
10829 (free_sese_info): Likewise.
10831 2017-10-18 Martin Liska <mliska@suse.cz>
10834 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10835 on an abnormal edge.
10837 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10839 * doc/invoke.texi (ffunction-sections and fdata-sections):
10842 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10844 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10845 the use statement can throw internally.
10847 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10849 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10850 any RTX present on the RHS of a SET.
10851 * compare-elim.c (try_eliminate_compare): Restore comment.
10853 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10855 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10856 may be also called on tcc_type nodes.
10857 * langhooks.c (lhd_tree_size): Likewise.
10859 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10861 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10862 format_warning_at_substring.
10863 (maybe_warn): Convert source_range * param to a location_t. Pass
10864 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10865 (format_directive): Remove code to extract source_ranges and
10866 source_range * in favor of just a location_t.
10867 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10869 * substring-locations.c (format_warning_va): Convert
10870 source_range * param to a location_t.
10871 (format_warning_at_substring): Likewise.
10872 * substring-locations.h (format_warning_va): Likewise.
10873 (format_warning_at_substring): Likewise.
10875 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10877 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10879 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10880 and scatter/gather ops.
10882 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10883 vec_gather_load and vec_scatter_store.
10884 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10885 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10886 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10887 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10888 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10889 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10891 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10893 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10894 Detect FTST instruction and handle its register pops. Only pop
10895 second operand if can_pop_second_op is true.
10896 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10897 set can_pop_second_op to false in the compare_for_stack_reg call.
10899 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10900 output_fp_compare for stack register operands.
10901 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10902 instructions here. Do not emit stack register pops here. Assert
10903 that FCOMPP pops next to top stack register. Rewrite function.
10905 2017-10-17 Nathan Sidwell <nathan@acm.org>
10907 PR middle-end/82577
10908 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10909 use DECL_ASSEMBLER_NAME_RAW.
10911 PR middle-end/82546
10912 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10915 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10916 Wilco Dijkstra <wilco.dijkstra@arm.com>
10918 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10919 converstion to Pmode from the buf_addr.
10921 2017-10-17 Richard Biener <rguenther@suse.de>
10923 * graphite-dependences.c (scop_get_reads_and_writes): Change
10924 output parameters to references.
10926 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10928 PR 71026/tree-optimization
10929 * fold-const.c (distribute_real_division): Removed.
10930 (fold_binary_loc): Remove calls to distribute_real_divison.
10932 2017-10-17 Richard Biener <rguenther@suse.de>
10934 * graphite-scop-detection.c
10935 (scop_detection::stmt_has_simple_data_refs_p): Always use
10936 the full nest as region.
10937 (try_generate_gimple_bb): Likewise.
10938 * sese.c (scalar_evolution_in_region): Simplify now that
10939 SCEV can handle instantiation in regions.
10940 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10941 in the non-loop part of a function if requested.
10943 2017-10-17 Richard Biener <rguenther@suse.de>
10945 PR tree-optimization/82563
10946 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10948 (graphite_regenerate_ast_isl): Call it.
10949 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10951 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10953 PR tree-optimization/82549
10954 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10955 Formatting fixes. Instead of calling make_bit_field_ref with negative
10958 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10960 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10961 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10962 _mm_maskz_reduce_ss): New.
10963 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10964 __builtin_ia32_reducess_mask): Ditto..
10965 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10966 * config/i386/sse.md (reduces<mode>): Renamed to ...
10967 (reduces<mode><mask_scalar_name>): ... this.
10968 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10970 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10971 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10973 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10975 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10976 * selftest-run-tests.c (selftest::run_tests): Call
10977 selftest::unique_ptr_tests_cc_tests.
10978 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10979 * unique-ptr-tests.cc: New file.
10981 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10984 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10986 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10987 (make_hard_regno_born, make_hard_regno_dead): Update
10988 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10990 2017-10-16 Jeff Law <law@redhat.com>
10992 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10994 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10996 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10998 2017-10-16 Olivier Hainque <hainque@adacore.com>
11000 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11001 with_cpu if we were configured for an e500v2 target cpu name.
11003 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11005 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11006 * doc/invoke.texi: Document +nodsp as a valid extension for
11009 2017-10-16 Martin Liska <mliska@suse.cz>
11011 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11012 (test_set_range): Likewise.
11013 (test_range_functions): Rename to ...
11014 (test_bit_in_range): ... this.
11015 (sbitmap_c_tests): Add new test.
11017 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11019 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11021 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11022 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11024 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11026 * config/aarch64/aarch64-builtins.c
11027 (aarch64_types_quadopu_lane_qualifiers): New.
11028 (TYPES_QUADOPU_LANE): New.
11029 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11030 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11031 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11032 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11033 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11034 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11035 (Vdottype, DOTPROD): New.
11036 (sur): Add SDOT and UDOT.
11038 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11040 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11041 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11042 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11043 Add TARGET_DOTPROD.
11044 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11045 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11046 Enable TARGET_DOTPROD.
11047 (cortex-a75.cortex-a55): Likewise.
11048 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11050 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11052 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11053 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11055 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11057 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11058 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11059 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11060 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11061 * config/arm/types.md (neon_dot, neon_dot_q): New.
11062 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11064 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11066 * config/arm/arm.h (TARGET_DOTPROD): New.
11067 * config/arm/arm.c (arm_arch_dotprod): New.
11068 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11069 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11070 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11071 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11072 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11073 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11074 * doc/invoke.texi (armv8.2-a): Document dotprod
11076 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11078 * i386.c (ix86_vec_cost): New function.
11079 (ix86_rtx_costs): Handle vector operations better.
11080 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11081 * x86-tune-costs.h: Add new costs to all tables.
11083 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11085 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11087 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11088 divsd, sqrtss and sqrtsd
11089 * x86-tune-costs.h: Add new entries to all costs.
11090 (znver1_cost): Fix to match real instruction latencies.
11092 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11093 Michael Collison <michael.collison@arm.com>
11095 * compare-elim.c: Include emit-rtl.h.
11096 (can_merge_compare_into_arith): New function.
11097 (try_validate_parallel): Likewise.
11098 (try_merge_compare): Likewise.
11099 (try_eliminate_compare): Call the above when no previous clobber
11101 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11104 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11106 PR middle-end/62263
11107 PR middle-end/82498
11108 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11109 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11111 PR middle-end/62263
11112 PR middle-end/82498
11113 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11114 to be any operand_equal_p operands. For & (B - 1) require
11115 B to be power of 2. Recognize
11116 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11118 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11121 * optabs.c (expand_memory_blockage): Fix call of
11122 targetm.have_memory_blockage.
11124 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11127 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11128 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11129 objects to extra_objs instead of overwriting it.
11131 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11133 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11134 Use any_fp_register_operand as operand[3] predicate. Simplify
11135 equality test for operands[2] and operands[4] memory location.
11136 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11137 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11138 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11139 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11140 any_fp_register_operand as operand[1] predicate. Simplify
11141 equality test for operands[0] and operands[3] memory location.
11142 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11143 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11144 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11146 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11148 * target-insns.def: Add memory_blockage.
11149 * optabs.c (expand_memory_blockage): New function.
11150 (expand_asm_memory_barrier): Rename ...
11151 (expand_asm_memory_blockage): ... to this.
11152 (expand_mem_thread_fence): Call expand_memory_blockage
11153 instead of expand_asm_memory_barrier.
11154 (expand_mem_singnal_fence): Ditto.
11155 (expand_atomic_load): Ditto.
11156 (expand_atomic_store): Ditto.
11157 * doc/md.texi (Standard Pattern Names For Generation):
11158 Document memory_blockage instruction pattern.
11160 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11162 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11163 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11164 * config/rl78/rl78.md: New define_expand "adddi3".
11166 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11168 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11171 2017-10-13 Jeff Law <law@redhat.com>
11173 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11175 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11178 * internal-fn.c (expand_mul_overflow): If both operands have
11179 the same highpart of -1 or 0 and the topmost bit of lowpart
11180 is different, overflow is if res <= 0 rather than res < 0.
11182 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11184 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11185 TARGET_P9_VECTOR code for unaligned_load case.
11187 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11189 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11191 2017-10-13 Nathan Sidwell <nathan@acm.org>
11193 * tree-core.h (tree_contains_struct): Make bool.
11194 * tree.c (tree_contains_struct): Likewise.
11195 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11196 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11197 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11198 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11199 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11201 2017-10-13 Richard Biener <rguenther@suse.de>
11203 * graphite-isl-ast-to-gimple.c
11204 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11205 parameters and dominance check.
11206 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11207 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11208 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11209 Do not update SSA form here or do intermediate IL verification.
11210 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11211 (graphite_initialize): Remove check on the number of loops in
11212 the function and inline into graphite_transform_loops.
11213 (graphite_finalize): Inline into graphite_transform_loops.
11214 (graphite_transform_loops): Perform SSA update and IL verification
11216 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11218 2017-10-13 Richard Biener <rguenther@suse.de>
11220 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11221 graphite_expression_type_precision): Avoid global constructor
11223 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11224 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11225 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11226 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11227 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11228 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11230 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11233 * config/i386/i386.h (ix86_red_zone_size): New.
11234 * config/i386/i386.md (push peephole2s): Replace
11235 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11237 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11238 Alan Hayward <alan.hayward@arm.com>
11239 David Sherwood <david.sherwood@arm.com>
11241 * combine.c (can_change_dest_mode): Reject changes in
11242 REGMODE_NATURAL_SIZE.
11244 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11245 Alan Hayward <alan.hayward@arm.com>
11246 David Sherwood <david.sherwood@arm.com>
11248 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11249 (expand_debug_source_expr): Likewise.
11250 * combine.c (combine_simplify_rtx): Likewise.
11251 * cse.c (fold_rtx): Likewise.
11252 * fwprop.c (canonicalize_address): Likewise.
11253 * targhooks.c (default_shift_truncation_mask): Likewise.
11255 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11256 Alan Hayward <alan.hayward@arm.com>
11257 David Sherwood <david.sherwood@arm.com>
11259 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11260 (widened_mode): Likewise.
11261 (expand_unop): Likewise.
11262 * ree.c (transform_ifelse): Likewise.
11263 (merge_def_and_ext): Likewise.
11264 (combine_reaching_defs): Likewise.
11265 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11267 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11268 Alan Hayward <alan.hayward@arm.com>
11269 David Sherwood <david.sherwood@arm.com>
11271 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11272 * combine.c (gen_lowpart_for_combine): Likewise.
11273 * dwarf2out.c (rtl_for_decl_location): Likewise.
11274 * final.c (alter_subreg): Likewise.
11275 * rtlhooks.c (gen_lowpart_general): Likewise.
11276 (gen_lowpart_if_possible): Likewise.
11278 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11279 Alan Hayward <alan.hayward@arm.com>
11280 David Sherwood <david.sherwood@arm.com>
11282 * calls.c (expand_call): Use subreg_lowpart_offset.
11283 * cse.c (cse_insn): Likewise.
11284 * regcprop.c (copy_value): Likewise.
11285 (copyprop_hardreg_forward_1): Likewise.
11287 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11290 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11291 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11292 =Q constraints to +Q and into insn condition add check
11293 that operands[0] and operands[1] are equal.
11294 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11295 =Q constraints to +Q and into insn condition add check
11296 that operands[0] is equal to either operands[1] or operands[2].
11299 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11300 instead of handling MINUS_EXPR twice (once for each argument),
11301 canonicalize operand order and handle just once, use rtype where
11302 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11305 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11306 any values of __C while still being pattern recognizable as a simple
11307 rotate instruction.
11309 2017-10-13 Richard Biener <rguenther@suse.de>
11311 PR tree-optimization/82451
11313 2017-10-02 Richard Biener <rguenther@suse.de>
11315 PR tree-optimization/82355
11316 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11317 a mapping for the enclosing loop but avoid generating one for
11318 the loop tree root.
11319 (copy_bb_and_scalar_dependences): Remove premature codegen
11320 error on PHIs in blocks duplicated into multiple places.
11321 * graphite-scop-detection.c
11322 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11323 in the region use it as loop and nest to analyze the DR in.
11324 (try_generate_gimple_bb): Likewise.
11325 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11326 (add_loop_constraints): For blocks in a loop not in the region
11327 create a dimension with a single iteration.
11328 * sese.h (gbb_loop_at_index): Remove assert.
11330 * cfgloop.c (loop_preheader_edge): For the loop tree root
11331 return the single successor of the entry block.
11332 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11333 Reset the SCEV hashtable and niters.
11334 * graphite-scop-detection.c
11335 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11336 assert that we only have POLYNOMIAL_CHREC that vary in loops
11337 contained in the region.
11338 (scop_detection::graphite_can_represent_expr): Adjust.
11339 (scop_detection::stmt_has_simple_data_refs_p): For loops
11340 not in the region set loop to NULL. The nest is now the
11341 entry edge to the region.
11342 (try_generate_gimple_bb): Likewise.
11343 * sese.c (scalar_evolution_in_region): Adjust for
11344 instantiate_scev change.
11345 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11346 Make nest parameter the edge into the region.
11347 (create_data_ref): Likewise.
11348 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11349 entry edge into a region and adjust instantiate_scev calls.
11350 (create_data_ref): Likewise.
11351 (graphite_find_data_references_in_stmt): Likewise.
11352 (find_data_references_in_stmt): Pass the loop preheader edge
11353 from the nest argument.
11354 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11355 parameter the edge into the region.
11356 (instantiate_parameters): Use the loop preheader edge as entry.
11357 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11359 (get_instantiated_value_entry): Make instantiate_below parameter
11360 the edge into the region.
11361 (instantiate_scev_name): Likewise. Adjust dominance checks,
11362 when we cannot use loop-based instantiation instantiate by
11363 walking use-def chains.
11364 (instantiate_scev_poly): Adjust.
11365 (instantiate_scev_binary): Likewise.
11366 (instantiate_scev_convert): Likewise.
11367 (instantiate_scev_not): Likewise.
11368 (instantiate_array_ref): Remove.
11369 (instantiate_scev_3): Likewise.
11370 (instantiate_scev_2): Likewise.
11371 (instantiate_scev_1): Likewise.
11372 (instantiate_scev_r): Do not blindly handle N-operand trees.
11373 Do not instantiate array-refs. Handle all constants and invariants.
11374 (instantiate_scev): Make instantiate_below parameter
11375 the edge into the region.
11376 (resolve_mixers): Use the loop preheader edge for the region
11377 parameter to instantiate_scev_r.
11378 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11380 2017-10-13 Richard Biener <rguenther@suse.de>
11382 PR tree-optimization/82525
11383 * graphite-isl-ast-to-gimple.c
11384 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11386 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11387 Fail code generation when we cannot represent the isl integer.
11388 (binary_op_to_tree): Elide modulo operations that are no-ops
11389 in the type we code generate. Remove now superfluous code
11392 2017-10-13 Richard Biener <rguenther@suse.de>
11394 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11395 (scop_detection::harmful_loop_in_region): Remove premature
11396 IV type restriction.
11397 (scop_detection::graphite_can_represent_scev): We can handle
11398 pointer IVs just fine.
11400 2017-10-13 Alan Modra <amodra@gmail.com>
11402 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11403 "Clobbers and Scratch Registers". Add paragraph on
11404 alternative to clobbers for scratch registers and OpenBLAS
11407 2017-10-13 Alan Modra <amodra@gmail.com>
11409 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11410 example of a memory input for a string of known length. Move
11411 commentary out of table. Add a number of new examples
11412 covering array memory inputs.
11414 2017-10-12 Martin Liska <mliska@suse.cz>
11416 PR tree-optimization/82493
11417 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11418 (test_range_functions): New function.
11419 (sbitmap_c_tests): Likewise.
11420 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11421 * selftest.h (sbitmap_c_tests): New function.
11423 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11425 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11427 * config/rs6000/amo.h: Fix spacing issue.
11429 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11432 * config/i386/i386.md (*ashl<mode>3_mask_1,
11433 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
11434 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
11437 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11439 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
11440 (profile_probability): Set max_probability
11441 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
11443 * profile-count.c (profile_probability::differs_from_p): Do not
11444 rely on max_probaiblity == 10000
11446 2017-10-12 Jeff Law <law@redhat.com>
11448 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
11451 2017-10-12 Martin Sebor <msebor@redhat.com>
11455 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
11456 (handle_alias_pairs): Call it.
11457 * common.opt (-Wattribute-alias): New option.
11458 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
11459 * doc/invoke.texi (-Wattribute-alias): Document.
11461 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
11464 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11466 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11468 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11469 (make_hard_regno_born, make_hard_regno_dead): Update
11470 bb_killed_pseudos and bb_gen_pseudos.
11472 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
11474 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
11476 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
11478 * config/alpha/alpha.c (alpha_split_conditional_move):
11479 Use std::swap instead of manually swapping.
11480 (alpha_stdarg_optimize_hook): Ditto.
11481 (alpha_canonicalize_comparison): Ditto.
11483 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11485 * tree-loop-distribution.c (struct builtin_info): New struct.
11486 (struct partition): Refactor fields into struct builtin_info.
11487 (partition_free): Free struct builtin_info.
11488 (build_size_arg_loc, build_addr_arg_loc): Delete.
11489 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
11490 information from struct builtin_info.
11491 (find_single_drs): New function refactored from classify_partition.
11492 Also moved builtin validity checks to this function.
11493 (compute_access_range, alloc_builtin): New functions.
11494 (classify_builtin_st, classify_builtin_ldst): New functions.
11495 (classify_partition): Refactor code into functions find_single_drs,
11496 classify_builtin_st and classify_builtin_ldst.
11497 (distribute_loop): Don't do runtime alias check when distributing
11499 (find_seed_stmts_for_distribution): New function.
11500 (pass_loop_distribution::execute): Refactor code finding seed
11501 stmts into above function. Support distribution for the innermost
11502 two-level loop nest. Adjust dump information.
11504 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11506 * tree-loop-distribution.c: Adjust the general comment.
11507 (NUM_PARTITION_THRESHOLD): New macro.
11508 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
11509 (classify_partition): Skip builtin pattern of loop nest's inner loop.
11510 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
11511 in call to build_partition_graph.
11512 (finalize_partitions): New parameter. Make loop distribution more
11513 conservative by fusing more partitions.
11514 (distribute_loop): Don't do runtime alias check in case of loop nest
11516 (find_seed_stmts_for_distribution): New function.
11517 (prepare_perfect_loop_nest): New function.
11518 (pass_loop_distribution::execute): Refactor code finding seed stmts
11519 and loop nest into above functions. Support loop nest distribution.
11520 Adjust dump information accordingly.
11522 2017-10-12 Bin Cheng <bin.cheng@arm.com>
11524 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
11525 and set PTYPE_SEQUENTIAL for merged partition.
11527 2017-10-12 Richard Biener <rguenther@suse.de>
11529 PR tree-optimization/69728
11531 2017-09-19 Richard Biener <rguenther@suse.de>
11533 PR tree-optimization/69728
11534 * graphite-sese-to-poly.c (schedule_error): New global.
11535 (add_loop_schedule): Handle empty domain by failing the
11537 (build_original_schedule): Handle schedule_error.
11539 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
11540 domain by returning an unchanged schedule.
11542 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11544 * genrecog.c (validate_pattern): For VEC_SELECT verify that
11545 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
11547 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
11549 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
11552 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11555 * expr.c (store_field): Don't optimize away bitsize == 0 store
11556 from CALL_EXPR with addressable return type.
11558 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11560 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
11561 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11562 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
11563 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
11564 TARGET_ISEL instead of TARGET_ISEL<sel>.
11566 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
11568 * config/rs6000/rs6000.c
11569 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
11571 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
11573 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
11575 (reg_or_cint_operand): Fix comment.
11576 (reg_or_zero_operand): New predicate.
11577 * config/rs6000/rs6000-protos.h (output_isel): Delete.
11578 * config/rs6000/rs6000.c (output_isel): Delete.
11579 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
11580 instead of reg_or_cint_operand. Output instruction directly (not via
11582 (isel_unsigned_<mode>): Ditto.
11583 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
11584 gpc_reg_operand. Add an instruction alternative for this. Output
11585 instruction directly.
11586 (*isel_reversed_unsigned_<mode>): Ditto.
11588 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11590 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
11591 (TARGET_CANONICALIZE_COMPARISON): Define.
11593 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
11596 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
11597 Check whether the dest is REG before adding REG_EQUIV note.
11599 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
11602 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11604 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11605 (make_hard_regno_born, make_hard_regno_dead): Update
11606 bb_killed_pseudos and bb_gen_pseudos.
11608 2017-10-11 Nathan Sidwell <nathan@acm.org>
11610 * incpath.h (enum incpath_kind): Name enum, prefix values.
11611 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
11612 * incpath.c (heads, tails): Use INC_MAX.
11613 (add_env_var_paths, add_standard_paths): Use incpath_kind.
11614 (merge_include_chains, split_quote_chain,
11615 register_include_chains): Update incpath_kind names.
11616 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
11617 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
11619 (add_framework_path, darwin_register_objc_includes): Likewise.
11620 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
11622 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11624 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
11625 Do not use float_operator operator predicate.
11626 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
11627 * config/i386/predicates.md (float_operator): Remove predicate.
11629 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
11631 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
11632 (*jccxf_i387): Ditto.
11633 (*jcc<mode>_i387): Ditto.
11634 (*jccu<mode>_i387): Ditto.
11635 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
11636 (*jcc_*_i387 splitters): Remove.
11637 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
11638 * config/i386/i386.c (ix86_split_fp_branch): Remove.
11639 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
11642 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11644 * profile-count.h (slow_safe_scale_64bit): New function.
11645 (safe_scale_64bit): New inline.
11646 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
11647 * profile-count.c: Include wide-int.h
11648 (slow_safe_scale_64bit): New.
11650 2017-10-11 Nathan Sidwell <nathan@acm.org>
11652 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
11653 HAS_DECL_ASSEMBLER_NAME_P.
11654 * gimple-expr.c (gimple_decl_printable_name: Check
11655 HAS_DECL_ASSEMBLER_NAME_P too.
11656 * ipa-utils.h (type_in_anonymous_namespace_p): Check
11657 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
11658 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
11659 * passes.c (rest_of_decl_compilation): Check
11660 HAS_DECL_ASSEMBLER_NAME_P too.
11661 * recog.c (verify_changes): Likewise.
11662 * tree-pretty-print.c (dump_decl_name): Likewise.
11663 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
11665 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
11666 (SET_DECL_ASSEMBLER_NAME): Use it.
11667 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
11668 (COPY_DECL_ASSEMBLER_NAME): Likewise.
11669 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
11671 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
11673 * config.gcc (i386, x86_64): Add extra objects.
11674 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
11675 (ix86_min_insn_size): Declare.
11676 (ix86_issue_rate): Declare.
11677 (ix86_adjust_cost): Declare.
11678 (ia32_multipass_dfa_lookahead): Declare.
11679 (ix86_macro_fusion_p): Declare.
11680 (ix86_macro_fusion_pair_p): Declare.
11681 (ix86_bd_has_dispatch): Declare.
11682 (ix86_bd_do_dispatch): Declare.
11683 (ix86_core2i7_init_hooks): Declare.
11684 (ix86_atom_sched_reorder): Declare.
11685 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
11686 (COSTS_N_BYTES): Move to x86-tune-costs.h.
11687 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
11688 (rip_relative_addr_p): Rename to ...
11689 (ix86_rip_relative_addr_p): ... this one; export.
11690 (memory_address_length): Update.
11691 (ix86_issue_rate): Move to x86-tune-sched.c.
11692 (ix86_flags_dependent): Move to x86-tune-sched.c.
11693 (ix86_agi_dependent): Move to x86-tune-sched.c.
11694 (exact_dependency_1): Move to x86-tune-sched.c.
11695 (exact_store_load_dependency): Move to x86-tune-sched.c.
11696 (ix86_adjust_cost): Move to x86-tune-sched.c.
11697 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
11698 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
11699 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
11700 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
11701 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
11702 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
11703 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
11704 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
11705 (min_insn_size): Rename to ...
11706 (ix86_min_insn_size): ... this one; export.
11707 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
11708 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
11709 (core2i7_first_cycle_multipass_backtrack): Move to
11710 x86-tune-sched-core.c.
11711 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
11712 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
11713 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
11714 (ix86_avoid_jump_mispredicts): Update.
11715 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
11716 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
11717 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
11718 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
11719 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
11720 (MAX_INSN): Move to ix86-tune-sched-bd.c.
11721 (MAX_IMM): Move to ix86-tune-sched-bd.c.
11722 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
11723 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
11724 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
11725 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
11726 (MAX_STORE): Move to ix86-tune-sched-bd.c.
11727 (BIG): Move to ix86-tune-sched-bd.c.
11728 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
11729 (enum insn_path): Move to ix86-tune-sched-bd.c.
11730 (get_mem_group): Move to ix86-tune-sched-bd.c.
11731 (is_cmp): Move to ix86-tune-sched-bd.c.
11732 (dispatch_violation): Move to ix86-tune-sched-bd.c.
11733 (is_branch): Move to ix86-tune-sched-bd.c.
11734 (is_prefetch): Move to ix86-tune-sched-bd.c.
11735 (init_window): Move to ix86-tune-sched-bd.c.
11736 (allocate_window): Move to ix86-tune-sched-bd.c.
11737 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
11738 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
11739 (process_end_window): Move to ix86-tune-sched-bd.c.
11740 (allocate_next_window): Move to ix86-tune-sched-bd.c.
11741 (find_constant): Move to ix86-tune-sched-bd.c.
11742 (get_num_immediates): Move to ix86-tune-sched-bd.c.
11743 (has_immediate): Move to ix86-tune-sched-bd.c.
11744 (get_insn_path): Move to ix86-tune-sched-bd.c.
11745 (get_insn_group): Move to ix86-tune-sched-bd.c.
11746 (count_num_restricted): Move to ix86-tune-sched-bd.c.
11747 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
11748 (add_insn_window): Move to ix86-tune-sched-bd.c.
11749 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
11750 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
11751 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
11752 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
11753 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
11754 (do_dispatch): Move to ix86-tune-sched-bd.c.
11755 (has_dispatch): Move to ix86-tune-sched-bd.c.
11756 * config/i386/t-i386: Add new object files.
11757 * config/i386/x86-tune-costs.h: New file.
11758 * config/i386/x86-tune-sched-atom.c: New file.
11759 * config/i386/x86-tune-sched-bd.c: New file.
11760 * config/i386/x86-tune-sched-core.c: New file.
11761 * config/i386/x86-tune-sched.c: New file.
11763 2017-10-11 Liu Hao <lh_mouse@126.com>
11765 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
11766 the generic version below instead.
11767 (should_colorize): Recognize Windows consoles as terminals
11769 * pretty-print.c [__MINGW32__] (write_all): New function.
11770 [__MINGW32__] (find_esc_head): Likewise.
11771 [__MINGW32__] (find_esc_terminator): Likewise.
11772 [__MINGW32__] (eat_esc_sequence): Likewise.
11773 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11775 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11778 2017-10-11 Richard Biener <rguenther@suse.de>
11780 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11781 Properly call analyze_scalar_evolution with the loop of the stmt.
11783 2017-10-11 Richard Biener <rguenther@suse.de>
11785 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11786 * tree-core.h (tree_base): Add chrec_var union member.
11787 * tree.h (CHREC_VAR): Remove.
11788 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11789 * tree-chrec.h (build_polynomial_chrec): Adjust.
11790 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11791 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11793 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11795 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11796 * match.pd: ... here.
11797 ((T) X == (T) Y): Relax condition.
11799 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11801 PR tree-optimization/82472
11802 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11804 (break_alias_scc_partitions): Update postorder number.
11806 2017-10-11 Martin Liska <mliska@suse.cz>
11809 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11811 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11813 2017-10-11 Martin Liska <mliska@suse.cz>
11815 * print-rtl.c (print_insn): Move declaration of idbuf
11816 to same scope as name.
11818 2017-10-11 Martin Liska <mliska@suse.cz>
11823 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11824 volatile arguments.
11826 2017-10-11 Martin Liska <mliska@suse.cz>
11829 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11830 volatile arguments.
11832 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11834 * config.gcc (default_gnu_indirect_function): Default to yes for
11835 arm*-*-linux* with glibc.
11837 2017-10-11 Richard Biener <rguenther@suse.de>
11839 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11840 default-defs and types we do not want to analyze.
11841 (interpret_loop_phi): Replace unreachable code with an assert.
11842 (compute_scalar_evolution_in_loop): Remove and inline ...
11843 (analyze_scalar_evolution_1): ... here, replacing condition with
11844 what makes the intent clearer. Remove handling of cases
11845 get_scalar_evolution now handles.
11847 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11849 PR rtl-optimization/81434
11850 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11851 comment for main loop. In sched_group_found if, also add checks for
11852 pass and min_cost_group.
11854 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11856 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11857 (rs6000_insn_cost): New function.
11858 * config/rs6000/rs6000.md (cost): New attribute.
11860 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11861 H.J. Lu <hongjiu.lu@intel.com>
11865 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11866 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11867 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11868 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11869 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11870 OPTION_MASK_ISA_MMX and some other bit set require both
11871 mmx and the ISAs without the mmx bit.
11872 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11873 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11874 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11875 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11876 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11877 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11878 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11879 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11880 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11881 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11882 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11883 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11884 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11885 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11886 Add OPTION_MASK_ISA_MMX.
11888 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11890 * config.gcc (armv7*-*-freebsd*): New target.
11891 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11893 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11895 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11896 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11899 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11901 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11902 a template parameter.
11903 (WIDE_INT_REF_FOR): Update accordingly.
11904 * tree.h (wi::int_traits <const_tree>): Delete.
11905 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11906 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11907 (wi::tree_to_wide_ref): New typedef.
11908 (wi::to_wide): New function.
11909 * calls.c (get_size_range): Use wi::to_wide when operating on
11910 trees as wide_ints.
11911 * cgraph.c (cgraph_node::create_thunk): Likewise.
11912 * config/i386/i386.c (ix86_data_alignment): Likewise.
11913 (ix86_local_alignment): Likewise.
11914 * dbxout.c (stabstr_O): Likewise.
11915 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11916 * expr.c (const_vector_from_tree): Likewise.
11917 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11918 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11919 (fold_negate_expr_1, int_const_binop_1, const_binop)
11920 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11921 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11922 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11923 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11924 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11925 (fold_not_const, round_up_loc): Likewise.
11926 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11927 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11928 (alloca_call_type): Likewise.
11929 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11930 * godump.c (go_output_typedef): Likewise.
11931 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11932 * internal-fn.c (get_min_precision): Likewise.
11933 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11934 * ipa-polymorphic-call.c
11935 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11936 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11937 (ipa_modify_call_arguments): Likewise.
11938 * match.pd: Likewise.
11939 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11940 * print-tree.c (print_node_brief, print_node): Likewise.
11941 * stmt.c (expand_case): Likewise.
11942 * stor-layout.c (layout_type): Likewise.
11943 * tree-affine.c (tree_to_aff_combination): Likewise.
11944 * tree-cfg.c (group_case_labels_stmt): Likewise.
11945 * tree-data-ref.c (dr_analyze_indices): Likewise.
11946 (prune_runtime_alias_test_list): Likewise.
11947 * tree-dump.c (dequeue_and_dump): Likewise.
11948 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11949 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11950 * tree-pretty-print.c (dump_generic_node): Likewise.
11951 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11952 (simple_iv_with_niters): Likewise.
11953 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11954 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11955 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11956 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11957 (refine_value_range_using_guard, number_of_iterations_ne_max)
11958 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11959 (get_cst_init_from_scev, record_nonwrapping_iv)
11960 (scev_var_range_cant_overflow): Likewise.
11961 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11962 * tree-ssa-pre.c (compute_avail): Likewise.
11963 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11964 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11965 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11966 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11967 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11968 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11969 (dump_case_nodes, try_switch_expansion): Likewise.
11970 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11971 (vect_do_peeling): Likewise.
11972 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11973 * tree-vect-stmts.c (vectorizable_load): Likewise.
11974 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11975 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11976 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11977 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11978 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11979 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11980 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11981 (evrp_dom_walker::before_dom_children): Likewise.
11982 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11983 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11984 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11985 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11986 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11987 (get_range_pos_neg): Likewise.
11988 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11989 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11990 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11991 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11992 * config/avr/avr.c (avr_fold_builtin): Likewise.
11993 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11994 * config/msp430/msp430.c (msp430_attr): Likewise.
11995 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11996 * config/powerpcspe/powerpcspe-c.c
11997 (altivec_resolve_overloaded_builtin): Likewise.
11998 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11999 (rs6000_expand_ternop_builtin): Likewise.
12000 * config/rs6000/rs6000-c.c
12001 (altivec_resolve_overloaded_builtin): Likewise.
12002 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12003 (rs6000_expand_ternop_builtin): Likewise.
12004 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12006 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12008 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12009 when copying loop nest with only one inner loop.
12011 2017-10-10 Richard Biener <rguenther@suse.de>
12013 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12014 blocks if SCEV is active.
12015 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12017 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12018 (scev_initialize): Assert we are not yet initialized.
12020 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12022 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12023 inner loop's exit stmt by making it always exit the loop, otherwise
12024 we would generate an infinite empty loop.
12026 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12028 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12029 renaming variables in new preheader if it's deleted.
12031 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12033 * tree-loop-distribution.c (struct partition): Remove unused field
12034 loops of the structure.
12035 (partition_alloc, partition_free): Ditto.
12036 (build_rdg_partition_for_vertex): Ditto.
12038 2017-10-09 Jeff Law <law@redhat.com>
12040 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12041 return type to match prototype and documentation.
12043 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12045 * config/rs6000/rs6000.c (processor_costs): Move to ...
12046 * config/rs6000/rs6000.h: ... here.
12047 (rs6000_cost): Declare.
12049 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12051 * except.c (setjmp_fn): New global variable.
12052 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12053 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12054 if DONT_USE_BUILTIN_SETJMP is defined.
12056 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12058 * target.def (insn_cost): New hook.
12059 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12060 * doc/tm.texi: Regenerate.
12061 * rtlanal.c (insn_cost): Use the new hook.
12063 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12065 * combine.c (combine_validate_cost): Compute the new insn_cost,
12066 not just pattern_cost.
12067 (try_combine): Adjust comment.
12069 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12071 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12073 * combine.c (uid_insn_cost): Adjust comment.
12074 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12076 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12077 * dse.c (find_shift_sequence): Ditto.
12078 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12079 (bb_valid_for_noce_process_p): Use pattern_cost.
12080 * rtl.h (insn_rtx_cost): Delete.
12081 (pattern_cost): New prototype.
12082 (insn_cost): New prototype.
12083 * rtlanal.c (insn_rtx_cost): Rename to...
12084 (pattern_cost): ... this.
12087 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12089 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12090 (*jcc<mode>_0_r_i387): Ditto.
12091 (*jccxf_r_i387): Ditto.
12092 (*jcc<mode>_r_i387): Ditto.
12093 (*jccu<mode>_r_i387): Ditto.
12094 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12095 (*jcc): Rename from *jcc_1.
12097 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12099 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12100 deferred rescans after the lvx/stvx recombination pre-pass.
12102 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12104 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12105 memory operation instruction support.
12106 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12107 (rs6000-ibm-aix[789]*): Likewise.
12108 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12109 Document new functions.
12111 2017-10-09 Richard Biener <rguenther@suse.de>
12113 PR tree-optimization/82397
12114 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12115 equality only for semantically equal trees.
12117 2017-10-09 Richard Biener <rguenther@suse.de>
12119 PR tree-optimization/82449
12120 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12121 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12122 allow constant addresses.
12123 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12126 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12128 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12131 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12134 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12137 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12140 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12142 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12145 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12146 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12148 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12150 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12151 (WI_BINARY_PREDICATE_RESULT): Likewise.
12152 (wi::binary_traits::operator_result): New type.
12153 (wi::binary_traits::predicate_result): Likewise.
12154 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12155 (generic_wide_int::operator==, generic_wide_int::operator!=)
12156 (generic_wide_int::operator&, generic_wide_int::and_not)
12157 (generic_wide_int::operator|, generic_wide_int::or_not)
12158 (generic_wide_int::operator^, generic_wide_int::operator+
12159 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12161 (operator~, unary operator-, operator==, operator!=, operator&)
12162 (operator|, operator^, operator+, binary operator-, operator*): New
12164 * expr.c (get_inner_reference): Use wi::bit_and_not.
12165 * fold-const.c (fold_binary_loc): Likewise.
12166 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12167 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12168 (bit_value_binop): Likewise.
12169 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12170 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12171 (extract_range_from_binary_expr_1): Likewise.
12172 (masked_increment): Likewise.
12173 (simplify_bit_ops_using_ranges): Likewise.
12175 2017-10-09 Martin Jambor <mjambor@suse.cz>
12178 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12179 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12180 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12182 (hsa_fixup_mov_insn_type): New function.
12183 (hsa_op_with_type::get_in_type): Use it.
12184 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12185 immediates in an assert.
12186 (hsa_op_with_type::extend_int_to_32bit): New method.
12187 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12188 types. Convert to dest type if necessary.
12189 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12190 (reg_for_gimple_ssa): Pass false as min32int to
12191 hsa_type_for_scalar_tree_type.
12192 (gen_hsa_addr): Fixup type when creating addresable temporary.
12193 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12194 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12195 necessary. Call hsa_fixup_mov_insn_type.
12196 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12197 extend operands and convert to dest type if necessary.
12198 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12199 to dest type if necessary.
12200 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12201 if conversion nt necessary and size matches.
12202 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12203 to dest type if necessary.
12204 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12205 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12207 (gen_hsa_clrsb): Likewise.
12208 (gen_hsa_ffs): Likewise.
12209 (gen_hsa_divmod): Extend operands and convert to dest type if
12211 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12213 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12215 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12216 Remove empty default arguments. Use a brace block as output
12218 (conditional return): Ditto.
12220 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12221 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12223 (group_ending_nop): Ditto.
12224 (doloop_end): Ditto.
12225 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12226 (splitters for those): Ditto.
12228 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12230 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12231 a conditional jump (and the compare for it) so that pc_rtx is the
12233 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12234 for the deleted and renamed ctr<mode>_internal[234] patterns.
12235 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12236 Delete second conditional return pattern.
12237 (ctr<mode>_internal2): Delete this second bdnz pattern.
12238 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12239 (ctr<mode>_internal4): Delete this second bdz pattern.
12241 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12243 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12244 (always_initialized_rtx_for_ssa_name_p): New predicate.
12245 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12246 (finish_out_of_ssa): Free new field of SA.
12247 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12248 * tree-ssa-coalesce.c: Include tree-ssa.h.
12249 (get_parm_default_def_partitions): Remove extern keyword.
12250 (get_undefined_value_partitions): New function.
12251 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12252 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12253 uninitialized bits.
12254 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12256 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12258 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12260 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12262 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12263 for targets that preffer 128bit.
12265 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12267 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12269 2017-10-08 Olivier Hainque <hainque@adacore.com>
12271 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12272 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12274 (thumb_set_return_address): Likewise.
12276 2017-10-08 Olivier Hainque <hainque@adacore.com>
12278 * common/config/arm/arm-common.c (arm_except_unwind_info):
12279 Handle DWARF2_UNWIND_INFO.
12281 2017-10-07 Michael Collison <michael.collison@arm.com>
12283 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12286 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12288 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12289 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12290 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12291 defined, force the creation of a new block for a dispatch label.
12293 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12295 * invoke.texi (Wsuggest-attribute=cold): Document.
12296 * common.opt (Wsuggest-attribute=cold): New
12297 * ipa-pure-const.c (warn_function_cold): New function.
12298 * predict.c (compute_function_frequency): Use it.
12299 * predict.h (warn_function_cold): Declare.
12301 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12303 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12306 2017-10-06 Martin Liska <mliska@suse.cz>
12308 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12309 keyword for member functions.
12310 (struct sanopt_tree_couple): New struct.
12311 (struct sanopt_tree_couple_hash): New function.
12312 (struct sanopt_ctx): Add new hash_map.
12313 (has_dominating_ubsan_ptr_check): New function.
12314 (record_ubsan_ptr_check_stmt): Likewise.
12315 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12316 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12317 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12319 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12322 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12323 aarch64_simd_valid_immediate on CONST_VECTORs.
12324 (aarch64_reg_or_bic_imm): Likewise.
12326 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12328 PR rtl-optimization/82396
12329 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12331 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12333 * graphite-dependences.c (scop_get_reads): Move code to...
12334 (scop_get_must_writes): Move code to...
12335 (scop_get_may_writes): Move code to...
12336 (scop_get_reads_and_writes): ... here.
12337 (scop_get_dependences): Call scop_get_reads_and_writes.
12339 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12341 PR tree-optimization/82434
12342 * fold-const.h (can_native_encode_type_p,
12343 can_native_encode_string_p): Remove.
12344 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12345 don't encode anything, just return what would be otherwise returned.
12346 (native_encode_fixed, native_encode_complex, native_encode_vector):
12348 (native_encode_string): Likewise. Inline by hand
12349 can_native_encode_string_p.
12350 (can_native_encode_type_p): Remove.
12351 (can_native_encode_string_p): Remove.
12352 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12353 STRING_CSTs using can_native_encode_string_p, test all
12354 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12355 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12356 argument from native_encode_expr.
12357 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12358 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12361 2017-10-06 Richard Biener <rguenther@suse.de>
12363 PR tree-optimization/82397
12364 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12365 operand_equal_p but rely on data_ref_compare_tree for detecting
12367 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12368 to match up with dr_group_sort_cmp.
12370 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12373 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12375 * config/s390/s390-builtin-types.def: Regenerate.
12377 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12380 * config/s390/s390-builtin-types.def: Regenerate.
12381 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12382 Change flag from B_VXE to B_VX.
12383 (s390_vec_min_dbl): Remove B_VXE flag.
12385 2017-10-06 Richard Biener <rguenther@suse.de>
12387 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12388 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12389 translate_isl_ast_to_gimple::is_valid_rename,
12390 translate_isl_ast_to_gimple::get_rename,
12391 translate_isl_ast_to_gimple::get_def_bb_for_const,
12392 translate_isl_ast_to_gimple::get_new_name,
12393 translate_isl_ast_to_gimple::collect_all_ssa_names,
12394 translate_isl_ast_to_gimple::copy_loop_phi_args,
12395 translate_isl_ast_to_gimple::collect_all_ssa_names,
12396 translate_isl_ast_to_gimple::copy_loop_phi_args,
12397 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12398 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12399 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12400 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12401 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12402 translate_isl_ast_to_gimple::copy_cond_phi_args,
12403 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12404 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12405 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12406 translate_isl_ast_to_gimple::rename_uses,
12407 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12408 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12409 (set_rename_for_each_def): Likewise.
12410 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12411 here. Handle rewriting SCEV analyzable uses here.
12412 (copy_bb_and_scalar_dependences): Generate code for PHI
12414 (graphite_regenerate_ast_isl): Adjust.
12415 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12416 (add_write, add_read): New functions.
12417 (build_cross_bb_scalars_def): Use it and simplify.
12418 (build_cross_bb_scalars_use): Likewise.
12419 (graphite_find_cross_bb_scalar_vars): Inline into...
12420 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12421 simulating out-of-SSA. Compute liveout and add dependencies.
12422 (build_scops): Force an empty entry block.
12423 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12425 (sese_build_liveouts): Declare.
12426 (sese_trivially_empty_bb_p): Likewise.
12427 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12428 compute liveout and debug_liveout.
12429 (sese_bad_liveouts_use): Remove.
12430 (sese_reset_debug_liveouts_bb): Likewise.
12431 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12432 (sese_build_liveouts): Build liveout and debug_liveout and store
12434 (new_sese_info): Adjust.
12435 (free_sese_info): Likewise.
12436 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
12437 do not build liveout here.
12438 (move_sese_in_condition): Adjust region entry.
12439 (scev_analyzable_p): Match up with chrec_apply requirements.
12440 (sese_trivially_empty_bb_p): New.
12441 * tree-into-ssa.c (get_reaching_def): Properly support generating
12442 default-defs for incremental rewrite of anonymous names.
12444 2017-10-06 Richard Biener <rguenther@suse.de>
12446 * graphite-sese-to-poly.c (extract_affine): For casts increasing
12447 precision do not perform modulo reduction.
12449 2017-10-06 Richard Biener <rguenther@suse.de>
12451 PR tree-optimization/82436
12452 * tree-vect-slp.c (vect_supported_load_permutation_p): More
12453 conservatively choose the vectorization factor when checking
12454 whether we can perform the required load permutation.
12455 (vect_transform_slp_perm_load): Assert when we may not fail.
12457 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
12459 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
12460 message for incompatible -msdata=* and -mcall-* options.
12462 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12464 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
12465 rate for post-reload scheduling.
12467 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12469 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
12471 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12473 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
12474 to improve monte carlo in scimark.
12476 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
12478 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
12479 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
12480 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
12481 pentium4_cost, nocona_cost): Set reassociation width to 1.
12482 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
12483 width to 2 for fp operations and 1 otherwise.
12484 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
12486 (atom_cost): Set reassociation width to 2.
12487 (slm_cost, generic_cost): Set fp reassociation width
12488 to 2 and 1 otherwise.
12489 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
12490 (core_cost): Set fp reassociation width to 4 and vector to 2.
12491 (ix86_reassociation_width): Rewrite using cost table; special case
12492 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
12493 and TARGET_AVX128_OPTIMAL.
12494 * config/i386/i386.h (processor_costs): Add
12495 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
12496 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
12497 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
12498 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
12499 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
12500 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
12502 2017-10-05 Nathan Sidwell <nathan@acm.org>
12504 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
12506 2017-10-05 Tamar Christina <tamar.christina@arm.com>
12508 * config/arm/arm.c (arm_test_fpu_data): New.
12509 (arm_run_selftests): Call arm_test_fpu_data.
12511 2017-10-04 Nathan Sidwell <nathan@acm.org>
12513 * toplev.c (toplev::main): Remove excess parens on pretty_printer
12515 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
12517 2017-10-04 Sudakshina Das <sudi.das@arm.com>
12519 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
12520 check type for aarch64_simd_valid_immediate.
12521 (aarch64_output_simd_mov_immediate): Update prototype.
12522 (aarch64_simd_valid_immediate): Update prototype.
12523 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
12524 support for ORR-immediate.
12525 (and<mode>3): modified pattern to add support for BIC-immediate.
12526 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
12527 now checks for valid immediate for BIC and ORR based on new enum
12529 (aarch64_output_simd_mov_immediate): Function now used to output
12530 BIC/ORR imm as well based on new enum argument.
12531 * config/aarch64/constraints.md (Do): New vector immediate constraint.
12533 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
12534 (aarch64_reg_or_bic_imm): Likewise.
12536 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12538 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
12539 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
12541 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12545 PR rtl-optimization/82396
12546 * haifa-sched.c (autopref_multipass_init): Simplify
12548 (autopref_rank_data): Simplify sort order.
12549 * sched-int.h (autopref_multipass_data_): Remove
12550 multi_mem_insn_p, min_offset and max_offset.
12552 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12554 * doc/sourcebuild.texi: Document vect_peeling_profitable.
12556 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12558 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
12559 vect_doubleint_cvt.
12561 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12563 * doc/sourcebuild.texi: Document vect_long_mult.
12565 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
12567 PR tree-optimization/82413
12568 * fold-const.c (build_range_check): Use widest_int when comparing
12569 the maximum ETYPE value with HIGH.
12571 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
12573 PR rtl-optimization/82396
12574 * haifa-sched.c (autopref_multipass_init): Simplify
12576 (autopref_rank_data): Simplify sort order.
12577 * sched-int.h (autopref_multipass_data_): Remove
12578 multi_mem_insn_p, min_offset and max_offset.
12580 2017-10-04 Jakub Jelinek <jakub@redhat.com>
12582 PR tree-optimization/82381
12583 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
12584 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
12587 PR tree-optimization/82374
12588 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
12589 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
12590 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
12591 current_function_decl to the new decl.
12593 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12595 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
12596 helper macro for IEEE float128 hardware built-in functions.
12597 (SQRTF128_ODD): Add built-in functions with the round-to-odd
12599 (TRUNCF128_ODD): Likewise.
12600 (ADDF128_ODD): Likewise.
12601 (SUBF128_ODD): Likewise.
12602 (MULF128_ODD): Likewise.
12603 (DIVF128_ODD): Likewise.
12604 (FMAF128_ODD): Likewise.
12605 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
12606 UNSPEC_TRUNC_ROUND_TO_ODD.
12607 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
12608 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
12609 floating point round to odd instructions.
12610 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
12611 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
12612 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
12613 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
12614 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
12615 (trunc<mode>sf2_hw): Change the truncate with round to odd
12616 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
12617 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
12618 to odd hardware instructions.
12619 (sub<mode>3_odd): Likewise.
12620 (mul<mode>3_odd): Likewise.
12621 (div<mode>3_odd): Likewise.
12622 (sqrt<mode>2_odd): Likewise.
12623 (fma<mode>4_odd): Likewise.
12624 (fms<mode>4_odd): Likewise.
12625 (nfma<mode>4_odd): Likewise.
12626 (nfms<mode>4_odd): Likewise.
12627 (trunc<mode>df2_odd): Change the truncate with round to odd
12628 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
12630 * doc/extend.texi (PowerPC built-in functions): Update documentation
12631 for existing IEEE float128-bit built-in functions. Add built-in
12632 functions that generate the IEEE 128-bit floating point round to
12635 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
12637 PR rtl-optimization/77729
12638 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
12639 to (X&(C1&~C2))|C2 transformations.
12641 2017-10-03 Martin Jambor <mjambor@suse.cz>
12643 PR tree-optimization/82363
12644 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
12645 mismatch, mark lacc written regardless of racc.
12647 2017-10-03 Jakub Jelinek <jakub@redhat.com>
12649 PR tree-optimization/82381
12650 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
12651 stmt_to_insert nor wheather SSA_NAMEs are default defs.
12652 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
12653 fallthrough into reassoc_stmt_dominates_stmt_p.
12656 * combine.c (combine_instructions): Don't combine in unreachable
12659 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
12662 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
12663 function to not use the have_cpu variable. Do not set cpu_index,
12664 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
12665 or the default cpu.
12666 (rs6000_valid_attribute_p): Remove duplicate initializations of
12667 old_optimize and func_optimize.
12668 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
12669 (rs6000_activate_target_options): Make global.
12670 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
12673 2017-10-02 Jakub Jelinek <jakub@redhat.com>
12675 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
12676 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
12677 Set *poffset to 0 and *psize and *pmax_size to -1 if
12678 *poffset + *psize overflows in HOST_WIDE_INT.
12680 PR tree-optimization/82387
12681 PR tree-optimization/82388
12682 PR tree-optimization/82389
12683 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
12684 instead of live_bytes non-NULL.
12686 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
12689 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
12692 2017-10-02 Richard Biener <rguenther@suse.de>
12694 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
12695 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
12696 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
12698 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12700 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
12701 requested precision matches the type's.
12702 * calls.c (alloc_max_size): Calculate the new candidate size as
12703 a widest_int and use wi::to_widest when comparing it with the
12704 current candidate size.
12705 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
12706 zero rather than integer_zero_node.
12707 * match.pd: Check for a no-op conversion before using wi::add
12708 rather than after. Use tree_to_uhwi when summing small shift
12709 counts into an unsigned int.
12711 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
12712 Alan Hayward <alan.hayward@arm.com>
12713 David Sherwood <david.sherwood@arm.com>
12716 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
12717 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
12718 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
12719 POINTER_AND_FP_REGS.
12721 2017-10-02 Richard Biener <rguenther@suse.de>
12723 PR tree-optimization/82355
12724 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12725 a mapping for the enclosing loop but avoid generating one for
12726 the loop tree root.
12727 (copy_bb_and_scalar_dependences): Remove premature codegen
12728 error on PHIs in blocks duplicated into multiple places.
12729 * graphite-scop-detection.c
12730 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12731 in the region use it as loop and nest to analyze the DR in.
12732 (try_generate_gimple_bb): Likewise.
12733 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12734 (add_loop_constraints): For blocks in a loop not in the region
12735 create a dimension with a single iteration.
12736 * sese.h (gbb_loop_at_index): Remove assert.
12738 2017-10-01 Kevin Buettner <kevinb@redhat.com>
12740 * omp-expand.c (adjust_context_scope): New function.
12741 (expand_parallel_call): Call adjust_context_scope.
12743 2017-10-01 Jeff Law <law@redhat.com>
12745 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
12746 dom_opt_dom_walker class with direct access to private members.
12747 Add comments. Call test_for_singularity.
12748 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
12749 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
12750 m_dummy_cond anymore.
12751 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
12753 (pass_dominator:execute): Build the dummy_cond here and pass it
12754 to the dom_opt_dom_walker ctor.
12755 (test_for_singularity): New function.
12757 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
12758 Maya Rashish <coypu@sdf.org>
12760 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
12761 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
12762 (alpha*-*-netbsd*) Use nbsd_tm_file.
12763 (arm*-*-netbsdelf*) Likewise.
12764 (i[34567]86-*-netbsdelf*) Likewise.
12765 (x86_64-*-netbsd*) Likewise.
12766 (mips*-*-netbsd*) Likewise.
12767 (powerpc-*-netbsd*) Likewise.
12768 (sh*-*-netbsd*) Likewise.
12769 (sparc-*-netbsdelf*) Likewise.
12770 (sparc64-*-netbsd*) Likewise.
12771 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12773 (vax-*-netbsdelf*) Likewise.
12774 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12775 (UINT_FAST8_TYPE) Likewise.
12776 (INT_FAST16_TYPE) Check CHAR_FAST16.
12777 (UINT_FAST16_TYPE) Likewise.
12779 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12782 * config/i386/i386.md
12783 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12784 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12785 *divmodsi4_zext_2): New define_insn_and_split.
12786 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12787 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12788 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12789 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12790 New define_insn_and_split.
12791 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12792 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12793 operands[1] having DImode when mode is SImode.
12795 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12796 always SImode for DIV and MOD in REG_EQUAL notes.
12798 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12800 PR middle-end/82319
12801 * match.pd: Fix handling of NaNs in pattern.
12803 2017-09-29 Jeff Law <law@redhat.com>
12805 * sbitmap.c (bitmap_bit_in_range_p): New function.
12806 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12807 * tree-ssa-dse.c (live_bytes_read): New function.
12808 (dse_classify_store): Ignore reads of dead bytes.
12810 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12811 typos and whitespace errors.
12812 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12813 * config/s390/s390.c (s390_emit_prologue): Likewise.
12815 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12818 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12819 with a symbol for LRA.
12821 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12823 PR rtl-optimization/82338
12824 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12826 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12828 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12829 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12830 (qsort_chk): Declare.
12831 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12832 (qsort_chk): New function.
12834 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12836 PR tree-optimization/82337
12837 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12838 phi definition if the PHI result appears in an abnormal PHI.
12839 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12840 basis appears in an abnormal PHI.
12842 2017-09-29 Richard Biener <rguenther@suse.de>
12844 * graphite-isl-ast-to-gimple.c
12845 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12846 (binary_op_to_tree): Use it.
12847 (get_rename_from_scev): Likewise.
12848 (copy_loop_phi_nodes): Likewise.
12849 (copy_bb_and_scalar_dependences): Likewise.
12850 (translate_pending_phi_nodes): Likewise.
12852 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12855 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12856 for movabsq $(i32 << shift), r64.
12858 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12860 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12861 index when encoding %esp as %rsp to avoid 0x67 prefix.
12863 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12865 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12866 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12868 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12870 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12871 Extensions with more than 16 double VFP registers.
12872 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12873 to_clear_mask and all code related to it. Replace the remaining
12874 entry by a sbitmap and adapt code accordingly.
12876 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12878 * brig-builtins.def: Change pure attributes to const.
12880 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12882 * config.gcc (default_gnu_indirect_function): Default to yes for
12883 sparc*-*-linux* with glibc.
12885 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12887 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12888 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12889 when creating .init_array and .fini_array sections with priority
12892 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12895 * config/aarch64/aarch64.c
12896 (aarch64_builtin_support_vector_misalignment): Always return false
12897 when misalignment is unknown.
12899 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12901 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12902 this function to return false if the definition used by the swap
12903 instruction is artificial, or if the memory address from which the
12904 constant value is loaded is not represented by a base address held
12905 in a register or if the base address register is a frame or stack
12906 pointer. Additionally, return false if the base address of the
12907 loaded constant is a SYMBOL_REF but is not considered to be a
12909 (replace_swapped_load_constant): New function.
12910 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12911 loaded constant vector with a load of a swapped constant vector.
12913 2017-09-27 Carl Love <cel@us.ibm.com>
12915 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12916 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12917 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12920 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12922 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12923 first, always call autopref_rank_data otherwise.
12925 2017-09-27 Richard Biener <rguenther@suse.de>
12927 * graphite-scop-detection.c (find_scop_parameters): Move
12928 loop bound handling ...
12929 (gather_bbs::before_dom_children): ... here, avoiding the need
12930 to build scop_info->loop_nest.
12931 (record_loop_in_sese): Remove.
12932 * sese.h (sese_info_t::loop_nest): Remove.
12933 * sese.c (new_sese_info): Do not allocate loop_nest.
12934 (free_sese_info): Do not free loop_nest.
12936 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12939 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12940 lhs from calls if the lhs has addressable type.
12942 2017-09-27 Richard Biener <rguenther@suse.de>
12944 * graphite.h (scop::max_alias_set): New member.
12945 * graphite-scop-detection.c: Remove references to non-existing
12946 --param in comments.
12947 (build_alias_sets): Record the maximum alias set used for drs.
12948 (build_scops): Support zero as unlimited for
12949 --param graphite-max-arrays-per-scop.
12950 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12951 and inline into ...
12952 (build_poly_sr_1): ... here. Compute alias set based on the
12953 maximum alias set used for drs rather than
12954 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12956 2017-09-27 Richard Biener <rguenther@suse.de>
12958 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12959 --param loop-block-tile-size=0 to disable tiling.
12961 2017-09-27 Richard Biener <rguenther@suse.de>
12963 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12964 (graphite-max-nb-scop-params): Document special value zero.
12965 * domwalk.h (dom_walker::STOP): New symbolical constant.
12966 (dom_walker::dom_walker): Add optional parameter for bb to
12968 (dom_walker::~dom_walker): Declare.
12969 (dom_walker::before_dom_children): Document STOP return value.
12970 (dom_walker::m_user_bb_to_rpo): New member.
12971 (dom_walker::m_bb_to_rpo): Likewise.
12972 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12973 mapping here if not provided by the user.
12974 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12975 provided by the user.
12976 (dom_walker::STOP): Define.
12977 (dom_walker::walk): Do not compute bb to RPO mapping here.
12978 Support STOP return value from before_dom_children to stop
12980 * graphite-optimize-isl.c (optimize_isl): If the schedule
12981 is the same still generate code if -fgraphite-identity
12982 or -floop-parallelize-all are given.
12983 * graphite-scop-detection.c: Include cfganal.h.
12984 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12986 (gather_bbs::before_dom_children): Return STOP for BBs
12988 (build_scops): Compute bb to RPO mapping and pass it to
12989 the domwalk. Treat --param graphite-max-nb-scop-params=0
12990 as not limiting the number of params.
12991 * graphite.c (graphite_initialize): Remove limit on the
12992 number of basic-blocks in a function.
12993 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12994 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12995 default value of 10.
12997 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12999 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13000 Adjust code to eliminate needing to do the shift right 32-bits
13001 operation after XSCVDPSPN.
13003 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13005 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13006 ((X / Y) != 0 -> X >= Y): Likewise.
13008 2017-09-26 Carl Love <cel@us.ibm.com>
13010 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13011 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13012 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13013 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13014 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13015 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13016 definitions and overloading.
13017 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13018 statement for P9V_BUILTIN_XST_LEN_R.
13019 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13020 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13021 define_expand and define_insn for the instructions and builtins.
13022 * doc/extend.texi: Update the built-in documentation file for the new
13023 built-in functions.
13024 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13025 define_insn for the instructions
13027 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13030 * gcc/config/netbsd-protos.h: New file.
13031 * gcc/config/netbsd.c: New file.
13032 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13033 * gcc/config/t-netbsd: New file.
13034 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13035 (tmake_file) Add t-netbsd.
13036 (extra_objs) Add netbsd.o.
13038 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13042 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13044 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13046 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13047 sign extension from a vector register to a GPR by doing a 32-bit
13048 direct move and then an EXTSW.
13049 (extendsi<mode>2 splitter): Likewise.
13050 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13051 right or vector extract after doing XSCVDPSPN. Use
13052 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13054 (movdi_from_sf_zero_ext): Likewise.
13055 (reload_gpr_from_vsxsf): Likewise.
13056 (p8_mfvsrd_4_disf): Delete, no longer used.
13057 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13058 then needing to move the SFmode to a GPR to use the XSCVDPSP
13059 instruction instead of FRSP and XSCVDPSPN.
13060 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13061 it is adjacent to the other XSCVSPDP insns.
13062 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13063 SFmode to be in traditional Altivec registers.
13064 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13065 (vsx_xscvspdpn): Likewise.
13066 (vsx_xscvspdpn_scalar): Likewise.
13068 2017-09-26 Martin Jambor <mjambor@suse.cz>
13070 * tree-sra.c (compare_access_positions): Put integral types first,
13071 stabilize sorting of integral types, remove conditions putting
13072 non-full-precision integers last.
13073 (sort_and_splice_var_accesses): Disable scalarization if a
13074 non-integert would be represented by a non-full-precision integer.
13076 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13078 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13079 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13080 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13081 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13082 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13083 conditionals inside the function instead of around it. Call
13084 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13085 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13087 2017-09-26 Richard Biener <rguenther@suse.de>
13089 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13091 (scop_detection::build_scop_breadth): ... this. Removed.
13092 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13093 (scop_detection::harmful_stmt_in_bb): Likewise.
13094 (scop_detection::graphite_can_represent_stmt): Likewise.
13095 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13096 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13097 (scop_detection::can_represent_loop_1): ... this. Removed.
13098 (scop_detection::harmful_loop_in_region): Simplify after inlining
13099 the above and remove more quadraticness.
13100 (build_scops): Adjust.
13101 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13104 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13107 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13108 REGNO (base) == SP_REG if base is a REG.
13110 PR middle-end/35691
13111 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13112 if it is different SSA_NAME.
13113 (optimize_range_tests_cmp_bitwise): New function.
13114 (optimize_range_tests): Call it.
13116 2017-09-26 Richard Biener <rguenther@suse.de>
13118 PR tree-optimization/82321
13119 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13120 for the def being inside the loop.
13122 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13124 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13126 * config/s390/s390-builtins.def: Fix constraint on op4.
13128 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13130 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13131 independent expanders.
13132 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13133 ("vec_ordered", "vec_unordered"): New expanders.
13135 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13137 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13140 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13142 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13143 vec_unpacks_lo_v16qi.
13144 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13146 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13148 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13149 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13150 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13152 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13154 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13156 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13158 ("*vec_slb<mode>"): New insn pattern.
13159 ("vec_shr_<mode>"): New expander.
13160 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13161 and force the shift count operand to V16QImode.
13162 ("vec_srb<mode>"): Set shift count mode to V16QI.
13164 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13166 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13167 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13168 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13170 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13173 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13175 2017-09-26 Richard Biener <rguenther@suse.de>
13177 PR tree-optimization/82320
13178 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13181 2017-09-25 Jeff Law <law@redhat.com>
13183 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13184 prototype for new argument.
13185 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13186 mostly extracted from rs6000_emit_allocate_stack.
13187 (rs6000_emit_probe_stack_range_stack_clash): New function.
13188 (rs6000_emit_allocate_stack): Call
13189 rs6000_emit_probe_stack_range_stack_clash as needed.
13190 (rs6000_emit_probe_stack_range): Add additional argument
13191 to call to gen_probe_stack_range{si,di}.
13192 (output_probe_stack_range): New.
13193 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13194 (output_probe_stack_range_stack_clash): New.
13195 (rs6000_emit_prologue): Emit notes into dump file as requested.
13196 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13197 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13198 Add additional operand and pass it to output_probe_stack_range.
13200 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13202 PR tree-optimization/82163
13203 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13204 (checking_verify_loop_closed_ssa): New parameter.
13205 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13206 (check_loop_closed_ssa_stmt): Delete.
13207 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13208 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13209 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13212 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13214 * brig-builtins.def: Treat HSAIL barrier builtins as
13215 setjmp/longjump style functions.
13217 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13219 * target.def (constant_alignment): New hook.
13220 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13221 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13222 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13223 * doc/tm.texi: Regenerate.
13224 * targhooks.h (default_constant_alignment): Declare.
13225 (constant_alignment_word_strings): Likewise.
13226 * targhooks.c (default_constant_alignment): New function.
13227 (constant_alignment_word_strings): Likewise.
13228 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13229 instead of CONSTANT_ALIGNMENT.
13230 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13231 (force_const_mem): Likewise.
13232 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13233 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13234 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13235 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13236 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13238 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13239 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13240 constant_alignment_word_strings.
13241 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13242 (CONSTANT_ALIGNMENT): Likewise.
13243 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13244 (arm_constant_alignment): New function.
13245 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13246 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13247 constant_alignment_word_strings.
13248 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13249 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13250 constant_alignment_word_strings.
13251 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13252 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13253 (cris_constant_alignment): New function.
13254 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13255 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13256 (epiphany_constant_alignment): New function.
13257 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13258 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13259 constant_alignment_word_strings.
13260 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13261 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13262 constant_alignment_word_strings.
13263 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13264 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13265 constant_alignment_word_strings.
13266 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13267 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13268 * config/i386/i386.c (ix86_constant_alignment): Make static.
13269 Use the same interface as the target hook.
13270 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13271 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13272 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13273 constant_alignment_word_strings.
13274 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13275 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13276 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13277 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13278 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13279 constant_alignment_word_strings.
13280 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13281 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13282 constant_alignment_word_strings.
13283 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13284 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13285 constant_alignment_word_strings.
13286 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13287 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13289 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13290 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13291 * config/mips/mips.c (mips_constant_alignment): New function.
13292 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13293 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13294 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13295 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13296 (mmix_constant_alignment): Make static. Use the same interface
13297 as the target hook.
13298 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13299 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13300 constant_alignment_word_strings.
13301 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13302 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13303 constant_alignment_word_strings.
13304 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13305 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13306 constant_alignment_word_strings.
13307 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13308 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13309 (rs6000_constant_alignment): New function.
13310 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13311 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13312 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13313 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13314 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13315 (rs6000_constant_alignment): New function.
13316 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13317 * config/s390/s390.c (s390_constant_alignment): New function.
13318 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13319 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13320 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13321 constant_alignment_word_strings.
13322 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13323 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13324 (sparc_constant_alignment): New function.
13325 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13326 * config/spu/spu.c (spu_constant_alignment): New function.
13327 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13328 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13329 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13330 constant_alignment_word_strings.
13331 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13332 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13333 constant_alignment_word_strings.
13334 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13335 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13336 constant_alignment_word_strings.
13337 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13338 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13339 (visium_constant_alignment): New function.
13340 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13341 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13342 (xtensa_constant_alignment): New function.
13343 * system.h (CONSTANT_ALIGNMENT): Poison.
13345 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13347 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13348 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13349 (rs6000_builtin_valid_without_lhs): New helper function.
13350 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13351 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13353 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13355 * target.h (vec_perm_indices): Use unsigned short rather than
13357 (auto_vec_perm_indices): Likewise.
13358 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13359 Use unsigned int rather than unsigned char.
13360 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13362 2017-09-25 Richard Biener <rguenther@suse.de>
13364 * cfgloop.h (sort_sibling_loops): Declare.
13365 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13366 (sort_sibling_loops): New function sorting the sibling loop list
13368 * graphite.c (graphite_transform_loops): Sort sibling loops.
13370 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13372 * target.def (vec_perm_const_ok): Change sel parameter to
13374 * optabs-query.c (can_vec_perm_p): Update accordingly.
13375 * doc/tm.texi: Regenerate.
13376 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13377 auto_vec_perm_indices and remove separate nelt field.
13378 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13379 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13380 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13381 (aarch64_expand_vec_perm_const): Update accordingly.
13382 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13383 to vec_perm_indices.
13384 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13385 auto_vec_perm_indices and remove separate nelt field.
13386 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13387 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13388 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13390 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13391 to vec_perm_indices.
13392 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13393 sel to vec_perm_indices.
13394 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13395 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13396 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13398 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13401 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13404 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13405 on the FUNCTION_DECL function context if it has a DIE that is a
13408 2017-09-25 Richard Biener <rguenther@suse.de>
13410 PR tree-optimization/82285
13411 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13414 2017-09-25 Tom de Vries <tom@codesourcery.com>
13418 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13421 2017-09-25 Richard Biener <rguenther@suse.de>
13423 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13424 ISL errors other than isl_error_quota happen. Dump if the
13425 schedule is the same.
13426 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13427 errors instead of aborting inside ISL.
13429 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13432 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13433 of libgcc_eh for m64.
13434 * config/i386/darwin64.h: Likewise.
13436 2017-09-25 Richard Biener <rguenther@suse.de>
13438 PR middle-end/82144
13439 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
13440 attribute for incomplete types nor twice for complete ones.
13442 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
13445 * config/i386/i386.c (ix86_print_operand_address_as): Encode
13446 %esp as %rsp to avoid 0x67 prefix if there is no index or base
13449 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
13452 * config/i386/i386.opt (mprefer-avx256): Use
13453 ix86_target_flags variable.
13454 * config/i386/i386.c (ix86_target_string): Move
13455 -mprefer-avx256 to flag2_opts.
13457 2017-09-22 Jakub Jelinek <jakub@redhat.com>
13459 PR middle-end/35691
13460 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
13461 and x != -1 | y != -1 into (x & y) != -1.
13463 2017-09-22 Steve Ellcey <sellcey@cavium.com>
13465 * config.gcc: Add new case statement to set
13466 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
13467 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
13468 s390x-*-linux* case statements. Added aarch64 to the list of
13469 supported architectures.
13471 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13473 PR tree-optimization/82289
13474 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
13475 STMT_VINFO_RELEVANT_P.
13477 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13478 Alan Hayward <alan.hayward@arm.com>
13479 David Sherwood <david.sherwood@arm.com>
13481 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
13482 for VR_RANGE only; don't allow VR_ANTI_RANGE.
13483 (extract_range_from_binary_expr_1): Don't call
13484 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
13486 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13487 Alan Hayward <alan.hayward@arm.com>
13488 David Sherwood <david.sherwood@arm.com>
13490 * target.def (preferred_vector_alignment): New hook.
13491 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
13493 * doc/tm.texi: Regenerate.
13494 * targhooks.h (default_preferred_vector_alignment): Declare.
13495 * targhooks.c (default_preferred_vector_alignment): New function.
13496 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
13498 (DR_TARGET_ALIGNMENT): New macro.
13499 (aligned_access_p): Update commentary.
13500 (vect_known_alignment_in_bytes): New function.
13501 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
13503 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
13504 Calculate the misalignment based on the target alignment rather than
13506 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
13507 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
13508 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
13509 the target alignment, rather than masking the element misalignment
13510 with the number of elements in a vector. Also use the target
13511 alignment when calculating the maximum number of peels.
13512 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
13513 instead of TYPE_ALIGN_UNIT.
13514 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
13515 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
13516 (vect_create_addr_base_for_vector_ref): Update call accordingly.
13517 (vect_create_data_ref_ptr): Likewise.
13518 (vect_setup_realignment): Realign by ANDing with
13519 -DR_TARGET_MISALIGNMENT.
13520 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
13521 the number of peels based on DR_TARGET_ALIGNMENT.
13522 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
13523 with the guaranteed alignment boundary when deciding whether
13525 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
13526 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
13527 (ensure_base_align): Remove stmt_info parameter. Get the
13528 target base alignment from DR_TARGET_ALIGNMENT.
13529 (vectorizable_store): Update call accordingly. Interpret
13530 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
13532 (vectorizable_load): Likewise.
13534 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
13535 Alan Hayward <alan.hayward@arm.com>
13536 David Sherwood <david.sherwood@arm.com>
13538 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
13539 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
13540 (vect_enhance_data_refs_alignment): Likewise.
13542 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13544 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
13545 error. Only quit immediately if parsing is complete.
13546 (BEGIN): Initialize fatal_err and parse_done.
13547 (begin fpu, end fpu): Check number of arguments.
13548 (begin arch, end arch): Likewise.
13549 (begin cpu, end cpu): Likewise.
13550 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
13551 (optalias): Likewise.
13553 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
13555 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
13556 * config/arm/arm-isa.h: Delete. Move definitions to ...
13557 * arm-cpus.in: ... here. Use new feature and fgroup values.
13558 * config/arm/arm.c (arm_option_override): Use lower case for feature
13560 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
13561 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
13562 * config/arm/parsecpu.awk (END): Add new command 'isa'.
13564 (print_isa_bits_for): New function.
13565 (gen_isa): New function.
13566 (gen_comm_data): Use print_isa_bits_for.
13567 (define feature): New keyword.
13568 (define fgroup): New keyword.
13569 * config/arm/t-arm (TM_H): Remove.
13570 (GTM_H): Add arm-isa.h.
13571 (arm-isa.h): Add rule to generate file.
13572 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
13573 case for feature bit names.
13575 2017-09-22 Richard Biener <rguenther@suse.de>
13577 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
13579 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
13580 print of no dependency loops ...
13581 * graphite.c (graphite_transform_loops): ... here.
13582 (canonicalize_loop_closed_ssa_form): Work from inner to outer
13584 (same_close_phi_node, remove_duplicate_close_phi,
13585 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
13586 (canonicalize_loop_closed_ssa): ... here and simplify.
13587 * graphite-optimize-isl.c: Include tree-vectorizer.h.
13588 (optimize_isl): Use dump_printf_loc to tell when we stopped
13589 optimizing because of an ISL timeout.
13591 2017-09-22 Richard Biener <rguenther@suse.de>
13593 PR tree-optimization/82291
13594 * tree-if-conv.c (predicate_mem_writes): Make sure to
13595 remove writes in blocks predicated with false.
13597 2017-09-22 Richard Biener <rguenther@suse.de>
13599 * sese.c: Include cfganal.h.
13600 (if_region_set_false_region): Remove.
13601 (create_if_region_on_edge): Likewise.
13602 (move_sese_in_condition): Re-implement without destroying
13605 2017-09-22 Richard Biener <rguenther@suse.de>
13607 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
13608 Verify both BBs contain loop PHI nodes before dispatching to
13609 copy_loop_phi_args.
13610 (graphite_regenerate_ast_isl): Do not recompute dominators,
13611 do not verify three times. Restructure for clarity.
13612 * graphite-scop-detection.c (same_close_phi_node,
13613 remove_duplicate_close_phi, make_close_phi_nodes_unique,
13614 defined_in_loop_p, canonicalize_loop_closed_ssa,
13615 canonicalize_loop_closed_ssa_form): Simplify, remove excess
13616 checking and SSA rewrite, move to ...
13617 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
13618 (graphite_initialize): Do not pass in ctx, do not reset the
13619 SCEV cache, compute only dominators.
13620 (graphite_transform_loops): Allocate ISL ctx after
13621 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
13622 Maintain post-dominators only around build_scops.
13623 * sese.c (if_region_set_false_region): Make static. Free
13624 and recompute dominators.
13625 (move_sese_in_condition): Assert we don't get called with
13626 post-dominators computed.
13627 * sese.h (if_region_set_false_region): Remove.
13629 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
13631 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
13632 mode attribute for TARGET_AVX512VL.
13634 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
13636 * config/i386/i386.opt (mprefer-avx256): New option.
13637 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
13639 (ix86_preferred_simd_mode): Return 256-bit AVX modes
13640 for TARGET_PREFER_AVX256.
13641 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
13643 2017-09-21 Jeff Law <law@redhat.com>
13645 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
13646 Fix dump output if the only stack space is for pushed registers.
13648 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13650 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
13653 2017-09-21 Martin Sebor <msebor@redhat.com>
13656 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
13657 code (in C++) or code that triggers warnings.
13659 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
13661 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
13663 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
13665 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
13666 * sched-rgn.c: Ditto.
13667 * sel-sched-ir.c: Ditto.
13669 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
13671 * toplev.h (set_random_seed): Adjust return type.
13672 * toplev.c (init_local_tick): Move eager initialization of random_seed
13673 to get_random_seed. Adjust comment.
13674 (init_random_seed): Inline to get_random_seed, delete.
13675 (get_random_seed): Initialize random_seed lazily.
13676 (set_random_seed): Do not return previous value.
13677 (print_switch_value): Do not call get_random_seed.
13679 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
13681 * cgraph.c (delete_function_version): New, broken out from...
13682 (cgraph_node::delete_function_version): ...here. Rename to
13683 cgraph_node::delete_function_version_by_decl. Update all uses.
13684 (cgraph_node::remove): Call delete_function_version.
13686 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13689 * tree-inline.c (expand_call_inline): Emit clobber stmts for
13690 VAR_DECLs to which addressable non-volatile parameters are mapped
13691 and for id->retvar after the return value assignment. Clear
13692 id->retval and id->retbnd after inlining.
13694 2017-09-21 Richard Biener <rguenther@suse.de>
13696 PR tree-optimization/82276
13697 PR tree-optimization/82244
13698 * tree-vrp.c (build_assert_expr_for): Set
13699 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
13701 (remove_range_assertions): Revert earlier change.
13703 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
13706 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
13708 2017-09-21 Richard Biener <rguenther@suse.de>
13710 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
13711 Restore valid IL after code generation errors.
13712 * graphite.c (graphite_transform_loops): Diagnose code
13713 generation issues as MSG_MISSED_OPTIMIZATION and continue
13714 with processing SCOPs.
13716 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13717 Alan Hayward <alan.hayward@arm.com>
13718 David Sherwood <david.sherwood@arm.com>
13720 * calls.c (compute_argument_addresses): Use simplify_gen_binary
13721 rather than choosing between plus_constant and gen_rtx_<CODE>.
13722 * expr.c (emit_push_insn): Likewise.
13723 (expand_expr_real_2): Likewise.
13725 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13726 Alan Hayward <alan.hayward@arm.com>
13727 David Sherwood <david.sherwood@arm.com>
13729 * loop-unroll.c (split_iv): Call copy_rtx on the step.
13731 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13732 Alan Hayward <alan.hayward@arm.com>
13733 David Sherwood <david.sherwood@arm.com>
13735 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
13736 calling tree_to_uhwi.
13738 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13739 Alan Hayward <alan.hayward@arm.com>
13740 David Sherwood <david.sherwood@arm.com>
13742 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
13743 INTEGER_CST rather than a negative test for ADDR_EXPR.
13745 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
13746 Alan Hayward <alan.hayward@arm.com>
13747 David Sherwood <david.sherwood@arm.com>
13749 * tree-vrp.c (extract_range_from_binary_expr_1): Check
13750 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
13752 2017-09-21 Richard Biener <rguenther@suse.de>
13754 PR tree-optimization/71351
13755 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
13756 graphite_create_new_loop_guard): Remove, fold remaining parts
13758 (translate_isl_ast_node_for): ... here and simplify.
13760 2017-09-21 Jakub Jelinek <jakub@redhat.com>
13763 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
13764 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
13765 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
13766 alternative always use QI mode, for -Os imov (=R,R) alternative
13767 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13770 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13771 Jeff Law <law@redhat.com>
13773 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13774 (allocate_stack_space): New function, partially extracted from
13775 s390_emit_prologue.
13776 (s390_emit_prologue): Track offset to most recent stack probe.
13777 Code to allocate space moved into allocate_stack_space.
13778 Dump actions when no stack is allocated.
13779 (s390_prologue_plus_offset): New function.
13780 (s390_emit_stack_probe): Likewise.
13782 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13784 * common.opt (Wa, Wl, Wp, g, gz=): Add
13786 (gno-column-info): Remove.
13787 (gcolumn-info): Drop RejectNegative.
13788 (gno-): New prefix.
13789 (gno-record-gcc-switches): Remove.
13790 (grecord-gcc-switches): Drop RejectNegative.
13791 (gno-split-dwarf): Remove.
13792 (gsplit-dwarf): Drop RejectNegative.
13793 (gno-strict-dwarf): Remove.
13794 (gstrict-dwarf): Drop RejectNegative.
13795 * config/darwin.opt (gfull, gused): Add RejectNegative.
13796 * dwarf2out.c (gen_producer_string): Drop
13797 gno-record-gcc-switches handler.
13798 * optc-gen.awk: Add g to prefixes with negative forms.
13799 * opts-common.c (remapping_prefix_p): New.
13800 (find_opt): Check it.
13801 (generate_canonical_option): Test g prefix.
13802 (option_map): Add -gno- mapping.
13803 (add_misspelling_candidates): Check remapping_prefix_p.
13805 2017-09-20 Jeff Law <law@redhat.com>
13807 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13808 thinko in stack clash protection support.
13810 * explow.c (compute_stack_clash_protection_loop_data): Use
13811 CONST_INT_P instead of explicit test. Verify object is a
13812 CONST_INT_P before looking at INTVAL.
13813 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13814 instead of explicit test.
13816 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13819 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13820 address instead of to r1 and r11.
13822 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13824 * config.gcc: Support "knm".
13825 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13826 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13828 * config/i386/i386.c (m_KNM): Define.
13829 (processor_target_table): Add "knm".
13831 (ix86_option_override_internal): Add "knm".
13832 (ix86_issue_rate): Add PROCESSOR_KNM.
13833 (ix86_adjust_cost): Ditto.
13834 (ia32_multipass_dfa_lookahead): Ditto.
13835 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13836 (fold_builtin_cpu): Add M_INTEL_KNM.
13837 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13838 (processor_type): Add PROCESSOR_KNM.
13839 * config/i386/x86-tune.def: Add m_KNM.
13840 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13842 2017-09-20 Richard Biener <rguenther@suse.de>
13844 PR tree-optimization/80213
13845 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13846 are allowed in empty BBs as well.
13847 (canonicalize_loop_closed_ssa): Also look for other complex
13849 (scop_detection::get_sese): Include the loop-closed PHI block
13851 (scop_detection::merge_sese): Remove code adding extra blocks.
13852 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13853 (build_scops): Assert the final returned scop is invalid.
13855 2017-09-20 Richard Biener <rguenther@suse.de>
13857 PR tree-optimization/82264
13858 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13860 (vn_phi_lookup): Likewise.
13861 (vn_phi_insert): Likewise.
13863 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13865 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13866 that fits into uhwi or shwi, add DW_AT_const_value regardless
13867 of early_dwarf without going through RTL, using add_AT_unsigned
13870 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13871 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13872 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13873 (DEBUG_MACRO_SECTION): Likewise.
13874 (DEBUG_LTO_MACRO_SECTION): Likewise.
13875 (DEBUG_STR_DWO_SECTION): Likewise.
13876 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13877 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13878 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13879 (DEBUG_LTO_LINE_STR_SECTION): Define.
13880 (init_sections_and_labels): Initialize debug_line_str_section
13881 variable. Initialize debug_loc_section for -gdwarf-5 to
13882 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13884 2017-09-20 Richard Biener <rguenther@suse.de>
13886 * graphite-sese-to-poly.c (extract_affine): Properly handle
13887 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13889 2017-09-20 Richard Biener <rguenther@suse.de>
13891 PR tree-optimization/81373
13892 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13893 Force SESE live-out defs to be handled even if they are
13896 2017-09-19 Jeff Law <law@redhat.com>
13898 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13899 nothing for stack adjustments with REG_STACK_CHECK.
13900 * sched-deps.c (parse_add_or_inc): Reject insns with
13901 REG_STACK_CHECK from dependency breaking.
13902 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13903 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13904 * reg-notes.def (STACK_CHECK): New note.
13906 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13907 (ix86_expand_prologue): Dump stack clash info as needed.
13908 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13910 * function.c (dump_stack_clash_frame_info): New function.
13911 * function.h (dump_stack_clash_frame_info): Prototype.
13912 (enum stack_clash_probes): New enum.
13914 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13915 flag_stack_clash_protection.
13916 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13917 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13918 (arm_frame_pointer_required): Likewise.
13919 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13920 (ia64_expand_prologue): Likewise.
13921 * config/mips/mips.c (mips_expand_prologue): Likewise.
13922 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13923 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13924 (sparc_flat_expand_prologue): Likewise.
13925 * config/spu/spu.c (spu_expand_prologue): Likewise.
13927 * explow.c: Include "params.h".
13928 (anti_adjust_stack_and_probe_stack_clash): New function.
13929 (get_stack_check_protect): Likewise.
13930 (compute_stack_clash_protection_loop_data): Likewise.
13931 (emit_stack_clash_protection_loop_start): Likewise.
13932 (emit_stack_clash_protection_loop_end): Likewise.
13933 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13934 Use anti_adjust_stack_and_probe_stack_clash.
13935 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13936 (emit_stack_clash_protection_loop_start): Likewise.
13937 (emit_stack_clash_protection_loop_end): Likewise.
13938 * rtl.h (get_stack_check_protect): Prototype.
13939 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13940 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13941 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13943 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13945 * doc/tm.texi: Rebuilt.
13946 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13947 get_stack_check_protect.
13948 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13949 * config/arm/arm.c (arm_expand_prologue): Likewise.
13950 (arm_frame_pointer_required): Likewise.
13951 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13952 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13953 * config/mips/mips.c (mips_expand_prologue): Likewise.
13954 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13955 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13956 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13957 (sparc_flat_expand_prologue): Likewise.
13959 * common.opt (-fstack-clash-protection): New option.
13960 * flag-types.h (enum stack_check_type): Note difference between
13961 -fstack-check= and -fstack-clash-protection.
13962 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13963 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13964 * toplev.c (process_options): Issue warnings/errors for cases
13965 not handled with -fstack-clash-protection.
13966 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13967 (-fstack-check): Note additional problem with -fstack-check=generic.
13968 Note that -fstack-check is primarily for Ada and refer users
13969 to -fstack-clash-protection for stack-clash-protection.
13970 Document new params for stack clash protection.
13972 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13974 * config/i386/i386.c (ix86_split_long_move): Do not handle
13975 address used for LEA in a special way.
13977 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13979 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13981 2017-09-19 Martin Sebor <msebor@redhat.com>
13984 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13985 of incompatible types.
13987 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13989 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13990 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13991 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13992 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13994 2017-09-19 Richard Biener <rguenther@suse.de>
13996 PR tree-optimization/82244
13997 * tree-vrp.c (remove_range_assertions): Do not propagate
13998 a constant to abnormals but replace the assert with a copy.
14000 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14002 PR rtl-optimization/57878
14003 PR rtl-optimization/68988
14004 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14005 avoidance test involving non_reload_pseudos. Move frequency test
14006 below the general fragmentation avoidance test.
14008 2017-09-19 Richard Biener <rguenther@suse.de>
14010 PR tree-optimization/69728
14011 * graphite-sese-to-poly.c (schedule_error): New global.
14012 (add_loop_schedule): Handle empty domain by failing the
14014 (build_original_schedule): Handle schedule_error.
14016 2017-09-19 Richard Biener <rguenther@suse.de>
14018 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14019 Do not iterate to sibling loops but only to siblings of inner
14022 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14025 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14026 that the registers are different.
14028 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14030 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14031 to processor_model and "amdfam17h" to arch_names_table.
14032 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14034 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14037 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14039 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14040 Alan Hayward <alan.hayward@arm.com>
14041 David Sherwood <david.sherwood@arm.com>
14043 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14045 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14047 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14048 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14049 vect_schedule_slp_instance.
14050 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14051 Update call to vect_slp_analyze_node_operations. Simplify return
14053 (vect_slp_analyze_bb_1): Update call accordingly.
14054 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14055 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14056 (vect_schedule_slp): Update call accordingly.
14058 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14059 Alan Hayward <alan.hayward@arm.com>
14060 David Sherwood <david.sherwood@arm.com>
14062 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14063 with types that aren't in fact scalar.
14065 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14067 * tree-vect-slp.c (vect_record_max_nunits): New function,
14069 (vect_build_slp_tree_1): ...here.
14070 (vect_build_slp_tree_2): Call it for phis too.
14072 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14074 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14075 to vect_get_vec_def_for_operand when getting the mask operand.
14077 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14078 Alan Hayward <alan.hayward@arm.com>
14079 David Sherwood <david.sherwood@arm.com>
14081 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14084 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14085 Alan Hayward <alan.hayward@arm.com>
14086 David Sherwood <david.sherwood@arm.com>
14088 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14089 calculation for vector booleans.
14091 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14092 Alan Hayward <alan.hayward@arm.com>
14093 David Sherwood <david.sherwood@arm.com>
14095 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14097 (vect_transform_stmt): ...here.
14098 (vect_analyze_stmt): Use it instead of calling
14099 vectorizable_live_operation directly.
14101 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14103 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14104 non-SIMT targets in acc vector loops.
14106 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14108 * configure.ac: Add arc and check if assembler supports gdwarf2.
14109 * configure: Regenerate.
14111 2017-09-18 Richard Biener <rguenther@suse.de>
14113 PR tree-optimization/82220
14114 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14115 epilogue niters from the min_profitable_iters compute.
14117 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14120 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14121 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14122 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14124 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14127 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14128 switching to a new text section.
14130 2017-09-18 Richard Biener <rguenther@suse.de>
14132 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14134 (build_alias_set): Reject aliases with no access function.
14136 2017-09-18 Richard Biener <rguenther@suse.de>
14138 PR tree-optimization/79622
14139 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14141 (build_cross_bb_scalars_use): Likewise.
14143 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14145 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14147 2017-09-18 Alan Modra <amodra@gmail.com>
14150 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14151 stack_pointer_rtx for count 0. Update comments. Break up
14152 large rtl expression.
14154 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14156 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14157 Increase to 20 bytes.
14158 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14159 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14160 or avx version of the stub.
14162 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14165 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14166 compute the minimum stack alignment. Also update preferred stack
14167 boundary for leaf functions.
14169 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14171 PR tree-optimization/82228
14172 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14175 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14177 * common/config/nds32/nds32-common.c
14178 (nds32_option_optimization_table): Refine formatting.
14179 (nds32_option_optimization_table): Use -fsched-pressure and
14180 -fomit-frame-pointer for specific optimization level.
14182 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14184 * config/nds32/nds32.c: Refine formatting and comments.
14185 * config/nds32/nds32.h: Likewise.
14186 * config/nds32/nds32.md: Likewise.
14187 * config/nds32/nds32-cost.c: Likewise.
14188 * config/nds32/nds32-isr.c: Likewise.
14189 * config/nds32/nds32-md-auxiliary.c: Likewise.
14190 * config/nds32/nds32-multiple.md: Likewise.
14191 * config/nds32/nds32-predicates.c: Likewise.
14193 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14194 Jakub Jelinek <jakub@redhat.com>
14196 Add support for -std=c++2a.
14197 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14199 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14201 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14204 * doc/extend.texi (Common Function Attributes): Add
14205 references to ARM, AArch64, and S/390 specific attributes.
14206 (Function Specific Option Pragmas): Add AArch64 and S/390
14207 to list of back ends that support the target pragma.
14209 2017-09-15 Nathan Sidwell <nathan@acm.org>
14211 * doc/standards.texi: Fix C++17 description. Update URLs for
14214 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14216 * common.opt (Wcast-align=strict): New warning option.
14217 * doc/invoke.texi: Document -Wcast-align=strict.
14219 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14221 * cgraph.h (cgraph_thunk_info): Add comments.
14222 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14223 assert for VIRTUAL_* arguments stricter.
14225 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14227 PR tree-optimization/71026
14228 * match.pd: Move RDIV patterns from fold-const.c
14229 * fold-const.c (distribute_real_division): Removed.
14230 (fold_binary_loc): Remove calls to distribute_real_divison.
14232 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14234 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14235 c++1z and gnu++1z as deprecated. Change other references to
14236 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14237 Change -Wc++1z-compat to -Wc++17-compat.
14238 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14239 * dwarf2out.c (highest_c_language): Handle C++17.
14240 (gen_compile_unit_die): Likewise.
14242 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14244 PR rtl-optimization/82192
14245 * combine.c (make_extraction): Don't look through non-paradoxical
14246 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14249 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14250 Alan Hayward <alan.hayward@arm.com>
14251 David Sherwood <david.sherwood@arm.com>
14253 * target.def (function_arg_offset): New hook.
14254 * targhooks.h (default_function_arg_offset): Declare.
14255 * targhooks.c (default_function_arg_offset): New function.
14256 * function.c (locate_and_pad_parm): Use
14257 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14258 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14259 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14260 * doc/tm.texi: Regenerate.
14261 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14262 * config/spu/spu.c (spu_function_arg_offset): New function.
14263 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14264 * system.h (FUNCTION_ARG_OFFSET): Poison.
14266 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14267 Alan Hayard <alan.hayward@arm.com>
14268 David Sherwood <david.sherwood@arm.com>
14270 * target.def (truly_noop_truncation): New hook.
14271 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14272 than TRULY_NOOP_TRUNCATION.
14273 * hooks.h (hook_bool_uint_uint_true): Declare.
14274 * hooks.c (hook_bool_uint_uint_true): New function.
14275 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14276 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14277 * doc/tm.texi: Regenerate.
14278 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14279 rather than TRULY_NOOP_TRUNCATION in comments.
14280 (simplify_comparison): Likewise.
14281 (record_truncated_value): Likewise.
14282 * expmed.c (extract_bit_field_1): Likewise.
14283 (extract_split_bit_field): Likewise.
14284 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14285 instead of TRULY_NOOP_TRUNCATION.
14286 * function.c (assign_parm_setup_block): Likewise.
14287 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14288 * rtlhooks.c: Include target.h.
14289 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14290 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14291 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14292 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14293 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14294 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14295 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14296 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14297 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14298 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14299 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14300 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14301 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14302 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14303 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14304 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14305 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14306 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14307 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14308 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14309 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14310 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14311 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14312 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14313 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14314 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14315 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14316 rather than TRULY_NOOP_TRUNCATION in comments.
14317 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14318 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14319 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14320 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14321 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14322 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14323 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14324 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14325 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14326 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14327 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14328 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14329 rather than TRULY_NOOP_TRUNCATION in comments.
14330 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14331 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14332 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14333 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14334 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14335 TRULY_NOOP_TRUNCATION condition.
14336 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14337 (TRULY_NOOP_TRUNCATION): Delete.
14338 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14339 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14340 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14341 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14342 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14343 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14344 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14345 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14346 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14347 rather than TRULY_NOOP_TRUNCATION in comments.
14348 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14349 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14350 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14351 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14352 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14353 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14355 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14358 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14359 (*cmp_ior): Likewise.
14360 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14361 (*ior_scc_scc_cmp): Likewise.
14362 (*and_scc_scc): Likewise.
14363 (*and_scc_scc_cmp): Likewise.
14365 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14366 Alan Hayard <alan.hayward@arm.com>
14367 David Sherwood <david.sherwood@arm.com>
14369 * target.def (can_change_mode_class): New hook.
14370 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14371 (hard_regno_nregs): Likewise.
14372 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14373 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14374 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14375 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14376 (LOAD_EXTEND_OP): Update accordingly.
14377 * doc/tm.texi: Regenerate.
14378 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14379 CANNOT_CHANGE_MODE_CLASS.
14380 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14381 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14382 * combine.c (simplify_set): Update accordingly.
14383 * emit-rtl.c (validate_subreg): Likewise.
14384 * recog.c (general_operand): Likewise.
14385 * regcprop.c (mode_change_ok): Likewise.
14386 * reload1.c (choose_reload_regs): Likewise.
14387 (inherit_piecemeal_p): Likewise.
14388 * rtlanal.c (simplify_subreg_regno): Likewise.
14389 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14390 instead of CANNOT_CHANGE_MODE_CLASS.
14391 (reload_cse_simplify_operands): Likewise.
14392 * reload.c (push_reload): Use targetm.can_change_mode_class
14393 instead of CANNOT_CHANGE_MODE_CLASS.
14394 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14395 REG_CANNOT_CHANGE_MODE_P.
14396 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14397 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14398 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14399 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14400 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14401 (arm_can_change_mode_class): New function.
14402 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14403 than CANNOT_CHANGE_MODE_CLASS in comments.
14404 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14405 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14406 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14407 (ix86_can_change_mode_class): ...this new function, inverting the
14408 sense of the return value.
14409 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14410 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14411 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14412 (ia64_can_change_mode_class): New function.
14413 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14414 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14415 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14416 (m32c_can_change_mode_class): ...this new function, inverting the
14417 sense of the return value.
14418 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14419 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14420 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14421 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14422 (mips_can_change_mode_class): ...this new function, inverting the
14423 sense of the return value.
14424 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14425 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14426 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14427 (msp430_can_change_mode_class): New function.
14428 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14429 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14430 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14431 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14432 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14433 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
14434 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14435 (pa_cannot_change_mode_class): Replace with...
14436 (pa_can_change_mode_class): ...this new function, inverting the
14437 sense of the return value.
14438 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14439 than CANNOT_CHANGE_MODE_CLASS in comments.
14440 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14441 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
14442 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14443 (pdp11_cannot_change_mode_class): Replace with...
14444 (pdp11_can_change_mode_class): ...this new function, inverting the
14445 sense of the return value.
14446 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14447 * config/powerpcspe/powerpcspe-protos.h
14448 (rs6000_cannot_change_mode_class_ptr): Delete.
14449 * config/powerpcspe/powerpcspe.c
14450 (rs6000_cannot_change_mode_class_ptr): Delete.
14451 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14452 (rs6000_option_override_internal): Assign to
14453 targetm.can_change_mode_class instead of
14454 rs6000_cannot_change_mode_class_ptr.
14455 (rs6000_cannot_change_mode_class): Replace with...
14456 (rs6000_can_change_mode_class): ...this new function, inverting the
14457 sense of the return value.
14458 (rs6000_debug_cannot_change_mode_class): Replace with...
14459 (rs6000_debug_can_change_mode_class): ...this new function.
14460 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14461 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
14462 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14463 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14464 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
14466 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
14467 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14468 (rs6000_option_override_internal): Assign to
14469 targetm.can_change_mode_class instead of
14470 rs6000_cannot_change_mode_class_ptr.
14471 (rs6000_cannot_change_mode_class): Replace with...
14472 (rs6000_can_change_mode_class): ...this new function, inverting the
14473 sense of the return value.
14474 (rs6000_debug_cannot_change_mode_class): Replace with...
14475 (rs6000_debug_can_change_mode_class): ...this new function.
14476 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14477 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
14478 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
14479 (s390_can_change_mode_class): ...this new function, inverting the
14480 sense of the return value.
14481 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14482 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14483 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
14484 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14485 (sh_cannot_change_mode_class): Replace with...
14486 (sh_can_change_mode_class): ...this new function, inverting the
14487 sense of the return value.
14488 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14489 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14490 (sparc_can_change_mode_class): New function.
14491 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14492 * config/spu/spu.c (spu_can_change_mode_class): New function.
14493 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14494 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14495 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14496 (visium_can_change_mode_class): New function.
14497 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
14499 2017-09-15 Richard Biener <rguenther@suse.de>
14501 PR tree-optimization/82217
14502 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
14503 but not undefined case.
14505 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14508 * postreload.c (reload_cse_simplify_operands): Skip
14509 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
14511 2017-09-15 Richard Biener <rguenther@suse.de>
14513 PR tree-optimization/68823
14514 * graphite-scop-detection.c (build_alias_set): If we have a
14515 possible dependence check whether we can handle them by just
14516 looking at the DRs DR_ACCESS_FNs.
14517 (build_scops): If build_alias_set fails, fail the SCOP.
14519 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
14521 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
14522 to support float128 built-in functions that require the ISA 3.0
14524 (BU_FLOAT128_3_HW): Likewise.
14525 (SQRTF128): Add support for the IEEE 128-bit square root and fma
14526 built-in functions.
14527 (FMAF128): Likewise.
14529 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14530 support for built-in functions that need the ISA 3.0 IEEE 128-bit
14531 floating point instructions.
14532 (rs6000_invalid_builtin): Likewise.
14533 (rs6000_builtin_mask_names): Likewise.
14534 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
14535 (RS6000_BTM_FLOAT128_HW): Likewise.
14536 (RS6000_BTM_COMMON): Likewise.
14537 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
14539 * doc/extend.texi (RS/6000 built-in functions): Document the
14540 IEEE 128-bit floating point square root and fused multiply-add
14541 built-in functions.
14543 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
14545 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
14546 reg (r2) isn't in the set of registers defined in the prologue.
14548 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14549 Alan Hayward <alan.hayward@arm.com>
14550 David Sherwood <david.sherwood@arm.com>
14552 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
14553 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
14554 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
14555 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
14556 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
14558 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14559 max_vectorization_factor.
14560 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
14562 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14563 Alan Hayward <alan.hayward@arm.com>
14564 David Sherwood <david.sherwood@arm.com>
14566 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
14567 (vect_worthwhile_without_simd_p): Declare.
14568 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
14569 (vectorizable_reduction): Use it.
14570 * tree-vect-stmts.c (vectorizable_shift): Likewise.
14571 (vectorizable_operation): Likewise.
14573 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14574 Alan Hayward <alan.hayward@arm.com>
14575 David Sherwood <david.sherwood@arm.com>
14577 * tree-vectorizer.h (vect_get_num_copies): New function.
14578 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
14579 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14580 (vectorizable_induction): Likewise.
14581 (vectorizable_live_operation): Likewise.
14582 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
14583 (vectorizable_bswap): Likewise.
14584 (vectorizable_call): Likewise.
14585 (vectorizable_conversion): Likewise.
14586 (vectorizable_assignment): Likewise.
14587 (vectorizable_shift): Likewise.
14588 (vectorizable_operation): Likewise.
14589 (vectorizable_store): Likewise.
14590 (vectorizable_load): Likewise.
14591 (vectorizable_condition): Likewise.
14592 (vectorizable_comparison): Likewise.
14593 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
14595 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14596 Alan Hayward <alan.hayward@arm.com>
14597 David Sherwood <david.sherwood@arm.com>
14599 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
14600 of vect_init_vector.
14602 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14603 Alan Hayward <alan.hayward@arm.com>
14604 David Sherwood <david.sherwood@arm.com>
14606 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
14607 an inline wrapper that provides a location.
14608 (gimple_build_vector): Likewise.
14609 * gimple-fold.c (gimple_build_vector_from_val): New function.
14610 (gimple_build_vector): Likewise.
14611 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
14612 functions to build the initial value. Always return a gimple value.
14613 (get_initial_defs_for_reduction): Likewise. Only compute
14615 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
14616 vect_init_vector on the results from get_initial_def(s)_for_reduction.
14617 (vectorizable_induction): Use gimple_build_vector rather than
14620 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14621 Alan Hayward <alan.hayward@arm.com>
14622 David Sherwood <david.sherwood@arm.com>
14624 * target.h (vec_perm_indices): New typedef.
14625 (auto_vec_perm_indices): Likewise.
14626 * optabs-query.h: Include target.h
14627 (can_vec_perm_p): Take a vec_perm_indices *.
14628 * optabs-query.c (can_vec_perm_p): Likewise.
14629 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
14630 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14631 * tree-vect-generic.c (lower_vec_perm): Likewise.
14632 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
14633 (vect_grouped_load_supported): Likewise.
14634 (vect_shift_permute_load_chain): Likewise.
14635 (vect_permute_store_chain): Use auto_vec_perm_indices.
14636 (vect_permute_load_chain): Likewise.
14637 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
14638 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
14639 Update uses of can_vec_perm_p.
14640 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
14641 mode with a number of elements. Take a vec_perm_indices *.
14642 (vect_create_epilog_for_reduction): Update accordingly.
14643 Use auto_vec_perm_indices.
14644 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
14645 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
14646 (vect_transform_slp_perm_load): Likewise.
14647 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
14648 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
14649 (vect_gen_perm_mask_checked): Likewise.
14650 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
14651 (vect_gen_perm_mask_checked): Likewise.
14652 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
14653 (vectorizable_store): Likewise.
14654 (vectorizable_load): Likewise.
14655 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
14656 (vectorizable_bswap): Likewise.
14658 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14659 Alan Hayward <alan.hayward@arm.com>
14660 David Sherwood <david.sherwood@arm.com>
14662 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
14663 * tree.c (build_vector): Likewise.
14664 (build_vector_from_ctor): Update accordingly.
14665 (build_vector_from_val): Likewise.
14666 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14667 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
14668 * tree-vect-generic.c (add_rshift): Likewise.
14669 (expand_vector_divmod): Likewise.
14670 (optimize_vector_constructor): Likewise.
14671 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14672 (vect_transform_slp_perm_load): Likewise.
14673 (vect_schedule_slp_instance): Likewise.
14674 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
14675 (vectorizable_call): Likewise.
14676 (vect_gen_perm_mask_any): Likewise. Add elements in order.
14677 * expmed.c (make_tree): Likewise.
14678 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
14679 a vector passed to build_vector.
14680 (fold_convert_const): Likewise.
14681 (exact_inverse): Likewise.
14682 (fold_ternary_loc): Likewise.
14683 (fold_relational_const): Likewise.
14684 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
14685 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
14686 (const_unop): Likewise. Store the reduction accumulator in a
14687 variable rather than an array.
14688 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
14689 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
14690 the new vector, rather than constructing it after the input arrays.
14691 (native_interpret_vector): Use auto_vec<tree> when building
14692 a vector passed to build_vector. Add elements in order.
14693 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
14694 auto_vec<tree> when building a vector passed to build_vector.
14695 (vect_create_epilog_for_reduction): Likewise.
14696 (vectorizable_induction): Likewise.
14697 (get_initial_def_for_reduction): Likewise. Fix indentation of
14699 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
14701 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
14702 passed to build_vector.
14704 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
14705 Alan Hayward <alan.hayward@arm.com>
14706 David Sherwood <david.sherwood@arm.com>
14708 * tree-core.h (tree_base::u): Add an "nelts" field.
14709 (tree_vector): Use VECTOR_CST_NELTS as the length.
14710 * tree.c (tree_size): Likewise.
14711 (make_vector): Initialize VECTOR_CST_NELTS.
14712 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
14713 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
14714 TYPE_VECTOR_SUBPARTS.
14715 * expr.c (const_vector_mask_from_tree): Consistently use "units"
14716 as the number of units, setting it from VECTOR_CST_NELTS.
14717 (const_vector_from_tree): Likewise.
14718 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
14719 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
14720 (fold_negate_expr_1): Likewise.
14721 (fold_convert_const): Likewise.
14722 (const_binop): Likewise. Differentiate the number of output and
14724 (const_unop): Likewise.
14725 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
14726 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
14727 in cases that did the opposite.
14729 2017-09-14 Richard Biener <rguenther@suse.de>
14731 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
14734 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
14736 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
14738 2017-09-14 Jakub Jelinek <jakub@redhat.com>
14741 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
14742 if and where to split a bb, except for splitting before debug insn
14743 sequences followed by non-label real insn. Delete debug insns
14744 in between basic blocks.
14746 * combine.c (make_compound_operation_int): Formatting fixes.
14748 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
14749 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14750 * config/netbsd.h (LINK_EH_SPEC): Likewise.
14751 * config/sol2.h (LINK_EH_SPEC): Likewise.
14752 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14753 * config/s390/linux.h (LINK_SPEC): Likewise.
14754 * config/freebsd.h (LINK_EH_SPEC): Likewise.
14755 * config/openbsd.h (LINK_EH_SPEC): Likewise.
14756 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14757 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
14758 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
14759 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14760 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
14761 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
14763 2017-09-13 Jakub Jelinek <jakub@redhat.com>
14765 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
14767 (ENDFILE_LINUX_SPEC): Likewise.
14768 (LINK_EH_SPEC): Likewise.
14769 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14770 (LINK_OS_LINUX_SPEC32): Likewise.
14771 (LINK_OS_LINUX_SPEC64): Likewise.
14772 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14773 (LINK_OS_LINUX_SPEC): Likewise.
14775 2017-09-13 Martin Liska <mliska@suse.cz>
14777 PR middle-end/82154
14778 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14779 CASE_HIGH is NULL_TREE.
14781 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14782 Alan Hayward <alan.hayward@arm.com>
14783 David Sherwood <david.sherwood@arm.com>
14785 * target.def (secondary_memory_needed): New hook.
14786 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14787 instead of SECONDARY_MEMORY_NEEDED.
14788 (secondary_memory_needed_mode): Likewise.
14789 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14790 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14791 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14792 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14793 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14794 * doc/tm.texi: Regenerate.
14795 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14796 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14797 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14798 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14799 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14800 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14801 mode argument first and change the reg_class arguments to reg_class_t.
14802 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14803 Make static. Update the call to inline_secondary_memory_needed.
14804 (ix86_register_move_cost): Update the call to
14805 inline_secondary_memory_needed.
14806 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14807 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14809 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14810 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14812 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14813 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14814 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14815 and match hook interface. Add comment from mips.h.
14816 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14817 * config/mmix/mmix.md (truncdfsf2): Refer to
14818 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14820 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14821 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14822 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14823 (pa_secondary_memory_needed): New function.
14824 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14825 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14826 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14827 (pdp11_secondary_memory_needed): Make static and match hook interface.
14828 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14829 * config/powerpcspe/powerpcspe-protos.h
14830 (rs6000_secondary_memory_needed_ptr): Delete.
14831 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14833 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14834 (rs6000_option_override_internal): Assign to
14835 targetm.secondary_memory_needed rather than
14836 rs6000_secondary_memory_needed_ptr.
14837 (rs6000_secondary_memory_needed): Match hook interface.
14838 (rs6000_debug_secondary_memory_needed): Likewise.
14839 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14840 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14841 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14842 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14843 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14844 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14846 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14847 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14848 (rs6000_option_override_internal): Assign to
14849 targetm.secondary_memory_needed rather than
14850 rs6000_secondary_memory_needed_ptr.
14851 (rs6000_secondary_memory_needed): Match hook interface.
14852 (rs6000_debug_secondary_memory_needed): Likewise.
14853 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14854 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14855 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14856 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14857 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14858 (sparc_secondary_memory_needed): New function.
14859 * lra-constraints.c (check_and_process_move): Refer to
14860 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14862 (curr_insn_transform): Likewise.
14863 (process_alt_operands): Use targetm.secondary_memory_needed
14864 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14865 (check_secondary_memory_needed_p): Likewise.
14866 (choose_split_class): Likewise.
14867 * reload.c: Unconditionally include code that was previously
14868 conditional on SECONDARY_MEMORY_NEEDED.
14869 (push_secondary_reload): Use targetm.secondary_memory_needed
14870 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14871 (push_reload): Likewise.
14872 * reload1.c: Unconditionally include code that was previously
14873 conditional on SECONDARY_MEMORY_NEEDED.
14874 (choose_reload_regs): Use targetm.secondary_memory_needed
14875 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14876 (gen_reload): Likewise.
14877 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14879 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14880 Alan Hayward <alan.hayward@arm.com>
14881 David Sherwood <david.sherwood@arm.com>
14883 * target.def (secondary_memory_needed_mode): New hook:
14884 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14885 * targhooks.h (default_secondary_memory_needed_mode): New function.
14886 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14887 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14888 * doc/tm.texi: Regenerate.
14889 * lra-constraints.c (check_and_process_move): Use
14890 targetm.secondary_memory_needed_mode instead of
14891 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14892 (curr_insn_transform): Likewise.
14893 * reload.c (get_secondary_mem): Likewise.
14894 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14895 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14897 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14898 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14899 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14900 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14901 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14903 * config/powerpcspe/powerpcspe-protos.h
14904 (rs6000_secondary_memory_needed_mode): Delete.
14905 * config/powerpcspe/powerpcspe.c
14906 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14907 (rs6000_secondary_memory_needed_mode): Make static.
14908 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14909 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14911 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14913 (rs6000_secondary_memory_needed_mode): Make static.
14914 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14915 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14916 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14917 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14918 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14920 (sparc_secondary_memory_needed_mode): New function.
14921 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14923 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14925 * config/aarch64/constraints.md (Umq): New constraint.
14926 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14928 (mov<mode>): Update condition.
14930 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14932 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14933 when bitposition is the same.
14935 2017-09-13 Richard Biener <rguenther@suse.de>
14937 * dwarf2out.c (output_die_symbol): Remove.
14938 (output_die): Do not output a DIEs symbol.
14940 2017-09-13 Richard Biener <rguenther@suse.de>
14942 PR middle-end/82128
14943 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14944 default-def to avoid breaking iterator update with the weird
14945 interaction with cgraph_update_edges_for_call_stmt_node.
14947 2017-09-13 Richard Biener <rguenther@suse.de>
14949 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14950 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14951 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14952 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14954 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14956 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14957 Disable pc relative literal load irrespective of
14958 TARGET_FIX_ERR_A53_84341 for default.
14960 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14962 * config/sparc/sparc.c (output_return): Output the source location of
14963 the insn in the delay slot, if any.
14964 (output_sibcall): Likewise.
14966 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14969 * common.opt (-static-pie): New alias.
14970 (shared): Negate static-pie.
14971 (-no-pie): Update help text.
14973 (static-pie): New option.
14974 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14975 -static-pie support.
14976 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14977 (LINK_EH_SPEC): Likewise.
14978 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14979 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14980 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14981 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14982 (init_gcc_specs): Likewise.
14983 (init_spec): Likewise.
14984 (display_help): Update help message for -pie.
14985 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14988 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14990 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14991 (movdi_aarch64): Likewise.
14992 (movti_aarch64): Likewise.
14994 2017-09-12 Simon Wright <simon@pushface.org>
14997 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14998 calculation of the minor version, always output as 0.
15000 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15003 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15004 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15005 on it early, rather than manual conversion late. For
15006 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15007 instead of performing manual conversion.
15009 2017-09-12 Carl Love <cel@us.ibm.com>
15011 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15012 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15014 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15015 VMULOSW): Add definitions.
15016 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15017 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15018 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15019 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15020 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15022 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15024 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15026 (movti_aarch64): Likewise.
15027 (movdf_aarch64): Likewise.
15028 (movtf_aarch64): Likewise.
15029 (load_pairdi): Likewise.
15030 (store_pairdi): Likewise.
15031 (load_pairdf): Likewise.
15032 (store_pairdf): Likewise.
15033 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15034 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15035 (ldr_got_small_<mode>): Likewise.
15036 (ldr_got_small_28k_<mode>): Likewise.
15037 (ldr_got_tiny): Likewise.
15038 * config/aarch64/iterators.md (ldst_sz): New.
15039 (ldpstp_sz): Likewise.
15040 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15042 (thunderx_load): Split load_8 to load_16.
15043 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15045 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15046 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15047 (xgene1_store_pair): Split store_8 to store_16.
15048 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15049 (falkor_st_0_st_sd): Split store_8 to store_16.
15051 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15053 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15054 and store1/2/3/4 to store_4/8/12/16.
15055 * config/aarch64/aarch64.md: Update for rename.
15056 * config/arm/arm.md: Likewise.: Likewise.
15057 * config/arm/arm.c: Likewise.
15058 * config/arm/thumb1.md: Likewise.
15059 * config/arm/thumb2.md: Likewise.
15060 * config/arm/vfp.md: Likewise.
15061 * config/arm/arm-generic.md: Likewise.
15062 * config/arm/arm1020e.md: Likewise.
15063 * config/arm/arm1026ejs.md: Likewise.
15064 * config/arm/arm1136jfs.md: Likewise.
15065 * config/arm/arm926ejs.md: Likewise.
15066 * config/arm/cortex-a15.md: Likewise.
15067 * config/arm/cortex-a17.md: Likewise.
15068 * config/arm/cortex-a5.md: Likewise.
15069 * config/arm/cortex-a53.md: Likewise.
15070 * config/arm/cortex-a57.md: Likewise.
15071 * config/arm/cortex-a7.md: Likewise.
15072 * config/arm/cortex-a8.md: Likewise.
15073 * config/arm/cortex-a9.md: Likewise.
15074 * config/arm/cortex-m4.md: Likewise.
15075 * config/arm/cortex-m7.md: Likewise.
15076 * config/arm/cortex-r4.md: Likewise.
15077 * config/arm/exynos-m1.md: Likewise.
15078 * config/arm/fa526.md: Likewise.
15079 * config/arm/fa606te.md: Likewise.
15080 * config/arm/fa626te.md: Likewise.
15081 * config/arm/fa726te.md: Likewise.
15082 * config/arm/fmp626.md: Likewise.
15083 * config/arm/iwmmxt.md: Likewise.
15084 * config/arm/ldmstm.md: Likewise.
15085 * config/arm/marvell-pj4.md: Likewise.
15086 * config/arm/xgene1.md: Likewise.
15087 * config/aarch64/thunderx.md: Likewise.
15088 * config/aarch64/thunderx2t99.md: Likewise.
15089 * config/aarch64/falkor.md: Likewise.
15091 2017-09-12 Martin Liska <mliska@suse.cz>
15093 * attribs.c (private_lookup_attribute): New function.
15094 * attribs.h (private_lookup_attribute): Declared here.
15095 (lookup_attribute): Called from this place.
15097 2017-09-12 Richard Biener <rguenther@suse.de>
15099 PR tree-optimization/82157
15100 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15101 stmts with side-effects.
15103 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15104 Alan Hayward <alan.hayward@arm.com>
15105 David Sherwood <david.sherwood@arm.com>
15107 * target.def (hard_regno_nregs): New hook.
15108 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15109 * targhooks.h (default_hard_regno_nregs): Declare.
15110 * targhooks.c (default_hard_regno_nregs): New function.
15111 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15112 (TARGET_HARD_REGNO_NREGS): ...this hook.
15113 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15114 (CLASS_MAX_NREGS): Likewise.
15115 * doc/tm.texi: Regenerate.
15116 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15117 instead of HARD_REGNO_NREGS.
15118 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15119 HARD_REGNO_NREGS in the comment.
15120 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15121 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15122 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15123 Return an unsigned int.
15124 (TARGET_HARD_REGNO_NREGS): Redefine.
15125 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15126 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15127 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15128 (arc_hard_regno_nregs): New function.
15129 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15130 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15131 (arm_hard_regno_nregs): New function.
15132 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15133 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15134 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15135 (TARGET_HARD_REGNO_NREGS): Redefine.
15136 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15137 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15138 (HARD_REGNO_NREGS): Delete.
15139 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15140 (cr16_hard_regno_nregs): New function.
15141 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15142 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15143 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15144 (cris_hard_regno_nregs): New function.
15145 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15146 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15147 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15148 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15149 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15150 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15151 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15152 (frv_hard_regno_nregs): Make static. Take and return an
15154 (frv_class_max_nregs): Remove outdated copy of documentation.
15155 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15156 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15157 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15158 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15159 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15160 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15161 (TARGET_HARD_REGNO_NREGS): Redefine.
15162 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15163 (CLASS_MAX_NREGS): Update comment.
15164 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15165 (ia64_hard_regno_nregs): New function.
15166 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15167 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15168 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15169 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15170 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15172 (m32c_hard_regno_nregs): Likewise. Make static.
15173 (TARGET_HARD_REGNO_NREGS): Redefine.
15174 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15175 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15176 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15177 (m68k_hard_regno_nregs): New function.
15178 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15179 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15180 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15181 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15182 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15183 Take and return an unsigned int.
15184 (TARGET_HARD_REGNO_NREGS): Redefine.
15185 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15186 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15187 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15188 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15189 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15190 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15191 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15192 (msp430_hard_regno_nregs): Make static. Take and return an
15194 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15195 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15196 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15197 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15198 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15199 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15200 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15201 (TARGET_HARD_REGNO_NREGS): Redefine.
15202 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15203 (PA_HARD_REGNO_NREGS): ...this.
15204 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15205 (PA_HARD_REGNO_NREGS): ...this.
15206 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15207 (pa_hard_regno_nregs): New function.
15208 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15209 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15210 (pdp11_hard_regno_nregs): New function.
15211 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15212 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15213 (rs6000_hard_regno_nregs_hook): New function.
15214 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15215 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15216 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15217 Take and return an unsigned int. Move earlier in file.
15218 (TARGET_HARD_REGNO_NREGS): Redefine.
15219 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15220 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15221 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15222 (rl78_hard_regno_nregs): Make static. Take and return an
15224 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15225 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15226 (rs6000_hard_regno_nregs_hook): New function.
15227 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15228 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15229 (TARGET_HARD_REGNO_NREGS): Redefine.
15230 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15231 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15232 instead of HARD_REGNO_NREGS.
15233 (s390_hard_regno_nregs): New function.
15234 (s390_hard_regno_mode_ok): Add comment from s390.h.
15235 (TARGET_HARD_REGNO_NREGS): Redefine.
15236 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15237 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15238 (sh_hard_regno_nregs): New function.
15239 (sh_pass_in_reg_p): Use it.
15240 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15241 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15242 (sparc_hard_regno_nregs): New function.
15243 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15244 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15245 (spu_function_arg_advance): Use it, supplying a valid register number.
15246 (TARGET_HARD_REGNO_NREGS): Redefine.
15247 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15248 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15249 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15250 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15251 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15252 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15253 (CLASS_MAX_NREGS): Remove copy of old documentation.
15254 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15255 (visium_hard_regno_nregs): New function.
15256 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15257 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15258 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15259 xtensa_hard_regno_nregs): New function.
15260 * system.h (HARD_REGNO_NREGS): Poison.
15262 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15264 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15265 hard_regno_nregs instead of HARD_REGNO_NREGS.
15266 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15267 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15268 (c6x_expand_epilogue): Likewise.
15269 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15270 (frv_read_iacc_argument): Likewise.
15271 * config/sh/sh.c: Include regs.h.
15272 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15273 (regs_used): Likewise.
15274 (output_stack_adjust): Likewise.
15275 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15276 * expmed.c: Include regs.h.
15277 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15278 * ree.c: Include regs.h.
15279 (combine_reaching_defs): Use hard_regno_nregs instead of
15281 (add_removable_extension): Likewise.
15283 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15285 * regs.h (hard_regno_nregs): Turn into a function.
15286 (end_hard_regno): Update accordingly.
15287 * caller-save.c (setup_save_areas): Likewise.
15288 (save_call_clobbered_regs): Likewise.
15289 (replace_reg_with_saved_mem): Likewise.
15290 (insert_restore): Likewise.
15291 (insert_save): Likewise.
15292 * combine.c (can_change_dest_mode): Likewise.
15293 (move_deaths): Likewise.
15294 (distribute_notes): Likewise.
15295 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15296 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15297 (rs6000_split_multireg_move): Likewise.
15298 (rs6000_register_move_cost): Likewise.
15299 (rs6000_memory_move_cost): Likewise.
15300 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15301 (rs6000_split_multireg_move): Likewise.
15302 (rs6000_register_move_cost): Likewise.
15303 (rs6000_memory_move_cost): Likewise.
15304 * cselib.c (cselib_reset_table): Likewise.
15305 (cselib_lookup_1): Likewise.
15306 * emit-rtl.c (set_mode_and_regno): Likewise.
15307 * function.c (aggregate_value_p): Likewise.
15308 * ira-color.c (setup_profitable_hard_regs): Likewise.
15309 (check_hard_reg_p): Likewise.
15310 (calculate_saved_nregs): Likewise.
15311 (assign_hard_reg): Likewise.
15312 (improve_allocation): Likewise.
15313 (calculate_spill_cost): Likewise.
15314 * ira-emit.c (modify_move_list): Likewise.
15315 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15316 (ira_hard_reg_in_set_p): Likewise.
15317 * ira.c (setup_reg_mode_hard_regset): Likewise.
15318 (clarify_prohibited_class_mode_regs): Likewise.
15319 (check_allocation): Likewise.
15320 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15321 (lra_setup_reg_renumber): Likewise.
15322 (setup_try_hard_regno_pseudos): Likewise.
15323 (spill_for): Likewise.
15324 (assign_hard_regno): Likewise.
15325 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15326 * lra-constraints.c (in_class_p): Likewise.
15327 (lra_constraint_offset): Likewise.
15328 (simplify_operand_subreg): Likewise.
15329 (lra_constraints): Likewise.
15330 (split_reg): Likewise.
15331 (split_if_necessary): Likewise.
15332 (invariant_p): Likewise.
15333 (inherit_in_ebb): Likewise.
15334 * lra-lives.c (process_bb_lives): Likewise.
15335 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15336 (get_hard_regs): Likewise.
15337 (do_remat): Likewise.
15338 * lra-spills.c (assign_spill_hard_regs): Likewise.
15339 * mode-switching.c (create_pre_exit): Likewise.
15340 * postreload.c (reload_combine_recognize_pattern): Likewise.
15341 * recog.c (peep2_find_free_register): Likewise.
15342 * regcprop.c (kill_value_regno): Likewise.
15343 (set_value_regno): Likewise.
15344 (copy_value): Likewise.
15345 (maybe_mode_change): Likewise.
15346 (find_oldest_value_reg): Likewise.
15347 (copyprop_hardreg_forward_1): Likewise.
15348 * regrename.c (check_new_reg_p): Likewise.
15349 (regrename_do_replace): Likewise.
15350 * reload.c (push_reload): Likewise.
15351 (combine_reloads): Likewise.
15352 (find_dummy_reload): Likewise.
15353 (operands_match_p): Likewise.
15354 (find_reloads): Likewise.
15355 (find_equiv_reg): Likewise.
15356 (reload_adjust_reg_for_mode): Likewise.
15357 * reload1.c (count_pseudo): Likewise.
15358 (count_spilled_pseudo): Likewise.
15359 (find_reg): Likewise.
15360 (clear_reload_reg_in_use): Likewise.
15361 (free_for_value_p): Likewise.
15362 (allocate_reload_reg): Likewise.
15363 (choose_reload_regs): Likewise.
15364 (reload_adjust_reg_for_temp): Likewise.
15365 (emit_reload_insns): Likewise.
15366 (delete_output_reload): Likewise.
15367 * rtlanal.c (subreg_get_info): Likewise.
15368 * sched-deps.c (sched_analyze_reg): Likewise.
15369 * sel-sched.c (init_regs_for_mode): Likewise.
15370 (mark_unavailable_hard_regs): Likewise.
15371 (choose_best_reg_1): Likewise.
15372 (verify_target_availability): Likewise.
15373 * valtrack.c (dead_debug_insert_temp): Likewise.
15374 * var-tracking.c (track_loc_p): Likewise.
15375 (emit_note_insn_var_location): Likewise.
15376 * varasm.c (make_decl_rtl): Likewise.
15377 * reginfo.c (choose_hard_reg_mode): Likewise.
15378 (init_reg_modes_target): Refer directly to
15379 this_target_regs->x_hard_regno_nregs.
15381 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15383 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15384 instead of hard_regno_nregs.
15386 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15388 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15389 end_hard_regno instead of hard_regno_nregs.
15390 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15391 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15392 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15393 * ira-color.c (improve_allocation): Likewise.
15394 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15395 * lra-lives.c (mark_regno_live): Likewise.
15396 (mark_regno_dead): Likewise.
15397 * lra-remat.c (operand_to_remat): Likewise.
15398 * lra.c (collect_non_operand_hard_regs): Likewise.
15399 * postreload.c (reload_combine_note_store): Likewise.
15400 (move2add_valid_value_p): Likewise.
15401 * reload.c (regno_clobbered_p): Likewise.
15403 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15405 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15407 * config/v850/v850.c (v850_reorg): Likewise.
15408 * reload.c (refers_to_regno_for_reload_p): Likewise.
15409 (find_equiv_reg): Likewise.
15410 * reload1.c (reload_reg_reaches_end_p): Likewise.
15412 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15414 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15416 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15417 * config/arm/arm.c (output_move_neon): Likewise.
15418 (arm_attr_length_move_neon): Likewise.
15419 (neon_split_vcombine): Likewise.
15420 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15421 (c6x_mark_reg_written): Likewise.
15422 (c6x_dwarf_register_span): Likewise.
15423 * config/i386/i386.c (ix86_save_reg): Likewise.
15424 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15425 (rws_access_reg): Likewise.
15426 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15427 * mode-switching.c (create_pre_exit): Likewise.
15428 * ree.c (combine_reaching_defs): Likewise.
15429 (add_removable_extension): Likewise.
15430 * regcprop.c (find_oldest_value_reg): Likewise.
15431 (copyprop_hardreg_forward_1): Likewise.
15432 * reload.c (reload_inner_reg_of_subreg): Likewise.
15433 (push_reload): Likewise.
15434 (combine_reloads): Likewise.
15435 (find_dummy_reload): Likewise.
15436 (reload_adjust_reg_for_mode): Likewise.
15437 * reload1.c (find_reload_regs): Likewise.
15438 (forget_old_reloads_1): Likewise.
15439 (reload_reg_free_for_value_p): Likewise.
15440 (reload_adjust_reg_for_temp): Likewise.
15441 (emit_reload_insns): Likewise.
15442 (delete_output_reload): Likewise.
15443 * sel-sched.c (choose_best_reg_1): Likewise.
15444 (choose_best_pseudo_reg): Likewise.
15446 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15447 Alan Hayward <alan.hayward@arm.com>
15448 David Sherwood <david.sherwood@arm.com>
15450 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
15451 * target.def (slow_unaligned_access): New hook.
15452 * targhooks.h (default_slow_unaligned_access): Declare.
15453 * targhooks.c (default_slow_unaligned_access): New function.
15454 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
15455 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
15456 * doc/tm.texi: Regenerate.
15457 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
15458 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
15459 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
15461 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
15462 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
15464 (rs6000_slow_unaligned_access): New function.
15465 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15466 (expand_block_compare): Likewise.
15467 (expand_strn_compare): Likewise.
15468 (rs6000_rtx_costs): Likewise.
15469 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
15470 (riscv_slow_unaligned_access): Likewise.
15471 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
15472 (riscv_slow_unaligned_access_p): ...this and make static.
15473 (riscv_option_override): Update accordingly.
15474 (riscv_slow_unaligned_access): New function.
15475 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15476 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
15477 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
15478 (rs6000_slow_unaligned_access): New function.
15479 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
15480 (rs6000_rtx_costs): Likewise.
15481 * config/rs6000/rs6000-string.c (expand_block_compare)
15482 (expand_strn_compare): Use targetm.slow_unaligned_access instead
15483 of SLOW_UNALIGNED_ACCESS.
15484 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
15485 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
15486 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
15487 of SLOW_UNALIGNED_ACCESS.
15488 * expmed.c (simple_mem_bitfield_p): Likewise.
15489 * expr.c (alignment_for_piecewise_move): Likewise.
15490 (emit_group_load_1): Likewise.
15491 (emit_group_store): Likewise.
15492 (copy_blkmode_from_reg): Likewise.
15493 (emit_push_insn): Likewise.
15494 (expand_assignment): Likewise.
15495 (store_field): Likewise.
15496 (expand_expr_real_1): Likewise.
15497 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15498 * lra-constraints.c (simplify_operand_subreg): Likewise.
15499 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15500 * gimple-ssa-store-merging.c: Likewise in block comment at start
15502 * tree-ssa-strlen.c: Include target.h.
15503 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
15504 of SLOW_UNALIGNED_ACCESS.
15505 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
15507 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15509 PR rtl-optimization/82185
15510 * expmed.c (emit_store_flag_int): Only test tem if it has been
15513 2017-09-12 Richard Biener <rguenther@suse.de>
15515 PR middle-end/82149
15516 * match.pd ((FTYPE) N CMP CST): Fix typo.
15518 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
15520 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
15522 (mips_near_type_p): Add 'short_call' attribute as a synonym
15524 * doc/extend.texi (short_call): Document new function attribute.
15526 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15529 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
15530 assertion check that in the condition.
15531 (get_atomic_generic_size): Likewise. Before testing if parameter
15532 has pointer type, if it has array type, call for C++
15533 default_conversion to perform array-to-pointer conversion.
15535 2017-09-12 Richard Biener <rguenther@suse.de>
15537 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
15538 for operations we cannot scalarize.
15540 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
15542 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
15543 vectors heap vectors. Clean up comments.
15544 Make visited_bbs a reference.
15545 (profitable_jump_thread_path): Make GC
15546 vectors heap vectors. Clean up comments.
15548 (convert_and_register_jump_thread_path): Make GC vectors heap
15550 (check_subpath_and_update_thread_path): Same. Clean up comments.
15551 Make visited_bbs a reference.
15552 (handle_phi): Abstract common code to to
15553 register_jump_thread_path_if_profitable.
15554 Rename VAR_BB to DEF_BB.
15556 Make GC vectors heap vectors.
15557 Make visited_bbs a reference.
15558 (handle_assignment): Same.
15559 (register_jump_thread_path_if_profitable): New.
15560 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
15562 Make GC vectors heap vectors. Clean up comments.
15563 Make visited_bbs a reference.
15564 (find_jump_threads_backwards): Make visited_bbs live in the stack.
15565 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
15568 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
15571 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
15572 words of E_DImode object are reachable by xtensa_uimm8x4 access.
15574 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
15576 Revert r251800 and r251799.
15578 2017-09-11 Martin Jambor <mjambor@suse.cz>
15581 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
15582 arguments in advance.
15583 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
15584 use it to find predecessor edges.
15585 (naive_outof_ssa): Collect vector of predecessors.
15587 2017-09-08 Jason Merrill <jason@redhat.com>
15589 PR c++/70029 - ICE with ref-qualifier and -flto
15590 * langhooks.h (struct lang_hooks_for_types): Add
15591 copy_lang_qualifiers.
15592 * attribs.c (build_type_attribute_qual_variant): Use it.
15593 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
15595 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
15596 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
15598 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
15601 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
15602 (*mulsi3_sp64): New instruction.
15603 (mulsi3): New expander.
15605 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
15607 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
15609 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15611 * sancov.c: Include memmodel.h.
15613 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
15616 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
15619 2017-09-07 Carl Love <cel@us.ibm.com>
15621 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
15622 the sldi instruction.
15624 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
15626 * sancov.c: Include tm_p.h.
15628 2017-09-07 Jakub Jelinek <jakub@redhat.com>
15631 * output.h (switch_to_other_text_partition): New declaration.
15632 * varasm.c (switch_to_other_text_partition): New function.
15633 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
15634 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
15635 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
15636 to the other text partition before emitting LCL label and switch back
15637 after emitting the word after it.
15639 2017-09-07 Richard Biener <rguenther@suse.de>
15641 * passes.def (pass_split_crit_edges): Remove instance before PRE.
15642 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
15643 critical edges here, after loop init.
15644 (pass_data_pre): Remove PROP_no_crit_edges flags.
15645 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
15646 for valueization of call args to avoid leaking VN_TOP.
15647 (visit_use): Assert we do not visit default defs.
15648 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
15649 Use error_mark_node to more easily detect leaking VN_TOP.
15650 All default-defs are varying, not VN_TOP. Mark them visited.
15651 (run_scc_vn): Make code match comment.
15653 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15655 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
15656 OPTION_MASK_FLOAT128_KEYWORD.
15657 (POWERPC_MASKS): Likewise.
15658 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
15659 support for the -mfloat128-type option, and make -mfloat128
15660 default on PowerPC Linux systems. Define or undefine
15661 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
15662 Define __float128 to be __ieee128 if IEEE 128-bit support is
15663 enabled, or undefine it.
15664 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
15665 Delete defining __FLOAT128_TYPE__.
15666 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
15667 -mfloat128-type option and make -mfloat128 default on PowerPC
15669 (TARGET_FLOAT128_TYPE): Likewise.
15670 (-mfloat128-type): Likewise.
15671 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15672 Delete the -mfloat128-type option and make -mfloat128 default on
15673 PowerPC Linux systems. Always use __ieee128 to be the keyword for
15674 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
15675 128-bit floating point is enabled. Change tests from using
15676 -mfloat128-type to -mfloat128.
15677 (rs6000_mangle_type): Use the correct mangling for the __float128
15678 type even if normal long double is restricted to 64-bits.
15679 (floatn_mode): Enable the _Float128 type by default on VSX Linux
15681 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
15682 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
15683 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
15685 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
15686 documentation for -mfloat128.
15688 2017-09-06 Olivier Hainque <hainque@adacore.com>
15690 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
15692 2017-09-06 Wish Wu <wishwu007@gmail.com>
15693 Jakub Jelinek <jakub@redhat.com>
15695 * asan.c (initialize_sanitizer_builtins): Add
15696 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
15697 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
15698 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
15699 BT_FN_VOID_UINT64_PTR variables.
15700 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
15701 (BT_FN_VOID_UINT16_UINT16): Likewise.
15702 (BT_FN_VOID_UINT32_UINT32): Likewise.
15703 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
15704 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
15705 (BT_FN_VOID_UINT64_PTR): Likewise.
15706 * common.opt (flag_sanitize_coverage): New variable.
15707 (fsanitize-coverage=trace-pc): Remove.
15708 (fsanitize-coverage=): Add.
15709 * flag-types.h (enum sanitize_coverage_code): New enum.
15710 * fold-const.c (fold_range_test): Disable non-short-circuit
15711 optimization if flag_sanitize_coverage.
15712 (fold_truth_andor): Likewise.
15713 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
15714 * opts.c (COVERAGE_SANITIZER_OPT): Define.
15715 (coverage_sanitizer_opts): New array.
15716 (get_closest_sanitizer_option): Add OPTS argument, handle also
15717 OPT_fsanitize_coverage_.
15718 (parse_sanitizer_options): Adjusted to also handle
15719 OPT_fsanitize_coverage_.
15720 (common_handle_option): Add OPT_fsanitize_coverage_.
15721 * sancov.c (instrument_comparison, instrument_switch): New function.
15722 (sancov_pass): Add trace-cmp support.
15723 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
15724 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
15725 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
15726 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
15727 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
15728 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
15729 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
15730 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
15731 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
15732 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
15734 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15736 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
15737 error. Only quit immediately if parsing is complete.
15738 (BEGIN): Initialize fatal_err and parse_done.
15739 (begin fpu, end fpu): Check number of arguments.
15740 (begin arch, end arch): Likewise.
15741 (begin cpu, end cpu): Likewise.
15742 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
15743 (optalias): Likewise.
15745 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
15747 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
15748 * config/arm/arm-isa.h: Delete. Move definitions to ...
15749 * arm-cpus.in: ... here. Use new feature and fgroup values.
15750 * config/arm/arm.c (arm_option_override): Use lower case for feature
15752 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
15753 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
15754 * config/arm/parsecpu.awk (END): Add new command 'isa'.
15756 (print_isa_bits_for): New function.
15757 (gen_isa): New function.
15758 (gen_comm_data): Use print_isa_bits_for.
15759 (define feature): New keyword.
15760 (define fgroup): New keyword.
15761 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
15762 (arm-isa.h): Add rule to generate file.
15763 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
15764 case for feature bit names.
15766 2017-09-06 Richard Biener <rguenther@suse.de>
15768 * tree-ssa-pre.c (NECESSARY): Remove.
15769 (create_expression_by_pieces): Do not touch pass-local flags.
15770 (insert_into_preds_of_block): Likewise.
15771 (do_pre_regular_insertion): Likewise.
15772 (eliminate_insert): Likewise.
15773 (eliminate_dom_walker::before_dom_children): Likewise.
15774 (fini_eliminate): Do not look at inserted_exprs.
15775 (mark_operand_necessary): Remove.
15776 (remove_dead_inserted_code): Replace with simple work-list
15777 algorithm based on inserted_exprs and SSA uses.
15778 (pass_pre::execute): Re-order fini_eliminate and
15779 remove_dead_inserted_code.
15781 2017-09-06 Olivier Hainque <hainque@adacore.com>
15783 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15784 for VxWorks 7. Adjust surrounding comments.
15786 2017-09-06 Richard Biener <rguenther@suse.de>
15788 * gimple-ssa-strength-reduction.c
15789 (find_candidates_dom_walker::before_dom_children): Also allow
15792 2017-09-06 Richard Biener <rguenther@suse.de>
15794 PR tree-optimization/82108
15795 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15796 for gap in the non-permutation SLP case.
15798 2017-09-06 Martin Jambor <mjambor@suse.cz>
15800 PR tree-optimization/82078
15801 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15802 add_access_to_work_queue...
15803 (build_accesses_from_assign): ...here.
15804 (propagate_all_subaccesses): Make sure racc is the group
15805 representative, if there is one.
15807 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15809 PR middle-end/82095
15810 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15813 2017-09-06 Richard Biener <rguenther@suse.de>
15815 * gimple-ssa-strength-reduction.c
15816 (find_candidates_dom_walker::before_doom_children): Use a
15817 type and not a mode check.
15819 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15822 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15824 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15826 2017-09-05 Jeff Law <law@redhat.com>
15828 PR tree-optimization/64910
15829 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15830 cases where we have 3 or more operands.
15832 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15834 PR middle-end/81768
15835 * omp-low.c (lower_omp_for): Recompute tree invariant if
15836 gimple_omp_for_initial/final is ADDR_EXPR.
15838 PR middle-end/81768
15839 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15840 into gimple val before gimplification fo the COND_EXPR.
15842 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15844 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15845 REGION_COPY argument.
15846 (thread_through_all_blocks): Remove unused argument to
15847 duplicate_thread_path.
15849 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15850 Alan Hayward <alan.hayward@arm.com>
15851 David Sherwood <david.sherwood@arm.com>
15853 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15854 Take a scalar_mode rather than a machine_mode.
15855 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15856 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15857 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15858 (aarch64_gen_adjusted_ldpstp): Likewise.
15859 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15861 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15862 Alan Hayward <alan.hayward@arm.com>
15863 David Sherwood <david.sherwood@arm.com>
15865 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15866 Take a scalar_int_mode instead of a machine_mode.
15867 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15868 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15869 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15870 (aarch64_simd_attr_length_rglist): Delete.
15871 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15872 a scalar_int_mode instead of a machine_mode.
15873 (aarch64_add_offset): Likewise.
15874 (aarch64_internal_mov_immediate): Likewise
15875 (aarch64_add_constant_internal): Likewise.
15876 (aarch64_add_constant): Likewise.
15877 (aarch64_movw_imm): Likewise.
15878 (aarch64_rtx_arith_op_extract_p): Likewise.
15879 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15880 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15881 Remove assert that the mode isn't a vector.
15882 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15883 (aarch64_expand_mov_immediate): Update calls after above changes.
15884 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15885 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15886 (aarch64_move_imm): Likewise.
15887 (aarch64_can_const_movi_rtx_p): Likewise.
15888 (aarch64_strip_extend): Likewise.
15889 (aarch64_extr_rtx_p): Likewise.
15890 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15891 a CONST_INT when the mode parameter is VOIDmode.
15892 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15894 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15896 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15897 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15898 (bitwise_type_for_mode): Update accordingly.
15900 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15902 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15903 * stor-layout.c (mode_for_size_tree): Likewise.
15904 (mode_for_array): Update accordingly.
15905 (layout_decl): Likewise.
15906 (compute_record_mode): Likewise. Only set the mode once.
15908 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15910 * target.def (get_mask_mode): Change return type to opt_mode.
15912 * doc/tm.texi: Regenerate.
15913 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15914 * targhooks.c (default_get_mask_mode): Likewise.
15915 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15916 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15917 targetm.get_mask_mode.
15918 * tree.c (build_truth_vector_type): Likewise.
15920 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15922 * machmode.h (mode_for_vector): Return an opt_mode.
15923 * stor-layout.c (mode_for_vector): Likewise.
15924 (mode_for_int_vector): Update accordingly.
15925 (layout_type): Likewise.
15926 * config/i386/i386.c (emit_memmov): Likewise.
15927 (ix86_expand_set_or_movmem): Likewise.
15928 (ix86_expand_vector_init): Likewise.
15929 (ix86_get_mask_mode): Likewise.
15930 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15932 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15933 * expmed.c (extract_bit_field_1): Likewise.
15934 * expr.c (expand_expr_real_2): Likewise.
15935 * optabs-query.c (can_vec_perm_p): Likewise.
15936 (can_vec_mask_load_store_p): Likewise.
15937 * optabs.c (expand_vec_perm): Likewise.
15938 * targhooks.c (default_get_mask_mode): Likewise.
15939 * tree-vect-stmts.c (vectorizable_store): Likewise.
15940 (vectorizable_load): Likewise.
15941 (get_vectype_for_scalar_type_and_size): Likewise.
15943 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15945 * machmode.h (mode_for_int_vector): New function.
15946 * stor-layout.c (mode_for_int_vector): Likewise.
15947 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15948 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15949 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15950 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15951 (s390_expand_vcond): Likewise.
15953 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15955 * machmode.h (opt_machine_mode): New type.
15956 (opt_mode<T>): Allow construction from anything that can be
15958 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15959 (mode_for_size): Return an opt_machine_mode.
15960 * stor-layout.c (mode_for_size): Likewise.
15961 (mode_for_size_tree): Update call accordingly.
15962 (bitwise_mode_for_mode): Likewise.
15963 (make_fract_type): Likewise.
15964 (make_accum_type): Likewise.
15965 * caller-save.c (replace_reg_with_saved_mem): Update call
15967 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15968 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15969 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15970 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15971 * expmed.c (extract_bit_field_1): Likewise.
15972 * reload.c (get_secondary_mem): Likewise.
15973 * varasm.c (assemble_integer): Likewise.
15974 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15977 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15979 * machmode.h (decimal_float_mode_for_size): New function.
15980 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15981 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15982 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15983 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15984 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15985 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15987 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15989 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15990 (get_builtin_sync_mode): Likewise.
15991 (expand_ifn_atomic_compare_exchange): Likewise.
15992 (expand_builtin_atomic_clear): Likewise.
15993 (expand_builtin_atomic_test_and_set): Likewise.
15994 (fold_builtin_atomic_always_lock_free): Likewise.
15995 * calls.c (compute_argument_addresses): Likewise.
15996 (emit_library_call_value_1): Likewise.
15997 (store_one_arg): Likewise.
15998 * combine.c (combine_instructions): Likewise.
15999 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16000 * config/arm/arm.c (arm_function_value): Likewise.
16001 (aapcs_allocate_return_reg): Likewise.
16002 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16003 * config/i386/i386.c (construct_container): Likewise.
16004 (ix86_gimplify_va_arg): Likewise.
16005 (ix86_expand_sse_cmp): Likewise.
16006 (emit_memmov): Likewise.
16007 (emit_memset): Likewise.
16008 (expand_small_movmem_or_setmem): Likewise.
16009 (ix86_expand_pextr): Likewise.
16010 (ix86_expand_pinsr): Likewise.
16011 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16012 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16014 * config/mips/mips.c (mips_function_value_1) Likewise.
16015 (mips_block_move_straight): Likewise.
16016 (mips_expand_ins_as_unaligned_store): Likewise.
16017 * config/powerpcspe/powerpcspe.c
16018 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16019 (rs6000_darwin64_record_arg_flush): Likewise.
16020 * config/rs6000/rs6000.c
16021 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16022 (rs6000_darwin64_record_arg_flush): Likewise.
16023 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16024 (sparc_function_value_1): Likewise.
16025 * config/spu/spu.c (adjust_operand): Likewise.
16026 (spu_emit_branch_or_set): Likewise.
16027 (arith_immediate_p): Likewise.
16028 * emit-rtl.c (gen_lowpart_common): Likewise.
16029 * expr.c (expand_expr_real_1): Likewise.
16030 * function.c (assign_parm_setup_block): Likewise.
16031 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16032 * reload1.c (alter_reg): Likewise.
16033 * stor-layout.c (mode_for_vector): Likewise.
16034 (layout_type): Likewise.
16036 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16038 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16039 (spu_convert_move): Likewise.
16040 * lower-subreg.c (resolve_simple_move): Likewise.
16042 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16045 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16046 define_insn to a define_expand.
16047 (altivec_vsum2sws_direct): New define_insn.
16048 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16050 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16052 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16055 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16060 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16061 realign stack if stack alignment needed is less than incoming
16064 2017-09-05 Marek Polacek <polacek@redhat.com>
16067 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16070 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16072 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16074 2017-09-05 Richard Biener <rguenther@suse.de>
16076 PR tree-optimization/82084
16077 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16079 2017-09-05 Richard Biener <rguenther@suse.de>
16081 PR tree-optimization/82102
16082 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16084 2017-09-05 Martin Liska <mliska@suse.cz>
16086 PR tree-optimization/82032
16087 * tree-cfg.c (generate_range_test): New function.
16088 * tree-cfg.h (generate_range_test): Declared here.
16089 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16090 (cleanup_control_expr_graph): Use it.
16091 * tree-switch-conversion.c (try_switch_expansion): Remove
16093 (emit_case_nodes): Use generate_range_test.
16095 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16098 * config/i386/i386.md (*<btsc><mode>_mask): Add
16099 TARGET_USE_BT to insn constraint.
16100 (*btr<mode>_mask): Ditto.
16102 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16104 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16105 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16107 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16110 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16111 TARGET_NEON and TARGET_IWMMXT.
16112 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16113 TARGET_NEON and TARGET_IWMMXT.
16114 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16116 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16118 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16119 (ix86_rewrite_tls_address): Ditto.
16120 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16121 (ix86_rewrite_tls_address_1): Ditto.
16122 (ix86_rewrite_tls_address): Ditto.
16123 * config/i386/predicates.md (tls_address_pattern): New predicate.
16124 * config/i386/i386.md (TLS address splitter): New splitter.
16126 2017-09-04 Richard Biener <rguenther@suse.de>
16128 PR tree-optimization/82084
16129 * fold-const.h (can_native_encode_string_p): Declare.
16130 * fold-const.c (can_native_encode_string_p): Factor out from ...
16131 (native_encode_string): ... here.
16132 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16133 vectorizing stores from constants we later cannot handle.
16135 2017-09-04 Marek Polacek <polacek@redhat.com>
16138 * doc/invoke.texi: Update -Wtautological-compare documentation.
16140 2017-09-04 Jeff Law <law@redhat.com>
16142 PR tree-optimization/64910
16143 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16144 swap the first and last operand if the last is a constant.
16146 2017-09-04 Marek Polacek <polacek@redhat.com>
16149 * convert.c (do_narrow): When sanitizing signed integer overflows,
16150 bail out for signed types.
16151 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16153 2017-09-04 Richard Biener <rguenther@suse.de>
16155 PR tree-optimization/82060
16156 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16157 Move devirtualization after stmt folding and before EH/AB/noreturn
16158 cleanup to get the stmt refs canonicalized. Use a bool instead
16159 of gimple_modified_p since that doesn't work for NOPs. Schedule
16160 NOPs generated by folding for removal.
16162 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16163 Alan Hayward <alan.hayward@arm.com>
16164 David Sherwood <david.sherwood@arm.com>
16166 * coretypes.h (pad_direction): New enum.
16167 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16168 (FUNCTION_ARG_PADDING): Likewise.
16169 * target.def (function_arg_padding): New hook.
16170 * targhooks.h (default_function_arg_padding): Declare.
16171 * targhooks.c (default_function_arg_padding): New function.
16172 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16173 (TARGET_FUNCTION_ARG_PADDING): ...this.
16174 * doc/tm.texi: Regenerate.
16175 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16176 instead of direction.
16177 (compute_argument_addresses): Likewise.
16178 (load_register_parameters): Likewise.
16179 (emit_library_call_value_1): Likewise.
16180 (store_one_arg): Use targetm.calls.function_arg_padding instead
16181 of FUNCTION_ARG_PADDING.
16182 (must_pass_in_stack_var_size_or_pad): Likewise.
16183 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16184 (emit_group_store): Likewise.
16185 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16186 instead of FUNCTION_ARG_PADDING.
16187 (emit_push_insn): Likewise, and propagate enum change throughout
16189 * function.h (direction): Delete.
16190 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16192 * function.c (assign_parm_find_stack_rtl): Likewise.
16193 (assign_parm_setup_block_p): Likewise.
16194 (assign_parm_setup_block): Likewise.
16195 (gimplify_parameters): Likewise.
16196 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16197 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16199 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16200 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16201 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16202 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16203 (aarch64_function_arg_padding): ...this new function.
16204 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16205 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16206 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16207 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16208 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16209 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16210 (arm_pad_arg_upward): Replace with...
16211 (arm_function_arg_padding): ...this new function.
16212 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16214 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16215 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16216 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16217 (ia64_hpux_function_arg_padding): Replace with...
16218 (ia64_function_arg_padding): ...this new function. Use pad_direction
16219 instead of direction. Check for TARGET_HPUX.
16220 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16221 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16222 (iq2000_function_arg_padding): New function.
16223 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16224 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16225 (mips_function_arg_padding): ...this new function.
16226 (mips_pad_reg_upward): Update accordingly.
16227 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16228 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16229 targetm.calls.function_arg_padding.
16230 (FUNCTION_ARG_PADDING): Delete.
16231 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16232 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16233 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16234 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16235 (nios2_block_reg_padding): Return pad_direction instead of direction.
16236 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16237 instead of direction.
16238 (nios2_function_arg_padding): Likewise. Make static.
16239 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16240 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16241 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16242 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16243 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16244 (pa_function_arg_padding): Make static. Return pad_direction instead
16246 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16247 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16248 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16249 instead of direction. Use targetm.calls.function_arg_padding.
16250 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16251 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16252 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16253 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16254 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16256 (function_arg_padding): Rename to...
16257 (rs6000_function_arg_padding): ...this. Make static. Return
16258 pad_direction instead of direction.
16259 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16260 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16261 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16262 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16263 instead of direction. Use targetm.calls.function_arg_padding.
16264 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16265 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16266 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16267 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16268 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16269 (function_arg_padding): Rename to...
16270 (rs6000_function_arg_padding): ...this. Make static. Return
16271 pad_direction instead of direction.
16272 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16273 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16274 * config/s390/s390.c (s390_function_arg_padding): New function.
16275 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16276 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16277 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16278 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16279 (function_arg_padding): Rename to...
16280 (sparc_function_arg_padding): ...this. Make static. Return
16281 pad_direction instead of direction.
16282 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16283 * config/spu/spu.c (spu_function_arg_padding): New function.
16284 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16285 * system.h (FUNCTION_ARG_PADDING): Poison.
16287 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16288 Alan Hayward <alan.hayward@arm.com>
16289 David Sherwood <david.sherwood@arm.com>
16291 * target.def (modes_tieable_p): New hook.
16292 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16293 (TARGET_MODES_TIEABLE_P): ...this.
16294 * doc/tm.texi.in: Regenerate.
16295 * hooks.h (hook_bool_mode_mode_true): Declare.
16296 * hooks.c (hook_bool_mode_mode_true): New function.
16297 * combine.c (subst): Use targetm.modes_tieable_p instead of
16299 * dse.c (find_shift_sequence): Likewise.
16300 * expmed.c (extract_low_bits): Likewise.
16301 * lower-subreg.c: Include target.h.
16302 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16304 * rtlanal.c (rtx_cost): Likewise.
16305 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16306 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16307 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16308 (TARGET_MODES_TIEABLE_P): Redefine.
16309 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16310 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16311 (TARGET_MODES_TIEABLE_P): Redefine.
16312 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16313 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16314 (arc_modes_tieable_p): New function.
16315 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16316 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16317 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16318 (arm_modes_tieable_p): Make static.
16319 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16320 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16321 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16322 (TARGET_MODES_TIEABLE_P): Redefine.
16323 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16324 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16325 (TARGET_MODES_TIEABLE_P): Redefine.
16326 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16327 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16328 (cr16_modes_tieable_p): New function.
16329 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16330 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16331 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16332 (TRULY_NOOP_TRUNCATION): Update comment.
16333 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16334 (TRULY_NOOP_TRUNCATION): Update comment.
16335 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16336 (frv_modes_tieable_p): New function.
16337 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16338 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16339 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16340 (TARGET_MODES_TIEABLE_P): Redefine.
16341 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16342 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16343 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16344 (TARGET_MODES_TIEABLE_P): Redefine.
16345 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16346 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16347 (ia64_modes_tieable_p): New function.
16348 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16349 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16350 (iq2000_modes_tieable_p): New function.
16351 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16352 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16353 (lm32_modes_tieable_p): New function.
16354 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16355 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16356 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16357 (TARGET_MODES_TIEABLE_P): Redefine.
16358 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16359 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16360 (m32r_modes_tieable_p): New function.
16361 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16362 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16363 (m68k_modes_tieable_p): New function.
16364 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16365 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16366 (mcore_modes_tieable_p): New function.
16367 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16368 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16370 (TARGET_MODES_TIEABLE_P): Redefine.
16371 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16372 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16373 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16374 (TARGET_MODES_TIEABLE_P): Redefine.
16375 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16376 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16377 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16378 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16379 (mn10300_modes_tieable_p): ...this and make static.
16380 (TARGET_MODES_TIEABLE_P): Redefine.
16381 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16382 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16383 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16384 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16385 (msp430_modes_tieable_p): Make static.
16386 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16387 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16388 (TARGET_MODES_TIEABLE_P): Redefine.
16389 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16390 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16391 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16392 (TARGET_MODES_TIEABLE_P): Redefine.
16393 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16394 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16395 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16396 (TARGET_MODES_TIEABLE_P): Redefine.
16397 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16398 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16399 (pdp11_modes_tieable_p): New function.
16400 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16401 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16402 (rs6000_modes_tieable_p): New function.
16403 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16404 * config/powerpcspe/powerpcspe.md: Update comment.
16405 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16406 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16407 (TARGET_MODES_TIEABLE_P): Redefine.
16408 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16409 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16410 (rl78_modes_tieable_p): New function.
16411 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16412 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16413 (rs6000_modes_tieable_p): New function.
16414 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16415 * config/rs6000/rs6000.md: Update comment.
16416 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16417 * config/rx/rx.c (rx_modes_tieable_p): New function.
16418 (TARGET_MODES_TIEABLE_P): Redefine.
16419 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16420 * config/s390/s390.c (s390_modes_tieable_p): New function.
16421 (TARGET_MODES_TIEABLE_P): Redefine.
16422 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16423 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16424 (sh_modes_tieable_p): New function.
16425 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16426 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16427 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16428 (sparc_modes_tieable_p): Make static.
16429 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16430 * config/spu/spu.c (spu_modes_tieable_p): New function.
16431 (TARGET_MODES_TIEABLE_P): Redefine.
16432 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16433 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
16434 (TARGET_MODES_TIEABLE_P): Redefine.
16435 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
16436 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
16437 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
16438 * config/v850/v850.c (v850_modes_tieable_p): New function.
16439 (TARGET_MODES_TIEABLE_P): Redefine.
16440 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
16441 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
16442 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
16443 (visium_modes_tieable_p): New function.
16444 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
16445 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
16446 (xtensa_modes_tieable_p): New function.
16447 * system.h (MODES_TIEABLE_P): Poison.
16449 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16450 Alan Hayward <alan.hayward@arm.com>
16451 David Sherwood <david.sherwood@arm.com>
16453 * target.def (hard_regno_mode_ok): New hook.
16454 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
16455 (TARGET_HARD_REGNO_MODE_OK): ...this.
16456 * doc/tm.texi.in: Regenerate.
16457 * hooks.h (hook_bool_uint_mode_true): Declare.
16458 * hooks.c (hook_bool_uint_mode_true): New function.
16459 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
16460 HARD_REGNO_MODE_OK.
16461 * genpreds.c (write_insn_preds_c): Add an include of target.h.
16462 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
16463 instead of HARD_REGNO_MODE_OK.
16464 * caller-save.c: Include target.h.
16465 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
16466 HARD_REGNO_MODE_OK.
16467 * combine.c (can_combine_p): Likewise.
16468 (combinable_i3pat): Likewise.
16469 (can_change_dest_mode): Likewise.
16470 * expr.c (init_expr_target): Likewise.
16471 (convert_move): Likewise.
16472 (convert_modes): Likewise.
16473 * ira.c (setup_prohibited_class_mode_regs): Likewise.
16474 (setup_prohibited_mode_move_regs): Likewise.
16475 * ira.h (target_ira): Likewise.
16476 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16477 * lra-constraints.c (process_alt_operands): Likewise.
16478 (split_reg): Likewise.
16479 * recog.c (peep2_find_free_register): Likewise.
16480 * ree.c (combine_reaching_defs): Likewise.
16481 * regcprop.c (maybe_mode_change): Likewise.
16482 * reginfo.c (init_reg_sets_1): Likewise.
16483 (choose_hard_reg_mode): Likewise.
16484 (simplifiable_subregs): Likewise.
16485 * regrename.c (check_new_reg_p): Likewise.
16486 * reload.c (find_valid_class): Likewise.
16487 (find_valid_class_1): Likewise.
16488 (reload_inner_reg_of_subreg): Likewise.
16489 (push_reload): Likewise.
16490 (combine_reloads): Likewise.
16491 (find_dummy_reload): Likewise.
16492 (find_reloads): Likewise.
16493 * reload1.c (find_reg): Likewise.
16494 (set_reload_reg): Likewise.
16495 (allocate_reload_reg): Likewise.
16496 (choose_reload_regs): Likewise.
16497 (reload_adjust_reg_for_temp): Likewise.
16498 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
16499 (simplify_subreg_regno): Likewise.
16500 * sel-sched.c (init_regs_for_mode): Likewise.
16501 * varasm.c (make_decl_rtl): Likewise.
16502 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
16503 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
16504 HARD_REGNO_MODE_OK.
16505 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
16506 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
16507 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16508 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
16509 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
16510 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16511 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
16512 (arc_mode_class): Delete.
16513 (HARD_REGNO_MODE_OK): Delete.
16514 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16515 (arc_hard_regno_mode_ok): Rename old array to...
16516 (arc_hard_regno_mode_ok_modes): ...this.
16517 (arc_conditional_register_usage): Update accordingly.
16518 (arc_mode_class): Make static.
16519 (arc_hard_regno_mode_ok): New function.
16520 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
16521 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
16522 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16523 (arm_hard_regno_mode_ok): Make static.
16524 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
16525 HARD_REGNO_MODE_OK.
16526 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
16527 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
16528 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
16530 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16531 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
16532 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
16533 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
16534 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
16535 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16536 * config/bfin/predicates.md (valid_reg_operand): Use
16537 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
16538 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
16539 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
16540 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16541 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
16542 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
16543 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16544 (cr16_hard_regno_mode_ok): Make static and return a bool.
16545 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
16546 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16547 (cris_hard_regno_mode_ok): New function.
16548 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
16549 (epiphany_mode_class): Delete.
16550 (HARD_REGNO_MODE_OK): Delete.
16551 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
16552 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16553 (hard_regno_mode_ok): Rename to...
16554 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
16555 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
16556 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
16557 HARD_REGNO_MODE_OK.
16558 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
16559 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
16560 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16561 (frv_hard_regno_mode_ok): Make static and return a bool.
16562 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
16563 HARD_REGNO_MODE_OK.
16564 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
16565 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
16566 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
16567 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
16569 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16570 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
16571 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
16572 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
16574 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16575 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
16576 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16577 (ia64_hard_regno_mode_ok): New function.
16578 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
16579 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16580 (iq2000_hard_regno_mode_ok): New function.
16581 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
16582 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16583 (lm32_hard_regno_mode_ok): New function.
16584 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
16585 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
16586 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
16587 instead of HARD_REGNO_MODE_OK.
16588 (m32c_hard_regno_ok): Rename to...
16589 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
16590 (m32c_cannot_change_mode_class): Update accordingly.
16591 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16592 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
16593 (m32r_mode_class): Delete.
16594 (HARD_REGNO_MODE_OK): Delete.
16595 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16596 (m32r_hard_regno_mode_ok): Rename to...
16597 (m32r_hard_regno_modes): ...this.
16598 (m32r_mode_class): Make static.
16599 (m32r_hard_regno_mode_ok): New function.
16600 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
16601 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
16602 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16603 (m68k_hard_regno_mode_ok): Make static.
16604 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
16605 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16606 (mcore_hard_regno_mode_ok): New function.
16607 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
16608 (HARD_REGNO_MODE_OK): Delete.
16609 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
16611 (microblaze_hard_regno_mode_ok_p): ...this and make static.
16612 (microblaze_hard_regno_mode_ok): New function.
16613 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16614 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
16615 (mips_hard_regno_mode_ok): Delete.
16616 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
16617 (mips_hard_regno_mode_ok_p): ...this and make static.
16618 (mips_hard_regno_mode_ok_p): Rename to...
16619 (mips_hard_regno_mode_ok_uncached): ...this.
16620 (mips_hard_regno_mode_ok): New function.
16621 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
16622 of HARD_REGNO_MODE_OK.
16623 (mips_option_override): Update after above name changes.
16624 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16625 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
16626 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
16627 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
16628 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
16629 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16630 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
16631 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
16632 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
16633 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16634 (msp430_hard_regno_mode_ok): Make static and return a bool.
16635 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
16636 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
16637 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
16639 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16640 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
16641 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
16642 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
16643 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
16644 (PA_HARD_REGNO_MODE_OK): ...this
16645 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
16646 (PA_HARD_REGNO_MODE_OK): ...this.
16647 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16648 (pa_hard_regno_mode_ok): New function.
16649 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
16650 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16651 (pdp11_hard_regno_mode_ok): New function.
16652 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
16653 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
16655 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
16657 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16658 (rs6000_hard_regno_mode_ok): Rename to...
16659 (rs6000_hard_regno_mode_ok_uncached): ...this.
16660 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16661 (rs6000_hard_regno_mode_ok): New function.
16662 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
16663 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
16664 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
16665 (riscv_hard_regno_mode_ok): ...this and make static.
16666 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16667 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
16668 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
16669 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16670 (rl78_hard_regno_mode_ok): Make static and return bool.
16671 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
16672 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
16674 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
16675 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16676 (rs6000_hard_regno_mode_ok): Rename to...
16677 (rs6000_hard_regno_mode_ok_uncached): ...this.
16678 (rs6000_init_hard_regno_mode_ok): Update accordingly.
16679 (rs6000_hard_regno_mode_ok): New function.
16680 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
16681 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
16682 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16683 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
16684 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
16685 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
16686 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16687 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
16688 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
16689 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16690 (sh_hard_regno_mode_ok): Make static.
16691 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
16692 instead of HARD_REGNO_MODE_OK.
16693 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
16694 (sparc_mode_class): Delete.
16695 (HARD_REGNO_MODE_OK): Delete.
16696 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16697 (hard_regno_mode_classes): Make static.
16698 (sparc_mode_class): Likewise.
16699 (sparc_hard_regno_mode_ok): New function.
16700 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
16701 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
16702 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
16704 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16705 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
16706 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
16707 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
16708 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
16709 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16710 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
16711 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
16712 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
16713 (visium_hard_regno_mode_ok): New function.
16714 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
16715 instead of HARD_REGNO_MODE_OK.
16716 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
16717 (HARD_REGNO_MODE_OK): Delete.
16718 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
16719 (xtensa_hard_regno_mode_ok_p): ...this and make static.
16720 (xtensa_option_override): Update accordingly.
16721 (TARGET_HARD_REGNO_MODE_OK): Redefine.
16722 (xtensa_hard_regno_mode_ok): New function.
16723 * system.h (HARD_REGNO_MODE_OK): Poison.
16725 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16726 Alan Hayward <alan.hayward@arm.com>
16727 David Sherwood <david.sherwood@arm.com>
16729 * target.def (hard_regno_call_part_clobbered): New hook.
16730 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
16731 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
16732 * doc/tm.texi: Regenerate.
16733 * hooks.h (hook_bool_uint_mode_false): Declare.
16734 * hooks.c (hook_bool_uint_mode_false): New function.
16735 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16736 * cselib.c (cselib_process_insn): Use
16737 targetm.hard_regno_call_part_clobbered instead of
16738 HARD_REGNO_CALL_PART_CLOBBERED.
16739 * ira-conflicts.c (ira_build_conflicts): Likewise.
16740 * ira-costs.c (ira_tune_allocno_costs): Likewise.
16741 * lra-constraints.c (need_for_call_save_p): Likewise.
16742 * lra-lives.c: Include target.h.
16743 (check_pseudos_live_through_calls): Use
16744 targetm.hard_regno_call_part_clobbered instead of
16745 HARD_REGNO_CALL_PART_CLOBBERED.
16746 * regcprop.c: Include target.h.
16747 (copyprop_hardreg_forward_1): Use
16748 targetm.hard_regno_call_part_clobbered instead of
16749 HARD_REGNO_CALL_PART_CLOBBERED.
16750 * reginfo.c (choose_hard_reg_mode): Likewise.
16751 * regrename.c (check_new_reg_p): Likewise.
16752 * reload.c (find_equiv_reg): Likewise.
16753 * reload1.c (emit_reload_insns): Likewise.
16754 * sched-deps.c (deps_analyze_insn): Likewise.
16755 * sel-sched.c (init_regs_for_mode): Likewise.
16756 (mark_unavailable_hard_regs): Likewise.
16757 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
16758 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16759 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
16761 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16762 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16763 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
16765 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
16767 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16768 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16769 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16771 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16772 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16773 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16775 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16776 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16778 * config/powerpcspe/powerpcspe.c
16779 (rs6000_hard_regno_call_part_clobbered): New function.
16780 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16781 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16782 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16784 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16785 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16786 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16788 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16789 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16790 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16792 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16793 Alan Hayward <alan.hayward@arm.com>
16794 David Sherwood <david.sherwood@arm.com>
16796 * rtl.h (subreg_memory_offset): Declare.
16797 * emit-rtl.c (subreg_memory_offset): New function.
16798 * expmed.c (store_bit_field_1): Use it.
16799 * expr.c (undefined_operand_subword_p): Likewise.
16800 * simplify-rtx.c (simplify_subreg): Likewise.
16802 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16804 PR rtl-optimization/57448
16807 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16808 using atomic_load pattern.
16809 (expand_atomic_store): Likewise.
16811 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16814 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16815 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16816 handling. Use replace_call_with_value with NULL instead of
16817 gsi_replace, unlink_stmt_vdef and release_defs.
16819 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16822 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16824 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16827 * rtl.h (emit_library_call_value_1): Declare.
16828 (emit_library_call): Replace declaration with a series of overloads.
16829 Remove the parameter count argument.
16830 (emit_library_call_value): Likewise.
16831 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16832 with an "rtx_mode_t *".
16833 (emit_library_call_value): Delete.
16834 (emit_library_call): Likewise.
16835 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16836 (asan_emit_allocas_unpoison): Likewise.
16837 * builtins.c (expand_builtin_powi): Likewise.
16838 (expand_asan_emit_allocas_unpoison): Likewise.
16839 * cfgexpand.c (expand_main_function): Likewise.
16840 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16841 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16842 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16843 * config/arm/arm.c (arm_trampoline_init): Likewise.
16844 (arm_call_tls_get_addr): Likewise.
16845 (arm_expand_divmod_libfunc): Likewise.
16846 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16847 (smulsi3_highpart): Likewise.
16848 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16849 (c6x_expand_compare): Likewise.
16850 (c6x_expand_movmem): Likewise.
16851 * config/frv/frv.c (frv_trampoline_init): Likewise.
16852 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16853 (ix86_expand_divmod_libfunc): Likewise.
16854 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16855 (ia64_expand_compare): Likewise.
16856 (ia64_profile_hook): Likewise.
16857 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16858 (nonlocal_goto): Likewise.
16859 (restore_stack_nonlocal): Likewise.
16860 * config/m32r/m32r.c (block_move_call): Likewise.
16861 (m32r_trampoline_init): Likewise.
16862 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16863 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16864 (m68k_call_m68k_read_tp): Likewise.
16865 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16866 (microblaze_expand_divide): Likewise.
16867 * config/mips/mips.h (mips_args): Likewise.
16868 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16869 (MIPS_ICACHE_SYNC): Likewise.
16870 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16871 (nios2_trampoline_init): Likewise.
16872 * config/pa/pa.c (hppa_tls_call): Likewise.
16873 (pa_trampoline_init): Likewise.
16874 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16875 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16876 (expand_strn_compare): Likewise.
16877 (rs6000_generate_compare): Likewise.
16878 (rs6000_expand_float128_convert): Likewise.
16879 (output_profile_hook): Likewise.
16880 (rs6000_trampoline_init): Likewise.
16881 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16882 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16883 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16884 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16885 (rs6000_generate_compare): Likewise.
16886 (rs6000_expand_float128_convert): Likewise.
16887 (output_profile_hook): Likewise.
16888 (rs6000_trampoline_init): Likewise.
16889 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16890 * config/sh/sh.c (sh_trampoline_init): Likewise.
16891 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16892 (sparc_emit_float_lib_cmp): Likewise.
16893 (sparc32_initialize_trampoline): Likewise.
16894 (sparc64_initialize_trampoline): Likewise.
16895 (sparc_profile_hook): Likewise.
16896 * config/spu/spu.c (ea_load_store): Likewise.
16897 * config/spu/spu.md (floatunssidf2): Likewise.
16898 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16899 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16900 * config/visium/visium.c (expand_block_move_4): Likewise.
16901 (expand_block_move_2): Likewise.
16902 (expand_block_move_1): Likewise.
16903 (expand_block_set_4): Likewise.
16904 (expand_block_set_2): Likewise.
16905 (expand_block_set_1): Likewise.
16906 (visium_trampoline_init): Likewise.
16907 (visium_profile_hook): Likewise.
16908 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16909 (xtensa_setup_frame_addresses): Likewise.
16910 (xtensa_trampoline_init): Likewise.
16911 * except.c (sjlj_emit_function_enter): Likewise.
16912 (sjlj_emit_function_exit): Likewise.
16913 * explow.c (allocate_dynamic_stack_space): Likewise.
16914 (probe_stack_range): Likewise.
16915 * expr.c (convert_mode_scalar): Likewise.
16916 * optabs.c (expand_binop): Likewise.
16917 (expand_twoval_binop_libfunc): Likewise.
16918 (expand_unop): Likewise.
16919 (prepare_cmp_insn): Likewise.
16920 (prepare_float_lib_cmp): Likewise.
16921 (expand_float): Likewise.
16922 (expand_fix): Likewise.
16923 (expand_fixed_convert): Likewise.
16924 (maybe_emit_sync_lock_test_and_set): Likewise.
16925 (expand_atomic_compare_and_swap): Likewise.
16926 (expand_mem_thread_fence): Likewise.
16927 (expand_atomic_fetch_op): Likewise.
16929 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16931 * doc/generic.texi (OpenACC): Adjust URL.
16932 * doc/invoke.texi (C Dialect Options): Ditto.
16934 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16936 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16937 predicate for operand 1. Add (m,<S>) constraint.
16938 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16939 Prevent memory operand 1 with register operand 2.
16941 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16943 PR rtl-optimization/82024
16944 * combine.c (try_combine): If the combination result is a PARALLEL,
16945 and we only need to retain the SET in there that would be placed
16946 at I2, check that we can place that at I3 instead, before doing so.
16948 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16951 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16953 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16954 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16957 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16958 Jeff Law <law@redhat.com>
16960 * varasm.c (bss_initializer_p): Do not put constants into .bss
16961 (categorize_decl_for_section): Handle bss_initializer_p returning
16962 false when DECL_INITIAL is NULL.
16964 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16967 * config/s390/s390.c (s390_can_inline_p): New function.
16969 2017-09-01 Jeff Law <law@redhat.com>
16971 PR tree-optimization/82052
16972 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16973 Always initialize the returned slot after a hash table miss
16974 when INSERT is true.
16976 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16978 * config/s390/s390.md (mem_signal_fence): Remove.
16979 * doc/md.texi (mem_signal_fence): Remove.
16980 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16982 * target-insns.def (mem_signal_fence): Remove.
16984 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16987 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16990 * asan.c (create_odr_indicator): Strip name encoding from assembler
16991 name before appending it after __odr_asan_.
16993 2017-09-01 Martin Liska <mliska@suse.cz>
16995 PR tree-optimization/82059
16996 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16997 frequency only when an edge is redirected.
16999 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17001 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17002 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17003 (arc_conditional_register_usage): Remove ARC600 lp_count
17005 (arc_file_start): Emit Tag_ARC_CPU_variation.
17006 (arc_can_use_doloop_p): New conditions to use ZOLs.
17007 (hwloop_fail): New function.
17008 (hwloop_optimize): Likewise.
17009 (hwloop_pattern_reg): Likewise.
17010 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17011 (arc_reorg_loops): New function, calls reorg_loops.
17012 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17013 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17015 (arc_loop_hazard): Remove function, functionality moved into
17017 (arc_hazard): Remove arc_loop_hazard call.
17018 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17019 into hwloop_optimize.
17020 (arc_label_align): Remove ZOL handling.
17021 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17022 * config/arc/arc.md (doloop_begin): Remove pattern.
17023 (doloop_begin_i): Likewise.
17024 (doloop_end_i): Likewise.
17025 (doloop_fallback): Likewise.
17026 (doloop_fallback_m): Likewise.
17027 (doloop_end): Reimplement expand.
17028 (arc_lp): New pattern for LP instruction.
17029 (loop_end): New pattern.
17030 (loop_fail): Likewise.
17031 (decrement_and_branch_until_zero): Likewise.
17032 * config/arc/arc.opt (mlpc-width): New option.
17033 * doc/invoke.texi (mlpc-width): Document option.
17035 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17037 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17038 (arc_ccfsm_advance): Fix checking for delay slots.
17039 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17041 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17043 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17045 (movhi_insn): Update store instruction constraint which are saving
17046 6-bit short immediates.
17047 (movsi_insn): Consider also short scaled load operations.
17048 (zero_extendhisi2_i): Use Usd constraint instead of T.
17049 (extendhisi2_i): Add q constraint.
17050 (arc_clzsi2): Add type and length attributes.
17051 (arc_ctzsi2): Likewise.
17052 * config/arc/constraints.md (Usc): Update constraint, the
17053 assembler can parse two relocations for a single instruction.
17055 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17057 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17058 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17060 2017-08-31 Olivier Hainque <hainque@adacore.com>
17062 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17063 match as powerpc-wrs-vxworks*.
17065 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17067 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17068 register constraint for by-element operand.
17069 (aarch64_mls_elt_merge<mode>): Likewise.
17071 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17073 * config/arc/arc.c (arc_can_follow_jump): Check for short
17076 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17078 * config.gcc: Use g.opt for arc.
17079 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17080 functionality moved to ...
17081 (legitimate_scaled_address_p): New function, ...here.
17082 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17083 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17084 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17086 (arc_override_options): Handle G option.
17087 (arc_output_pic_addr_const): Correct function definition.
17088 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17089 (arc_decl_anon_ns_mem_p): Delete.
17090 (arc_in_small_data_p): Overhaul this function to take into
17091 consideration the value given via G option.
17092 (arc_rewrite_small_data_1): Renamed and corrected old
17093 arc_rewrite_small_data function.
17094 (arc_rewrite_small_data): New function.
17095 (small_data_pattern): Don't use pic_offset_table_rtx.
17096 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17097 * config/arc/simdext.md (movmisalignv2hi): Use
17098 prepare_move_operands function.
17100 (movmisalign*): Likewise.
17101 * doc/invoke.texi (ARC options): Document -G option.
17103 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17105 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17107 * config/arc/arc.c (arc_print_operand): Output scalled address for
17108 sdata whenever is possible.
17109 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17110 load/stores are available.
17111 (compact_sda_memory_operand): Check for the alignment required by
17112 code density instructions.
17113 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17115 * config/arc/constraints.md (Usd): Update constraint.
17116 (Us0): New constraint.
17117 (Usc): Update constraint.
17119 2017-08-31 Richard Biener <rguenther@suse.de>
17121 PR middle-end/82054
17122 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17123 function only once.
17125 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17127 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17128 Resize type_signature.
17130 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17131 Alan Hayward <alan.hayward@arm.com>
17132 David Sherwood <david.sherwood@arm.com>
17134 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17135 subregs whose inner modes can be stored in GPRs.
17136 (aarch64_classify_index): Likewise.
17138 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17139 Alan Hayward <alan.hayward@arm.com>
17140 David Sherwood <david.sherwood@arm.com>
17142 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17143 (V_INT_EQUIV): ...this.
17144 (v_cmp_result): Rename to...
17145 (v_int_equiv): ...this.
17146 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17147 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17148 (copysign<mode>3): Likewise.
17149 (aarch64_simd_bsl<mode>_internal): Likewise.
17150 (aarch64_simd_bsl<mode>): Likewise.
17151 (vec_cmp<mode><mode>): Likewise.
17152 (vcond<mode><mode>): Likewise.
17153 (vcond<v_cmp_mixed><mode>): Likewise.
17154 (vcondu<mode><v_cmp_mixed>): Likewise.
17155 (aarch64_cm<optab><mode>): Likewise.
17156 (aarch64_cmtst<mode>): Likewise.
17157 (aarch64_fac<optab><mode>): Likewise.
17158 (vec_perm_const<mode>): Likewise.
17159 (vcond_mask_<mode><v_cmp_result>): Rename to...
17160 (vcond_mask_<mode><v_int_equiv>): ...this.
17161 (vec_cmp<mode><v_cmp_result>): Rename to...
17162 (vec_cmp<mode><v_int_equiv>): ...this.
17164 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17165 Alan Hayward <alan.hayward@arm.com>
17166 David Sherwood <david.sherwood@arm.com>
17168 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17170 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17171 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17172 (UNSPEC_LD4_DREG): New unspecs.
17173 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17174 (aarch64_ld2<mode>_dreg_be): Replace with...
17175 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17177 (aarch64_ld3<mode>_dreg_le)
17178 (aarch64_ld3<mode>_dreg_be): Replace with...
17179 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17181 (aarch64_ld4<mode>_dreg_le)
17182 (aarch64_ld4<mode>_dreg_be): Replace with...
17183 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17186 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17188 PR tree-optimization/81987
17189 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17190 insert an initializer in a location not dominated by the stride
17193 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17195 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17196 on the entire header of the finally block in the fallthru case.
17198 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17200 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17202 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17204 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17205 rs6000_emit_move_from_cr and call renamed function.
17206 (rs6000_emit_prologue): Call renamed functions.
17207 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17208 movesi_from_cr, remove volatile CRs.
17210 2017-08-30 Jon Beniston <jon@beniston.com>
17211 Richard Biener <rguenther@suse.de>
17213 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17214 of VECTOR_MODE_P check.
17215 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17216 element vector types.
17218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17220 * df.h (df_read_modify_subreg_p): Remove in favor of...
17221 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17222 const_rtx instead of an rtx.
17223 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17224 * df-problems.c (df_word_lr_mark_ref): Likewise.
17225 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17226 (mark_pseudo_reg_dead): Likewise.
17227 (mark_ref_dead): Likewise.
17228 * reginfo.c (init_subregs_of_mode): Likewise.
17229 * sched-deps.c (sched_analyze_1): Likewise.
17230 * df-scan.c (df_def_record_1): Likewise.
17231 (df_uses_record): Likewise.
17232 (df_read_modify_subreg_p): Remove in favor of...
17233 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17234 const_rtx instead of an rtx.
17236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17237 Alan Hayward <alan.hayward@arm.com>
17238 David Sherwood <david.sherwood@arm.com>
17240 * rtl.h (partial_subreg_p): New function.
17241 * caller-save.c (save_call_clobbered_regs): Use it.
17242 * calls.c (expand_call): Likewise.
17243 * combine.c (combinable_i3pat): Likewise.
17244 (simplify_set): Likewise.
17245 (make_extraction): Likewise.
17246 (make_compound_operation_int): Likewise.
17247 (gen_lowpart_or_truncate): Likewise.
17248 (force_to_mode): Likewise.
17249 (make_field_assignment): Likewise.
17250 (reg_truncated_to_mode): Likewise.
17251 (record_truncated_value): Likewise.
17252 (move_deaths): Likewise.
17253 * cse.c (record_jump_cond): Likewise.
17254 (cse_insn): Likewise.
17255 * cselib.c (cselib_lookup_1): Likewise.
17256 * expmed.c (extract_bit_field_using_extv): Likewise.
17257 * function.c (assign_parm_setup_reg): Likewise.
17258 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17259 * ira-build.c (create_insn_allocnos): Likewise.
17260 * lra-coalesce.c (merge_pseudos): Likewise.
17261 * lra-constraints.c (match_reload): Likewise.
17262 (simplify_operand_subreg): Likewise.
17263 (curr_insn_transform): Likewise.
17264 * lra-lives.c (process_bb_lives): Likewise.
17265 * lra.c (new_insn_reg): Likewise.
17266 (lra_substitute_pseudo): Likewise.
17267 * regcprop.c (mode_change_ok): Likewise.
17268 (maybe_mode_change): Likewise.
17269 (copyprop_hardreg_forward_1): Likewise.
17270 * reload.c (push_reload): Likewise.
17271 (find_reloads): Likewise.
17272 (find_reloads_subreg_address): Likewise.
17273 * reload1.c (alter_reg): Likewise.
17274 (eliminate_regs_1): Likewise.
17275 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17277 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17279 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17280 back to if statements, including unpack.
17282 2017-08-30 Martin Liska <mliska@suse.cz>
17284 PR inline-asm/82001
17285 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17287 (func_checker::compare_asm_inputs_outputs): ... this function.
17288 (func_checker::compare_gimple_asm): Use the function to compare
17289 also ASM constrains.
17290 * ipa-icf-gimple.h: Rename the function.
17292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17293 Alan Hayward <alan.hayward@arm.com>
17294 David Sherwood <david.sherwood@arm.com>
17296 * coretypes.h (complex_mode): New type.
17297 * gdbhooks.py (build_pretty_printer): Handle it.
17298 * machmode.h (complex_mode): New class.
17299 (complex_mode::includes_p): New function.
17300 (is_complex_int_mode): Likewise.
17301 (is_complex_float_mode): Likewise.
17302 * genmodes.c (get_mode_class): Handle complex mode classes.
17303 * function.c (expand_function_end): Use is_complex_int_mode.
17305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17306 Alan Hayward <alan.hayward@arm.com>
17307 David Sherwood <david.sherwood@arm.com>
17309 * coretypes.h (scalar_mode_pod): New typedef.
17310 * gdbhooks.py (build_pretty_printer): Handle it.
17311 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17312 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17313 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17314 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17315 as_a <scalar_mode>.
17317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17318 Alan Hayward <alan.hayward@arm.com>
17319 David Sherwood <david.sherwood@arm.com>
17321 * machmode.h (mode_for_vector): Take a scalar_mode instead
17323 * stor-layout.c (mode_for_vector): Likewise.
17324 * explow.c (promote_mode): Use as_a <scalar_mode>.
17325 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
17327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17328 Alan Hayward <alan.hayward@arm.com>
17329 David Sherwood <david.sherwood@arm.com>
17331 * target.def (preferred_simd_mode): Take a scalar_mode
17332 instead of a machine_mode.
17333 * targhooks.h (default_preferred_simd_mode): Likewise.
17334 * targhooks.c (default_preferred_simd_mode): Likewise.
17335 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17336 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17337 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17338 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17339 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17340 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17341 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17342 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17344 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17345 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17346 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17347 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17348 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17349 * doc/tm.texi: Regenerate.
17350 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17354 Alan Hayward <alan.hayward@arm.com>
17355 David Sherwood <david.sherwood@arm.com>
17357 * target.def (scalar_mode_supported_p): Take a scalar_mode
17358 instead of a machine_mode.
17359 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17360 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17361 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17362 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17363 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17364 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17365 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17366 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17367 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17368 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17369 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17370 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17371 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17372 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17374 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17375 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17376 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17377 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17378 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17380 * doc/tm.texi: Regenerate.
17382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17383 Alan Hayward <alan.hayward@arm.com>
17384 David Sherwood <david.sherwood@arm.com>
17386 * coretypes.h (opt_scalar_mode): New typedef.
17387 * gdbhooks.py (build_pretty_printers): Handle it.
17388 * machmode.h (mode_iterator::get_2xwider): Add overload for
17390 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17392 * expr.c (convert_mode_scalar): Likewise.
17393 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17394 * optabs.c (expand_float): Likewise.
17395 (expand_fix): Likewise.
17396 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17399 Alan Hayward <alan.hayward@arm.com>
17400 David Sherwood <david.sherwood@arm.com>
17402 * optabs.c (expand_float): Explicitly check for scalars before
17403 using a branching expansion.
17404 (expand_fix): Likewise.
17406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17407 Alan Hayward <alan.hayward@arm.com>
17408 David Sherwood <david.sherwood@arm.com>
17410 * expr.c (convert_mode): Split scalar handling out into...
17411 (convert_mode_scalar): ...this new function. Treat the modes
17414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17415 Alan Hayward <alan.hayward@arm.com>
17416 David Sherwood <david.sherwood@arm.com>
17418 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17420 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17423 Alan Hayward <alan.hayward@arm.com>
17424 David Sherwood <david.sherwood@arm.com>
17426 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17427 rather than a machine_mode.
17428 (fixed_from_string): Likewise.
17429 (fixed_convert): Likewise.
17430 (fixed_convert_from_int): Likewise.
17431 (fixed_convert_from_real): Likewise.
17432 (real_convert_from_fixed): Likewise.
17433 * fixed-value.c (fixed_from_double_int): Likewise.
17434 (fixed_from_string): Likewise.
17435 (fixed_convert): Likewise.
17436 (fixed_convert_from_int): Likewise.
17437 (fixed_convert_from_real): Likewise.
17438 (real_convert_from_fixed): Likewise.
17439 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
17441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17442 Alan Hayward <alan.hayward@arm.com>
17443 David Sherwood <david.sherwood@arm.com>
17445 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
17446 of separate mode class checks. Do not allow vector modes here.
17447 (immed_wide_int_const): Use as_a <scalar_mode>.
17448 * explow.c (trunc_int_for_mode): Likewise.
17449 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
17450 (wi::shwi): Likewise.
17451 (wi::min_value): Likewise.
17452 (wi::max_value): Likewise.
17453 * dwarf2out.c (loc_descriptor): Likewise.
17454 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
17455 for CONST_WIDE_INT.
17457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17458 Alan Hayward <alan.hayward@arm.com>
17459 David Sherwood <david.sherwood@arm.com>
17461 * tree.h (SCALAR_TYPE_MODE): New macro.
17462 * expr.c (expand_expr_addr_expr_1): Use it.
17463 (expand_expr_real_2): Likewise.
17464 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
17465 (fold_convert_const_fixed_from_int): Likewise.
17466 (fold_convert_const_fixed_from_real): Likewise.
17467 (native_encode_fixed): Likewise
17468 (native_encode_complex): Likewise
17469 (native_encode_vector): Likewise.
17470 (native_interpret_fixed): Likewise.
17471 (native_interpret_real): Likewise.
17472 (native_interpret_complex): Likewise.
17473 (native_interpret_vector): Likewise.
17474 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
17475 (simd_clone_adjust_argument_types): Likewise.
17476 (simd_clone_init_simd_arrays): Likewise.
17477 (simd_clone_adjust): Likewise.
17478 * stor-layout.c (layout_type): Likewise.
17479 * tree.c (build_minus_one_cst): Likewise.
17480 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17481 * tree-inline.c (estimate_move_cost): Likewise.
17482 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
17483 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
17484 (vectorizable_reduction): Likewise.
17485 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
17486 (vect_recog_mixed_size_cond_pattern): Likewise.
17487 (check_bool_pattern): Likewise.
17488 (adjust_bool_pattern): Likewise.
17489 (search_type_for_mask_1): Likewise.
17490 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
17491 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17492 (vectorizable_load): Likewise.
17493 (vectorizable_store): Likewise.
17494 * ubsan.c (ubsan_encode_value): Likewise.
17495 * varasm.c (output_constant): Likewise.
17497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17498 Alan Hayward <alan.hayward@arm.com>
17499 David Sherwood <david.sherwood@arm.com>
17501 * coretypes.h (scalar_mode): New class.
17502 * machmode.h (scalar_mode): Likewise.
17503 (scalar_mode::includes_p): New function.
17504 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
17505 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
17506 * genmodes.c (get_mode_class): Handle remaining scalar modes.
17507 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
17508 * expmed.c (store_bit_field_1): Likewise.
17509 (extract_bit_field_1): Likewise.
17510 * expr.c (write_complex_part): Likewise.
17511 (read_complex_part): Likewise.
17512 (emit_move_complex_push): Likewise.
17513 (expand_expr_real_2): Likewise.
17514 * function.c (assign_parm_setup_reg): Likewise.
17515 (assign_parms_unsplit_complex): Likewise.
17516 * optabs.c (expand_binop): Likewise.
17517 * rtlanal.c (subreg_get_info): Likewise.
17518 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17519 * varasm.c (output_constant_pool_2): Likewise.
17521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17522 Alan Hayward <alan.hayward@arm.com>
17523 David Sherwood <david.sherwood@arm.com>
17525 * expmed.c (extract_high_half): Use scalar_int_mode and remove
17527 (expmed_mult_highpart_optab): Likewise.
17528 (expmed_mult_highpart): Likewise.
17530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17531 Alan Hayward <alan.hayward@arm.com>
17532 David Sherwood <david.sherwood@arm.com>
17534 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
17535 instead of a machine_mode.
17536 (builtin_memset_read_str): Likewise.
17537 * builtins.c (c_readstr): Likewise.
17538 (builtin_memcpy_read_str): Likewise.
17539 (builtin_strncpy_read_str): Likewise.
17540 (builtin_memset_read_str): Likewise.
17541 (builtin_memset_gen_str): Likewise.
17542 (expand_builtin_signbit): Use scalar_int_mode for local variables.
17543 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
17544 instead of a machine_mode.
17545 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
17547 (make_extraction): Likewise.
17548 (try_widen_shift_mode): Take and return scalar_int_modes instead
17550 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
17551 a scalar_int_mode instead of a machine_mode.
17552 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
17553 (avr_addr_space_pointer_mode): Likewise.
17554 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
17555 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
17556 (msp430_unwind_word_mode): Likewise.
17557 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
17558 (spu_addr_space_pointer_mode): Likewise.
17559 (spu_addr_space_address_mode): Likewise.
17560 (spu_libgcc_cmp_return_mode): Likewise.
17561 (spu_libgcc_shift_count_mode): Likewise.
17562 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
17563 (rl78_addr_space_pointer_mode): Likewise.
17564 (fl78_unwind_word_mode): Likewise.
17565 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
17567 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
17568 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
17569 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
17570 (mips_valid_pointer_mode): Likewise.
17571 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
17572 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
17573 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
17575 (ft32_addr_space_address_mode): Likewise.
17576 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
17577 scalar_int_mode instead of a machine_mode.
17578 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
17580 (m32c_addr_space_address_mode): Likewise.
17581 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
17582 (rs6000_eh_return_filter_mode): Likewise.
17583 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
17584 (rs6000_eh_return_filter_mode): Likewise.
17585 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
17586 (s390_libgcc_shift_count_mode): Likewise.
17587 (s390_unwind_word_mode): Likewise.
17588 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
17590 * target.def (mode_rep_extended): Likewise.
17591 (valid_pointer_mode): Likewise.
17592 (addr_space.valid_pointer_mode): Likewise.
17593 (eh_return_filter_mode): Return a scalar_int_mode rather than
17595 (libgcc_cmp_return_mode): Likewise.
17596 (libgcc_shift_count_mode): Likewise.
17597 (unwind_word_mode): Likewise.
17598 (addr_space.pointer_mode): Likewise.
17599 (addr_space.address_mode): Likewise.
17600 * doc/tm.texi: Regenerate.
17601 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
17603 (do_jump): Use scalar_int_mode for local variables.
17604 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
17605 rather than a machine_mode.
17606 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
17607 (scompare_loc_descriptor_wide): Likewise.
17608 (scompare_loc_descriptor_narrow): Likewise.
17609 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
17611 * except.c (sjlj_emit_dispatch_table): Likewise.
17612 (expand_builtin_eh_copy_values): Likewise.
17613 * explow.c (convert_memory_address_addr_space_1): Likewise.
17614 Take a scalar_int_mode rather than a machine_mode.
17615 (convert_memory_address_addr_space): Take a scalar_int_mode rather
17616 than a machine_mode.
17617 (memory_address_addr_space): Use scalar_int_mode for local variables.
17618 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
17619 rather than a machine_mode.
17620 * expmed.c (mask_rtx): Likewise.
17621 (init_expmed_one_conv): Likewise.
17622 (expand_mult_highpart_adjust): Likewise.
17623 (extract_high_half): Likewise.
17624 (expmed_mult_highpart_optab): Likewise.
17625 (expmed_mult_highpart): Likewise.
17626 (expand_smod_pow2): Likewise.
17627 (expand_sdiv_pow2): Likewise.
17628 (emit_store_flag_int): Likewise.
17629 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
17631 (extract_low_bits): Likewise.
17632 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
17634 * expr.c (pieces_addr::adjust): Likewise.
17635 (can_store_by_pieces): Likewise.
17636 (store_by_pieces): Likewise.
17637 (clear_by_pieces_1): Likewise.
17638 (expand_expr_addr_expr_1): Likewise.
17639 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
17640 (expand_expr_real_1): Likewise.
17641 (try_casesi): Likewise.
17642 * final.c (shorten_branches): Likewise.
17643 * fold-const.c (fold_convert_const_int_from_fixed): Change the
17644 type of "mode" to machine_mode.
17645 * internal-fn.c (expand_arith_overflow_result_store): Take a
17646 scalar_int_mode rather than a machine_mode.
17647 (expand_mul_overflow): Use scalar_int_mode for local variables.
17648 * loop-doloop.c (doloop_modify): Likewise.
17649 (doloop_optimize): Likewise.
17650 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
17651 than a machine_mode.
17652 (expand_doubleword_shift_condmove): Likewise.
17653 (expand_doubleword_shift): Likewise.
17654 (expand_doubleword_clz): Likewise.
17655 (expand_doubleword_popcount): Likewise.
17656 (expand_doubleword_parity): Likewise.
17657 (expand_absneg_bit): Use scalar_int_mode for local variables.
17658 (prepare_float_lib_cmp): Likewise.
17659 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
17660 rather than a machine_mode.
17661 (convert_memory_address_addr_space): Likewise.
17662 (get_mode_bounds): Likewise.
17663 (get_address_mode): Return a scalar_int_mode rather than a
17665 * rtlanal.c (get_address_mode): Likewise.
17666 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
17667 than a machine_mode.
17668 * targhooks.c (default_mode_rep_extended): Likewise.
17669 (default_valid_pointer_mode): Likewise.
17670 (default_addr_space_valid_pointer_mode): Likewise.
17671 (default_eh_return_filter_mode): Return a scalar_int_mode rather
17672 than a machine_mode.
17673 (default_libgcc_cmp_return_mode): Likewise.
17674 (default_libgcc_shift_count_mode): Likewise.
17675 (default_unwind_word_mode): Likewise.
17676 (default_addr_space_pointer_mode): Likewise.
17677 (default_addr_space_address_mode): Likewise.
17678 * targhooks.h (default_eh_return_filter_mode): Likewise.
17679 (default_libgcc_cmp_return_mode): Likewise.
17680 (default_libgcc_shift_count_mode): Likewise.
17681 (default_unwind_word_mode): Likewise.
17682 (default_addr_space_pointer_mode): Likewise.
17683 (default_addr_space_address_mode): Likewise.
17684 (default_mode_rep_extended): Take a scalar_int_mode rather than
17686 (default_valid_pointer_mode): Likewise.
17687 (default_addr_space_valid_pointer_mode): Likewise.
17688 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
17690 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
17691 rather than a machine_mode.
17692 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
17693 for local variables.
17694 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17695 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
17696 than a machine_mode.
17698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17699 Alan Hayward <alan.hayward@arm.com>
17700 David Sherwood <david.sherwood@arm.com>
17702 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
17703 the mode argument to scalar_int_mode.
17704 (do_jump_by_parts_zero_rtx): Likewise.
17705 (do_jump_by_parts_equality_rtx): Likewise.
17706 (do_jump_by_parts_greater): Take a mode argument.
17707 (do_jump_by_parts_equality): Likewise.
17708 (do_jump_1): Update calls accordingly.
17710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17711 Alan Hayward <alan.hayward@arm.com>
17712 David Sherwood <david.sherwood@arm.com>
17714 * is-a.h (safe_dyn_cast): New function.
17715 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
17716 (jump_table_for_label): Likewise.
17717 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
17718 instead of an rtx_insn *.
17719 (shorten_branches): Use dyn_cast instead of LABEL_P and
17720 JUMP_TABLE_DATA_P. Use jump_table_for_label and
17721 rtx_jump_table_data::get_data_mode.
17722 (final_scan_insn): Likewise.
17724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17725 Alan Hayward <alan.hayward@arm.com>
17726 David Sherwood <david.sherwood@arm.com>
17728 * combine.c (try_combine): Use is_a <scalar_int_mode> when
17729 trying to combine a full-register integer set with a subreg
17732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17733 Alan Hayward <alan.hayward@arm.com>
17734 David Sherwood <david.sherwood@arm.com>
17736 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
17737 that is always either address_mode or pointer_mode.
17739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17740 Alan Hayward <alan.hayward@arm.com>
17741 David Sherwood <david.sherwood@arm.com>
17743 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
17744 when the two are known to be equal.
17746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17747 Alan Hayward <alan.hayward@arm.com>
17748 David Sherwood <david.sherwood@arm.com>
17750 * simplify-rtx.c (simplify_const_unary_operation): Use
17751 is_a <scalar_int_mode> instead of checking for a nonzero
17752 precision. Forcibly convert op_mode to a scalar_int_mode
17753 in that case. More clearly differentiate the operand and
17754 result modes and use the former when deciding what the value
17755 of a count-bits operation should be. Use is_int_mode instead
17756 of checking for a MODE_INT. Remove redundant check for whether
17757 this mode has a zero precision.
17759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17760 Alan Hayward <alan.hayward@arm.com>
17761 David Sherwood <david.sherwood@arm.com>
17763 * optabs.c (widen_leading): Change the type of the mode argument
17764 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
17765 (widen_bswap): Likewise.
17766 (expand_parity): Likewise.
17767 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
17768 (expand_ffs): Likewise.
17769 (epand_unop): Check for scalar integer modes before calling the
17772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17773 Alan Hayward <alan.hayward@arm.com>
17774 David Sherwood <david.sherwood@arm.com>
17776 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
17778 (expand_expr_real_1): Update call accordingly.
17780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17781 Alan Hayward <alan.hayward@arm.com>
17782 David Sherwood <david.sherwood@arm.com>
17784 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17785 value_mode arguments. Use scalar_int_mode internally.
17786 (store_bit_field_1): Rename the new integer mode from imode
17787 to op0_mode and use it instead of GET_MODE (op0). Update calls
17788 to store_split_bit_field, store_bit_field_using_insv and
17789 store_fixed_bit_field.
17790 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17791 Use scalar_int_mode internally. Use a bit count rather than a mode
17792 when calculating the largest bit size for get_best_mode.
17793 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17794 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17795 Remove assertion that OP0 has a scalar integer mode.
17796 (store_split_bit_field): Add op0_mode and value_mode arguments.
17797 Update calls to extract_fixed_bit_field.
17798 (extract_bit_field_using_extv): Add an op0_mode argument.
17799 Use scalar_int_mode internally.
17800 (extract_bit_field_1): Rename the new integer mode from imode to
17801 op0_mode and use it instead of GET_MODE (op0). Update calls to
17802 extract_split_bit_field, extract_bit_field_using_extv and
17803 extract_fixed_bit_field.
17804 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17805 to extract_split_bit_field and extract_fixed_bit_field_1.
17806 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17807 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17808 on the target mode.
17809 (extract_split_bit_field): Add an op0_mode argument. Update call
17810 to extract_fixed_bit_field.
17812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17813 Alan Hayward <alan.hayward@arm.com>
17814 David Sherwood <david.sherwood@arm.com>
17816 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17817 * explow.c (hard_function_value): Likewise.
17818 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17819 convert_to_mode call outside the loop.
17820 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17821 for the mode iterator. Require the mode specified by max_pieces
17823 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17825 (copy_blkmode_to_reg): Likewise.
17826 (set_storage_via_setmem): Likewise.
17827 * optabs.c (prepare_cmp_insn): Likewise.
17828 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17829 * stor-layout.c (finish_bitfield_representative): Likewise.
17831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17832 Alan Hayward <alan.hayward@arm.com>
17833 David Sherwood <david.sherwood@arm.com>
17835 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17836 * expr.c (convert_move): Use them.
17837 (convert_modes): Likewise.
17838 (store_expr_with_bounds): Likewise.
17840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17841 Alan Hayward <alan.hayward@arm.com>
17842 David Sherwood <david.sherwood@arm.com>
17844 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17845 parameter for the mode of "x". Remove the "known_x", "known_mode"
17846 and "known_ret" arguments. Change the type of the mode argument
17847 to scalar_int_mode.
17848 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17849 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17850 (reg_num_sign_bit_copies_for_combine): Likewise.
17851 * rtlanal.c (nonzero_bits1): Likewise.
17852 (num_sign_bit_copies1): Likewise.
17853 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17854 (reg_num_sign_bit_copies_general): Likewise.
17855 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17856 (reg_nonzero_bits_general): Likewise.
17858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17859 Alan Hayward <alan.hayward@arm.com>
17860 David Sherwood <david.sherwood@arm.com>
17862 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17863 than in subroutines. Return 1 for non-integer modes.
17864 (cached_num_sign_bit_copies): Change the type of the mode parameter
17865 to scalar_int_mode.
17866 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17867 classes. Handle CONST_INT_P first and then check whether X also
17868 has a scalar integer mode. Check the same thing for inner registers
17869 of a SUBREG and for values that are being extended or truncated.
17871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17872 Alan Hayward <alan.hayward@arm.com>
17873 David Sherwood <david.sherwood@arm.com>
17875 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17876 in subroutines. Return the mode mask for non-integer modes.
17877 (cached_nonzero_bits): Change the type of the mode parameter
17878 to scalar_int_mode.
17879 (nonzero_bits1): Likewise. Remove early exit for other mode
17880 classes. Handle CONST_INT_P first and then check whether X
17881 also has a scalar integer mode.
17883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17884 Alan Hayward <alan.hayward@arm.com>
17885 David Sherwood <david.sherwood@arm.com>
17887 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17888 Return a scalar_int_mode and assert that the size is greater than
17890 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17891 (op_by_pieces_d::op_by_pieces_d): Likewise.
17892 (op_by_pieces_d::run): Likewise.
17893 (can_store_by_pieces): Likewise.
17895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17896 Alan Hayward <alan.hayward@arm.com>
17897 David Sherwood <david.sherwood@arm.com>
17899 * combine.c (extract_left_shift): Add a mode argument and update
17901 (make_compound_operation_int): Change the type of the mode parameter
17902 to scalar_int_mode and update the call to extract_left_shift.
17904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17905 Alan Hayward <alan.hayward@arm.com>
17906 David Sherwood <david.sherwood@arm.com>
17908 * combine.c (simplify_and_const_int): Change the type of the mode
17909 parameter to scalar_int_mode.
17910 (simplify_and_const_int_1): Likewise. Update recursive call.
17912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17913 Alan Hayward <alan.hayward@arm.com>
17914 David Sherwood <david.sherwood@arm.com>
17916 * combine.c (simplify_compare_const): Check that the mode is a
17917 scalar_int_mode (rather than VOIDmode) before testing its
17919 (simplify_comparison): Move COMPARISON_P handling out of the
17920 loop and restrict the latter part of the loop to scalar_int_modes.
17921 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17922 and when considering SUBREG_REGs. Use is_int_mode instead of
17923 checking GET_MODE_CLASS against MODE_INT.
17925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17926 Alan Hayward <alan.hayward@arm.com>
17927 David Sherwood <david.sherwood@arm.com>
17929 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17930 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17931 shift_unit_mode and for modes involved in scalar shifts.
17933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17934 Alan Hayward <alan.hayward@arm.com>
17935 David Sherwood <david.sherwood@arm.com>
17937 * combine.c (force_int_to_mode): New function, split out from...
17938 (force_to_mode): ...here. Keep xmode up-to-date and use it
17939 instead of GET_MODE (x).
17941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17942 Alan Hayward <alan.hayward@arm.com>
17943 David Sherwood <david.sherwood@arm.com>
17945 * optabs-query.h (extraction_insn::struct_mode): Change type to
17946 opt_scalar_int_mode and update comment.
17947 (extraction_insn::field_mode): Change type to scalar_int_mode.
17948 (extraction_insn::pos_mode): Likewise.
17949 * combine.c (make_extraction): Update accordingly.
17950 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17951 (get_optab_extraction_insn): Likewise.
17952 * recog.c (simplify_while_replacing): Likewise.
17953 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17954 parameter to opt_scalar_int_mode.
17956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17957 Alan Hayward <alan.hayward@arm.com>
17958 David Sherwood <david.sherwood@arm.com>
17960 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17961 to a scalar_int_mode instead of a machine_mode.
17962 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17963 (get_best_mode): Return a boolean and use a pointer argument to store
17964 the selected mode. Replace the limit mode parameter with a bit limit.
17965 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17966 for the values returned by bit_field_mode_iterator::next_mode.
17967 (store_bit_field): Update call to get_best_mode.
17968 (store_fixed_bit_field): Likewise.
17969 (extract_fixed_bit_field): Likewise.
17970 * expr.c (optimize_bitfield_assignment_op): Likewise.
17971 * fold-const.c (optimize_bit_field_compare): Likewise.
17972 (fold_truth_andor_1): Likewise.
17973 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17974 Update for new type of m_mode.
17975 (get_best_mode): As above.
17977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17978 Alan Hayward <alan.hayward@arm.com>
17979 David Sherwood <david.sherwood@arm.com>
17981 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17982 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17983 (store_bit_field): Check is_a <scalar_int_mode> before calling
17984 strict_volatile_bitfield_p.
17985 (extract_bit_field): Likewise.
17987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17988 Alan Hayward <alan.hayward@arm.com>
17989 David Sherwood <david.sherwood@arm.com>
17991 * target.def (cstore_mode): Return a scalar_int_mode.
17992 * doc/tm.texi: Regenerate.
17993 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17994 * targhooks.h (default_cstore_mode): Likewise.
17995 * targhooks.c (default_cstore_mode): Likewise, using a forced
17997 * expmed.c (emit_cstore): Expect the target of the cstore to be
18000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18001 Alan Hayward <alan.hayward@arm.com>
18002 David Sherwood <david.sherwood@arm.com>
18004 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18006 (niter_desc): Likewise mode.
18007 (iv_analyze): Add a mode parameter.
18009 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18010 and change its type to scalar_int_mode.
18011 * loop-iv.c: Update commentary at head of file.
18012 (iv_constant): Pass the mode paraeter before the rtx it describes
18013 and change its type to scalar_int_mode. Remove VOIDmode handling.
18014 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18015 (iv_extend): Likewise.
18016 (shorten_into_mode): Likewise.
18017 (iv_add): Use scalar_int_mode.
18018 (iv_mult): Likewise.
18019 (iv_shift): Likewise.
18020 (canonicalize_iv_subregs): Likewise.
18021 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18022 it describes and change its mode to scalar_int_mode. Also change
18023 the type of the returned inner_mode to scalar_int_mode.
18024 (get_biv_step): Likewise, turning outer_mode from a pointer
18025 into a direct parameter. Update call to get_biv_step_1.
18026 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18027 iv_constant and get_biv_step.
18028 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18029 and change its type to scalar_int_mode. Don't initialise iv->mode
18030 to VOIDmode and remove later checks for its still being VOIDmode.
18031 Update calls to iv_analyze_op and iv_analyze_expr. Check
18032 is_a <scalar_int_mode> when changing the mode under consideration.
18033 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18034 Update call to iv_analyze_expr.
18035 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18036 inner register is not also a scalar_int_mode. Update call to
18038 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18039 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18040 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18041 separate mode class checks. Update calls to iv_analyze. Remove
18042 fix-up of VOIDmodes after iv_analyze_biv.
18043 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18044 don't have a scalar_int_mode. Update call to biv_p.
18046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18047 Alan Hayward <alan.hayward@arm.com>
18048 David Sherwood <david.sherwood@arm.com>
18050 * cfgexpand.c (convert_debug_memory_address): Use
18051 as_a <scalar_int_mode>.
18052 * combine.c (expand_compound_operation): Likewise.
18053 (make_extraction): Likewise.
18054 (change_zero_ext): Likewise.
18055 (simplify_comparison): Likewise.
18056 * cse.c (cse_insn): Likewise.
18057 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18058 (mem_loc_descriptor): Likewise.
18059 (loc_descriptor): Likewise.
18060 * expmed.c (init_expmed_one_mode): Likewise.
18061 (synth_mult): Likewise.
18062 (emit_store_flag_1): Likewise.
18063 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18064 of a comparison with size.
18065 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18066 (reduce_to_bit_field_precision): Likewise.
18067 * function.c (expand_function_end): Likewise.
18068 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18069 * loop-doloop.c (doloop_modify): Likewise.
18070 * optabs.c (expand_binop): Likewise.
18071 (expand_unop): Likewise.
18072 (expand_copysign_absneg): Likewise.
18073 (prepare_cmp_insn): Likewise.
18074 (maybe_legitimize_operand): Likewise.
18075 * recog.c (const_scalar_int_operand): Likewise.
18076 * rtlanal.c (get_address_mode): Likewise.
18077 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18078 (simplify_cond_clz_ctz): Likewise.
18079 * tree-nested.c (get_nl_goto_field): Likewise.
18080 * tree.c (build_vector_type_for_mode): Likewise.
18081 * var-tracking.c (use_narrower_mode): Likewise.
18083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18084 Alan Hayward <alan.hayward@arm.com>
18085 David Sherwood <david.sherwood@arm.com>
18087 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18088 * builtins.c (expand_builtin_signbit): Use it.
18089 * cfgexpand.c (expand_debug_expr): Likewise.
18090 * dojump.c (do_jump): Likewise.
18091 (do_compare_and_jump): Likewise.
18092 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18093 * expmed.c (make_tree): Likewise.
18094 * expr.c (expand_expr_real_2): Likewise.
18095 (expand_expr_real_1): Likewise.
18096 (try_casesi): Likewise.
18097 * fold-const-call.c (fold_const_call_ss): Likewise.
18098 * fold-const.c (unextend): Likewise.
18099 (extract_muldiv_1): Likewise.
18100 (fold_single_bit_test): Likewise.
18101 (native_encode_int): Likewise.
18102 (native_encode_string): Likewise.
18103 (native_interpret_int): Likewise.
18104 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18105 * internal-fn.c (expand_addsub_overflow): Likewise.
18106 (expand_neg_overflow): Likewise.
18107 (expand_mul_overflow): Likewise.
18108 (expand_arith_overflow): Likewise.
18109 * match.pd: Likewise.
18110 * stor-layout.c (layout_type): Likewise.
18111 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18112 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18113 * tree-ssanames.c (get_range_info): Likewise.
18114 * tree-switch-conversion.c (array_value_type) Likewise.
18115 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18116 (vect_recog_divmod_pattern): Likewise.
18117 (vect_recog_mixed_size_cond_pattern): Likewise.
18118 * tree-vrp.c (extract_range_basic): Likewise.
18119 (simplify_float_conversion_using_ranges): Likewise.
18120 * tree.c (int_fits_type_p): Likewise.
18121 * ubsan.c (instrument_bool_enum_load): Likewise.
18122 * varasm.c (mergeable_string_section): Likewise.
18123 (narrowing_initializer_constant_valid_p): Likewise.
18124 (output_constant): Likewise.
18126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18127 Alan Hayward <alan.hayward@arm.com>
18128 David Sherwood <david.sherwood@arm.com>
18130 * machmode.h (NARROWEST_INT_MODE): New macro.
18131 * expr.c (alignment_for_piecewise_move): Use it instead of
18132 GET_CLASS_NARROWEST_MODE (MODE_INT).
18133 (push_block): Likewise.
18134 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18136 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18139 Alan Hayward <alan.hayward@arm.com>
18140 David Sherwood <david.sherwood@arm.com>
18142 * postreload.c (move2add_valid_value_p): Change the type of the
18143 mode parameter to scalar_int_mode.
18144 (move2add_use_add2_insn): Add a mode parameter and use it instead
18146 (move2add_use_add3_insn): Likewise.
18147 (reload_cse_move2add): Update accordingly.
18149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18150 Alan Hayward <alan.hayward@arm.com>
18151 David Sherwood <david.sherwood@arm.com>
18153 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18155 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18156 * optabs.c (expand_unop): Likewise.
18158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18159 Alan Hayward <alan.hayward@arm.com>
18160 David Sherwood <david.sherwood@arm.com>
18162 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18163 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18164 (popcount_loc_descriptor): Likewise.
18165 (bswap_loc_descriptor): Likewise.
18166 (rotate_loc_descriptor): Likewise.
18167 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18168 calling the functions above.
18170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18171 Alan Hayward <alan.hayward@arm.com>
18172 David Sherwood <david.sherwood@arm.com>
18174 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18176 (try_combine): Likewise.
18177 (simplify_if_then_else): Likewise.
18178 * cse.c (cse_insn): Likewise.
18179 * dwarf2out.c (mem_loc_descriptor): Likewise.
18180 * emit-rtl.c (gen_lowpart_common): Likewise.
18181 * simplify-rtx.c (simplify_truncation): Likewise.
18182 (simplify_binary_operation_1): Likewise.
18183 (simplify_const_relational_operation): Likewise.
18184 (simplify_ternary_operation): Likewise.
18185 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18188 Alan Hayward <alan.hayward@arm.com>
18189 David Sherwood <david.sherwood@arm.com>
18191 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18192 * reload.c (push_reload): Likewise.
18193 (find_reloads): Likewise.
18195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18196 Alan Hayward <alan.hayward@arm.com>
18197 David Sherwood <david.sherwood@arm.com>
18199 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18200 (make_compound_operation_int): Likewise.
18201 (change_zero_ext): Likewise.
18202 * expr.c (convert_move): Likewise.
18203 (convert_modes): Likewise.
18204 * fwprop.c (forward_propagate_subreg): Likewise.
18205 * loop-iv.c (get_biv_step_1): Likewise.
18206 * optabs.c (widen_operand): Likewise.
18207 * postreload.c (move2add_valid_value_p): Likewise.
18208 * recog.c (simplify_while_replacing): Likewise.
18209 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18210 (simplify_binary_operation_1): Likewise. Remove redundant
18211 mode equality check.
18213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18214 Alan Hayward <alan.hayward@arm.com>
18215 David Sherwood <david.sherwood@arm.com>
18217 * combine.c (combine_simplify_rtx): Add checks for
18218 is_a <scalar_int_mode>.
18219 (simplify_if_then_else): Likewise.
18220 (make_field_assignment): Likewise.
18221 (simplify_comparison): Likewise.
18222 * ifcvt.c (noce_try_bitop): Likewise.
18223 * loop-invariant.c (canonicalize_address_mult): Likewise.
18224 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18227 Alan Hayward <alan.hayward@arm.com>
18228 David Sherwood <david.sherwood@arm.com>
18230 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18231 is_a <scalar_int_mode> instead of != BLKmode.
18233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18234 Alan Hayward <alan.hayward@arm.com>
18235 David Sherwood <david.sherwood@arm.com>
18237 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18238 instead of != VOIDmode.
18239 * combine.c (if_then_else_cond): Likewise.
18240 (change_zero_ext): Likewise.
18241 * dwarf2out.c (mem_loc_descriptor): Likewise.
18242 (loc_descriptor): Likewise.
18243 * rtlanal.c (canonicalize_condition): Likewise.
18244 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18247 Alan Hayward <alan.hayward@arm.com>
18248 David Sherwood <david.sherwood@arm.com>
18250 * simplify-rtx.c (simplify_binary_operation_1): Use
18251 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18254 Alan Hayward <alan.hayward@arm.com>
18255 David Sherwood <david.sherwood@arm.com>
18257 * wide-int.h (int_traits<unsigned char>) New class.
18258 (int_traits<unsigned short>) Likewise.
18259 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18260 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18262 * combine.c (set_nonzero_bits_and_sign_copies): Use
18263 is_a <scalar_int_mode>.
18264 (find_split_point): Likewise.
18265 (combine_simplify_rtx): Likewise.
18266 (simplify_logical): Likewise.
18267 (expand_compound_operation): Likewise.
18268 (expand_field_assignment): Likewise.
18269 (make_compound_operation): Likewise.
18270 (extended_count): Likewise.
18271 (change_zero_ext): Likewise.
18272 (simplify_comparison): Likewise.
18273 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18274 (ucompare_loc_descriptor): Likewise.
18275 (minmax_loc_descriptor): Likewise.
18276 (mem_loc_descriptor): Likewise.
18277 (loc_descriptor): Likewise.
18278 * expmed.c (init_expmed_one_mode): Likewise.
18279 * lra-constraints.c (lra_constraint_offset): Likewise.
18280 * optabs.c (prepare_libcall_arg): Likewise.
18281 * postreload.c (move2add_note_store): Likewise.
18282 * reload.c (operands_match_p): Likewise.
18283 * rtl.h (load_extend_op): Likewise.
18284 * rtlhooks.c (gen_lowpart_general): Likewise.
18285 * simplify-rtx.c (simplify_truncation): Likewise.
18286 (simplify_unary_operation_1): Likewise.
18287 (simplify_binary_operation_1): Likewise.
18288 (simplify_const_binary_operation): Likewise.
18289 (simplify_const_relational_operation): Likewise.
18290 (simplify_subreg): Likewise.
18291 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18292 * var-tracking.c (adjust_mems): Likewise.
18293 (prepare_call_arguments): Likewise.
18295 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18296 Alan Hayward <alan.hayward@arm.com>
18297 David Sherwood <david.sherwood@arm.com>
18299 * machmode.h (is_int_mode): New fuction.
18300 * combine.c (find_split_point): Use it.
18301 (combine_simplify_rtx): Likewise.
18302 (simplify_if_then_else): Likewise.
18303 (simplify_set): Likewise.
18304 (simplify_shift_const_1): Likewise.
18305 (simplify_comparison): Likewise.
18306 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18307 * cse.c (notreg_cost): Likewise.
18308 (cse_insn): Likewise.
18309 * cselib.c (cselib_lookup_1): Likewise.
18310 * dojump.c (do_jump_1): Likewise.
18311 (do_compare_rtx_and_jump): Likewise.
18312 * dse.c (get_call_args): Likewise.
18313 * dwarf2out.c (rtl_for_decl_init): Likewise.
18314 (native_encode_initializer): Likewise.
18315 * expmed.c (emit_store_flag_1): Likewise.
18316 (emit_store_flag): Likewise.
18317 * expr.c (convert_modes): Likewise.
18318 (store_field): Likewise.
18319 (expand_expr_real_1): Likewise.
18320 * fold-const.c (fold_read_from_constant_string): Likewise.
18321 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18322 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18323 * optabs.c (expand_binop): Likewise.
18324 (expand_unop): Likewise.
18325 (expand_abs_nojump): Likewise.
18326 (expand_one_cmpl_abs_nojump): Likewise.
18327 * simplify-rtx.c (mode_signbit_p): Likewise.
18328 (val_signbit_p): Likewise.
18329 (val_signbit_known_set_p): Likewise.
18330 (val_signbit_known_clear_p): Likewise.
18331 (simplify_relational_operation_1): Likewise.
18332 * tree.c (vector_type_mode): Likewise.
18334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18335 Alan Hayward <alan.hayward@arm.com>
18336 David Sherwood <david.sherwood@arm.com>
18338 * machmode.h (smallest_mode_for_size): Fix formatting.
18339 (smallest_int_mode_for_size): New function.
18340 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18341 instead of smallest_mode_for_size.
18342 * combine.c (make_extraction): Likewise.
18343 * config/arc/arc.c (arc_expand_movmem): Likewise.
18344 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18345 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18346 * config/s390/s390.c (s390_expand_insv): Likewise.
18347 * config/sparc/sparc.c (assign_int_registers): Likewise.
18348 * config/spu/spu.c (spu_function_value): Likewise.
18349 (spu_function_arg): Likewise.
18350 * coverage.c (get_gcov_type): Likewise.
18351 (get_gcov_unsigned_t): Likewise.
18352 * dse.c (find_shift_sequence): Likewise.
18353 * expmed.c (store_bit_field_1): Likewise.
18354 * expr.c (convert_move): Likewise.
18355 (store_field): Likewise.
18356 * internal-fn.c (expand_arith_overflow): Likewise.
18357 * optabs-query.c (get_best_extraction_insn): Likewise.
18358 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18359 * stor-layout.c (layout_type): Likewise.
18360 (initialize_sizetypes): Likewise.
18361 * targhooks.c (default_get_mask_mode): Likewise.
18362 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18365 Alan Hayward <alan.hayward@arm.com>
18366 David Sherwood <david.sherwood@arm.com>
18368 * machmode.h (opt_mode::else_blk): New function.
18369 (int_mode_for_mode): Declare.
18370 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18371 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18373 * cfgexpand.c (expand_debug_expr): Likewise.
18374 * combine.c (gen_lowpart_or_truncate): Likewise.
18375 (gen_lowpart_for_combine): Likewise.
18376 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18377 * config/avr/avr.c (avr_to_int_mode): Likewise.
18378 (avr_out_plus_1): Likewise.
18379 (avr_out_plus): Likewise.
18380 (avr_out_round): Likewise.
18381 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18382 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18383 (s390_expand_vcond): Likewise.
18384 * config/spu/spu.c (spu_split_immediate): Likewise.
18385 (spu_expand_mov): Likewise.
18386 * dse.c (get_stored_val): Likewise.
18387 * expmed.c (store_bit_field_1): Likewise.
18388 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18390 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18391 (extract_low_bits): Likewise.
18392 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18393 handling rather than repeating the check.
18394 (emit_group_store): Likewise.
18395 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18396 * optabs.c (expand_absneg_bit): Likewise.
18397 (expand_copysign_absneg): Likewise.
18398 (expand_copysign_bit): Likewise.
18399 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18400 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18401 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18402 * var-tracking.c (prepare_call_arguments): Likewise.
18403 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18404 int_mode_for_mode instead of mode_for_size.
18405 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18408 Alan Hayward <alan.hayward@arm.com>
18409 David Sherwood <david.sherwood@arm.com>
18411 * machmode.h (int_mode_for_size): New function.
18412 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18413 instead of mode_for_size.
18414 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18416 * combine.c (expand_field_assignment): Use int_mode_for_size
18417 instead of mode_for_size.
18418 (make_extraction): Likewise.
18419 (simplify_shift_const_1): Likewise.
18420 (simplify_comparison): Likewise.
18421 * dojump.c (do_jump): Likewise.
18422 * dwarf2out.c (mem_loc_descriptor): Likewise.
18423 * emit-rtl.c (init_derived_machine_modes): Likewise.
18424 * expmed.c (flip_storage_order): Likewise.
18425 (convert_extracted_bit_field): Likewise.
18426 * expr.c (copy_blkmode_from_reg): Likewise.
18427 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18428 * internal-fn.c (expand_mul_overflow): Likewise.
18429 * lower-subreg.c (simple_move): Likewise.
18430 * optabs-libfuncs.c (init_optabs): Likewise.
18431 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18432 * tree.c (vector_type_mode): Likewise.
18433 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18434 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
18435 * tree-vect-generic.c (expand_vector_parallel): Likewise.
18436 * tree-vect-stmts.c (vectorizable_load): Likewise.
18437 (vectorizable_store): Likewise.
18439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18440 Alan Hayward <alan.hayward@arm.com>
18441 David Sherwood <david.sherwood@arm.com>
18443 * coretypes.h (pod_mode): New type.
18444 (scalar_int_mode_pod): New typedef.
18445 * machmode.h (pod_mode): New class.
18446 (int_n_data_t::m): Change type to scalar_int_mode_pod.
18447 * genmodes.c (emit_mode_int_n): Update accordingly.
18448 * lower-subreg.h (target_lower_subreg): Change type to
18449 scalar_int_mode_pod.
18450 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
18451 scalar_int_mode_pod.
18453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18454 Alan Hayward <alan.hayward@arm.com>
18455 David Sherwood <david.sherwood@arm.com>
18457 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
18458 machine_mode to scalar_int_mode.
18459 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
18460 (rs6000_option_override_internal): Remove cast to int.
18461 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
18462 machine_mode to scalar_int_mode.
18463 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
18464 (rs6000_option_override_internal): Remove cast to int.
18465 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
18466 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
18468 * config/s390/s390.c (s390_expand_builtin): Likewise.
18469 * coretypes.h (scalar_int_mode): New type.
18470 (opt_scalar_int_mode): New typedef.
18471 * machmode.h (scalar_int_mode): New class.
18472 (scalar_int_mode::includes_p): New function.
18473 (byte_mode): Change type to scalar_int_mode.
18474 (word_mode): Likewise.
18475 (ptr_mode): Likewise.
18476 * emit-rtl.c (byte_mode): Likewise.
18477 (word_mode): Likewise.
18478 (ptr_mode): Likewise.
18479 (init_derived_machine_modes): Update accordingly.
18480 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
18481 and MODE_PARTIAL_INT.
18482 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
18483 opt_scalar_int_mode.
18485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18486 Alan Hayward <alan.hayward@arm.com>
18487 David Sherwood <david.sherwood@arm.com>
18489 * target.def (libgcc_floating_mode_supported_p): Take a
18491 * doc/tm.texi: Regenerate.
18492 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
18494 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18495 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
18498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18499 Alan Hayward <alan.hayward@arm.com>
18500 David Sherwood <david.sherwood@arm.com>
18502 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
18503 * doc/tm.texi: Regenerate.
18504 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
18505 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
18506 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
18507 * targhooks.h (default_floatn_mode): Likewise.
18508 * targhooks.c (default_floatn_mode): Likewise.
18509 * tree.c (build_common_tree_nodes): Update accordingly.
18511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18512 Alan Hayward <alan.hayward@arm.com>
18513 David Sherwood <david.sherwood@arm.com>
18515 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
18516 (mode_iterator::iterate_p): Likewise.
18517 (mode_iterator::get_wider): Likewise.
18518 * expr.c (init_expr_target): Use opt_scalar_float_mode.
18520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18521 Alan Hayward <alan.hayward@arm.com>
18522 David Sherwood <david.sherwood@arm.com>
18524 * coretypes.h (opt_scalar_float_mode): New typedef.
18525 * machmode.h (float_mode_for_size): New function.
18526 * emit-rtl.c (double_mode): Delete.
18527 (init_emit_once): Use float_mode_for_size.
18528 * stor-layout.c (layout_type): Likewise.
18529 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
18531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18532 Alan Hayward <alan.hayward@arm.com>
18533 David Sherwood <david.sherwood@arm.com>
18535 * output.h (assemble_real): Take a scalar_float_mode.
18536 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
18537 * config/arm/arm.md (consttable_4): Likewise.
18538 (consttable_8): Likewise.
18539 (consttable_16): Likewise.
18540 * config/mips/mips.md (consttable_float): Likewise.
18541 * config/s390/s390.c (s390_output_pool_entry): Likewise.
18542 * varasm.c (assemble_real): Take a scalar_float_mode.
18543 (output_constant_pool_2): Update accordingly.
18544 (output_constant): Likewise.
18546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18547 Alan Hayward <alan.hayward@arm.com>
18548 David Sherwood <david.sherwood@arm.com>
18550 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
18551 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
18552 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
18553 (native_encode_real): Likewise.
18554 (native_interpret_real): Likewise.
18555 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
18556 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18559 Alan Hayward <alan.hayward@arm.com>
18560 David Sherwood <david.sherwood@arm.com>
18562 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
18563 <scalar_float_mode>. Simplify.
18564 (gen_extend_conv_libfunc): Likewise.
18566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18567 Alan Hayward <alan.hayward@arm.com>
18568 David Sherwood <david.sherwood@arm.com>
18570 * coretypes.h (scalar_float_mode): New type.
18571 * machmode.h (mode_traits::from_int): Use machine_mode if
18572 USE_ENUM_MODES is defined.
18573 (is_a): New function.
18575 (dyn_cast): Likewise.
18576 (scalar_float_mode): New class.
18577 (scalar_float_mode::includes_p): New function.
18578 (is_float_mode): Likewise.
18579 * gdbhooks.py (MachineModePrinter): New class.
18580 (build_pretty_printer): Use it for scalar_float_mode.
18581 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
18582 (format_helper::format_helper): Turn into a template.
18583 * genmodes.c (get_mode_class): New function.
18584 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
18585 or machine_mode if none.
18586 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
18587 as_a <scalar_float_mode>.
18588 * dwarf2out.c (mem_loc_descriptor): Likewise.
18589 (insert_float): Likewise.
18590 (add_const_value_attribute): Likewise.
18591 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18592 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
18593 (expand_unop): Update accordingly.
18594 (expand_abs_nojump): Likewise.
18595 (expand_copysign_absneg): Take a scalar_float_mode.
18596 (expand_copysign_bit): Likewise.
18597 (expand_copysign): Update accordingly.
18599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18600 Alan Hayward <alan.hayward@arm.com>
18601 David Sherwood <david.sherwood@arm.com>
18603 * coretypes.h (opt_mode): New class.
18604 * machmode.h (opt_mode): Likewise.
18605 (opt_mode::else_void): New function.
18606 (opt_mode::require): Likewise.
18607 (opt_mode::exists): Likewise.
18608 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
18609 (GET_MODE_2XWIDER_MODE): Likewise.
18610 (mode_iterator::get_wider): Update accordingly.
18611 (mode_iterator::get_2xwider): Likewise.
18612 (mode_iterator::get_known_wider): Likewise, turning into a template.
18613 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
18614 forcing a wider mode to exist.
18615 * config/cr16/cr16.h (LONG_REG_P): Likewise.
18616 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18617 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
18618 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
18619 * lower-subreg.c (init_lower_subreg): Likewise.
18620 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
18621 on the final iteration.
18622 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
18623 a wider mode exists before asking for a move pattern.
18624 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
18625 forcing a wider mode to exist.
18626 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
18627 returning false if no such mode exists.
18628 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
18629 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
18630 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
18631 Avoid checking for a MODE_INT if we already know the mode is not a
18633 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
18634 forcing a wider mode to exist.
18635 (expmed_mult_highpart_optab): Likewise.
18636 (expmed_mult_highpart): Likewise.
18637 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
18639 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18640 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
18641 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
18642 * internal-fn.c (expand_mul_overflow): Update use of
18643 GET_MODE_2XWIDER_MODE.
18644 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18645 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
18646 GET_MODE_WIDER_MODE.
18647 (convert_plusminus_to_widen): Likewise.
18648 * tree-switch-conversion.c (array_value_type): Likewise.
18649 * var-tracking.c (emit_note_insn_var_location): Likewise.
18650 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18651 Return false inside rather than outside the loop if no wider mode
18653 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
18654 and GET_MODE_2XWIDER_MODE
18655 (can_compare_p): Use else_void.
18656 * gdbhooks.py (OptMachineModePrinter): New class.
18657 (build_pretty_printer): Use it for opt_mode.
18659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18660 Alan Hayward <alan.hayward@arm.com>
18661 David Sherwood <david.sherwood@arm.com>
18663 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
18664 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
18666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18667 Alan Hayward <alan.hayward@arm.com>
18668 David Sherwood <david.sherwood@arm.com>
18670 * machmode.h (mode_traits): New structure.
18671 (get_narrowest_mode): New function.
18672 (mode_iterator::start): Likewise.
18673 (mode_iterator::iterate_p): Likewise.
18674 (mode_iterator::get_wider): Likewise.
18675 (mode_iterator::get_known_wider): Likewise.
18676 (mode_iterator::get_2xwider): Likewise.
18677 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
18678 (FOR_EACH_MODE): Likewise.
18679 (FOR_EACH_MODE_FROM): Likewise.
18680 (FOR_EACH_MODE_UNTIL): Likewise.
18681 (FOR_EACH_WIDER_MODE): Likewise.
18682 (FOR_EACH_2XWIDER_MODE): Likewise.
18683 * builtins.c (expand_builtin_strlen): Use new mode iterators.
18684 * combine.c (simplify_comparison): Likewise
18685 * config/i386/i386.c (type_natural_mode): Likewise.
18686 * cse.c (cse_insn): Likewise.
18687 * dse.c (find_shift_sequence): Likewise.
18688 * emit-rtl.c (init_derived_machine_modes): Likewise.
18689 (init_emit_once): Likewise.
18690 * explow.c (hard_function_value): Likewise.
18691 * expmed.c (extract_fixed_bit_field_1): Likewise.
18692 (extract_bit_field_1): Likewise.
18693 (expand_divmod): Likewise.
18694 (emit_store_flag_1): Likewise.
18695 * expr.c (init_expr_target): Likewise.
18696 (convert_move): Likewise.
18697 (alignment_for_piecewise_move): Likewise.
18698 (widest_int_mode_for_size): Likewise.
18699 (emit_block_move_via_movmem): Likewise.
18700 (copy_blkmode_to_reg): Likewise.
18701 (set_storage_via_setmem): Likewise.
18702 (compress_float_constant): Likewise.
18703 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18704 * optabs-query.c (get_best_extraction_insn): Likewise.
18705 * optabs.c (expand_binop): Likewise.
18706 (expand_twoval_unop): Likewise.
18707 (expand_twoval_binop): Likewise.
18708 (widen_leading): Likewise.
18709 (widen_bswap): Likewise.
18710 (expand_parity): Likewise.
18711 (expand_unop): Likewise.
18712 (prepare_cmp_insn): Likewise.
18713 (prepare_float_lib_cmp): Likewise.
18714 (expand_float): Likewise.
18715 (expand_fix): Likewise.
18716 (expand_sfix_optab): Likewise.
18717 * postreload.c (move2add_use_add2_insn): Likewise.
18718 * reg-stack.c (reg_to_stack): Likewise.
18719 * reginfo.c (choose_hard_reg_mode): Likewise.
18720 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18721 * stor-layout.c (mode_for_size): Likewise.
18722 (smallest_mode_for_size): Likewise.
18723 (mode_for_vector): Likewise.
18724 (finish_bitfield_representative): Likewise.
18725 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
18726 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18727 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18728 * var-tracking.c (prepare_call_arguments): Likewise.
18730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18731 Alan Hayward <alan.hayward@arm.com>
18732 David Sherwood <david.sherwood@arm.com>
18734 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
18735 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
18736 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
18737 * machmode.h (mode_size): Move earlier in file.
18738 (mode_precision): Likewise.
18739 (mode_inner): Likewise.
18740 (mode_nunits): Likewise.
18741 (mode_unit_size): Likewise.
18742 (unit_unit_precision): Likewise.
18743 (mode_wider): Likewise.
18744 (mode_2xwider): Likewise.
18745 (machine_mode): New class.
18746 (mode_to_bytes): New function.
18747 (mode_to_bits): Likewise.
18748 (mode_to_precision): Likewise.
18749 (mode_to_inner): Likewise.
18750 (mode_to_unit_size): Likewise.
18751 (mode_to_unit_precision): Likewise.
18752 (mode_to_nunits): Likewise.
18753 (GET_MODE_SIZE): Use mode_to_bytes.
18754 (GET_MODE_BITSIZE): Use mode_to_bits.
18755 (GET_MODE_PRECISION): Use mode_to_precision.
18756 (GET_MODE_INNER): Use mode_to_inner.
18757 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
18758 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
18759 (GET_MODE_NUNITS): Use mode_to_nunits.
18760 * system.h (ALWAYS_INLINE): New macro.
18761 * config/powerpcspe/powerpcspe-c.c
18762 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
18763 int for arg1_mode and arg2_mode.
18765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18766 Alan Hayward <alan.hayward@arm.com>
18767 David Sherwood <david.sherwood@arm.com>
18769 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18770 Prefix mode names with E_ in case statements.
18771 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18772 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18773 (aarch64_split_simd_move): Likewise.
18774 (aarch64_gen_storewb_pair): Likewise.
18775 (aarch64_gen_loadwb_pair): Likewise.
18776 (aarch64_gen_store_pair): Likewise.
18777 (aarch64_gen_load_pair): Likewise.
18778 (aarch64_get_condition_code_1): Likewise.
18779 (aarch64_constant_pool_reload_icode): Likewise.
18780 (get_rsqrte_type): Likewise.
18781 (get_rsqrts_type): Likewise.
18782 (get_recpe_type): Likewise.
18783 (get_recps_type): Likewise.
18784 (aarch64_gimplify_va_arg_expr): Likewise.
18785 (aarch64_simd_container_mode): Likewise.
18786 (aarch64_emit_load_exclusive): Likewise.
18787 (aarch64_emit_store_exclusive): Likewise.
18788 (aarch64_expand_compare_and_swap): Likewise.
18789 (aarch64_gen_atomic_cas): Likewise.
18790 (aarch64_emit_bic): Likewise.
18791 (aarch64_emit_atomic_swap): Likewise.
18792 (aarch64_emit_atomic_load_op): Likewise.
18793 (aarch64_evpc_trn): Likewise.
18794 (aarch64_evpc_uzp): Likewise.
18795 (aarch64_evpc_zip): Likewise.
18796 (aarch64_evpc_ext): Likewise.
18797 (aarch64_evpc_rev): Likewise.
18798 (aarch64_evpc_dup): Likewise.
18799 (aarch64_gen_ccmp_first): Likewise.
18800 (aarch64_gen_ccmp_next): Likewise.
18801 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18802 (alpha_emit_xfloating_libcall): Likewise.
18803 (emit_insxl): Likewise.
18804 (alpha_arg_type): Likewise.
18805 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18806 (arc_preferred_simd_mode): Likewise.
18807 (arc_secondary_reload): Likewise.
18808 (get_arc_condition_code): Likewise.
18809 (arc_print_operand): Likewise.
18810 (arc_legitimate_constant_p): Likewise.
18811 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18812 * config/arc/arc.md (casesi_load): Likewise.
18813 (casesi_compact_jump): Likewise.
18814 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18815 (cc_use_register): Likewise.
18816 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18817 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18818 (arm_init_iwmmxt_builtins): Likewise.
18819 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18820 (neon_expand_vector_init): Likewise.
18821 (arm_attr_length_move_neon): Likewise.
18822 (maybe_get_arm_condition_code): Likewise.
18823 (arm_emit_vector_const): Likewise.
18824 (arm_preferred_simd_mode): Likewise.
18825 (arm_output_iwmmxt_tinsr): Likewise.
18826 (thumb1_output_casesi): Likewise.
18827 (thumb2_output_casesi): Likewise.
18828 (arm_emit_load_exclusive): Likewise.
18829 (arm_emit_store_exclusive): Likewise.
18830 (arm_expand_compare_and_swap): Likewise.
18831 (arm_evpc_neon_vuzp): Likewise.
18832 (arm_evpc_neon_vzip): Likewise.
18833 (arm_evpc_neon_vrev): Likewise.
18834 (arm_evpc_neon_vtrn): Likewise.
18835 (arm_evpc_neon_vext): Likewise.
18836 (arm_validize_comparison): Likewise.
18837 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18838 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18839 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18840 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18841 (c6x_preferred_simd_mode): Likewise.
18842 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18843 (epiphany_rtx_costs): Likewise.
18844 * config/epiphany/predicates.md (proper_comparison_operator):
18846 * config/frv/frv.c (condexec_memory_operand): Likewise.
18847 (frv_emit_move): Likewise.
18848 (output_move_single): Likewise.
18849 (output_condmove_single): Likewise.
18850 (frv_hard_regno_mode_ok): Likewise.
18851 (frv_matching_accg_mode): Likewise.
18852 * config/h8300/h8300.c (split_adds_subs): Likewise.
18853 (h8300_rtx_costs): Likewise.
18854 (h8300_print_operand): Likewise.
18855 (compute_mov_length): Likewise.
18856 (output_logical_op): Likewise.
18857 (compute_logical_op_length): Likewise.
18858 (compute_logical_op_cc): Likewise.
18859 (h8300_shift_needs_scratch_p): Likewise.
18860 (output_a_shift): Likewise.
18861 (compute_a_shift_length): Likewise.
18862 (compute_a_shift_cc): Likewise.
18863 (expand_a_rotate): Likewise.
18864 (output_a_rotate): Likewise.
18865 * config/i386/i386.c (classify_argument): Likewise.
18866 (function_arg_advance_32): Likewise.
18867 (function_arg_32): Likewise.
18868 (function_arg_64): Likewise.
18869 (function_value_64): Likewise.
18870 (ix86_gimplify_va_arg): Likewise.
18871 (ix86_legitimate_constant_p): Likewise.
18872 (put_condition_code): Likewise.
18873 (split_double_mode): Likewise.
18874 (ix86_avx256_split_vector_move_misalign): Likewise.
18875 (ix86_expand_vector_logical_operator): Likewise.
18876 (ix86_split_idivmod): Likewise.
18877 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18878 (ix86_build_const_vector): Likewise.
18879 (ix86_build_signbit_mask): Likewise.
18880 (ix86_match_ccmode): Likewise.
18881 (ix86_cc_modes_compatible): Likewise.
18882 (ix86_expand_branch): Likewise.
18883 (ix86_expand_sse_cmp): Likewise.
18884 (ix86_expand_sse_movcc): Likewise.
18885 (ix86_expand_int_sse_cmp): Likewise.
18886 (ix86_expand_vec_perm_vpermi2): Likewise.
18887 (ix86_expand_vec_perm): Likewise.
18888 (ix86_expand_sse_unpack): Likewise.
18889 (ix86_expand_int_addcc): Likewise.
18890 (ix86_split_to_parts): Likewise.
18891 (ix86_vectorize_builtin_gather): Likewise.
18892 (ix86_vectorize_builtin_scatter): Likewise.
18893 (avx_vpermilp_parallel): Likewise.
18894 (inline_memory_move_cost): Likewise.
18895 (ix86_tieable_integer_mode_p): Likewise.
18896 (x86_maybe_negate_const_int): Likewise.
18897 (ix86_expand_vector_init_duplicate): Likewise.
18898 (ix86_expand_vector_init_one_nonzero): Likewise.
18899 (ix86_expand_vector_init_one_var): Likewise.
18900 (ix86_expand_vector_init_concat): Likewise.
18901 (ix86_expand_vector_init_interleave): Likewise.
18902 (ix86_expand_vector_init_general): Likewise.
18903 (ix86_expand_vector_set): Likewise.
18904 (ix86_expand_vector_extract): Likewise.
18905 (emit_reduc_half): Likewise.
18906 (ix86_emit_i387_round): Likewise.
18907 (ix86_mangle_type): Likewise.
18908 (ix86_expand_round_sse4): Likewise.
18909 (expand_vec_perm_blend): Likewise.
18910 (canonicalize_vector_int_perm): Likewise.
18911 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18912 (expand_vec_perm_1): Likewise.
18913 (expand_vec_perm_interleave3): Likewise.
18914 (expand_vec_perm_even_odd_pack): Likewise.
18915 (expand_vec_perm_even_odd_1): Likewise.
18916 (expand_vec_perm_broadcast_1): Likewise.
18917 (ix86_vectorize_vec_perm_const_ok): Likewise.
18918 (ix86_expand_vecop_qihi): Likewise.
18919 (ix86_expand_mul_widen_hilo): Likewise.
18920 (ix86_expand_sse2_abs): Likewise.
18921 (ix86_expand_pextr): Likewise.
18922 (ix86_expand_pinsr): Likewise.
18923 (ix86_preferred_simd_mode): Likewise.
18924 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18925 * config/i386/sse.md (*andnot<mode>3): Likewise.
18926 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18927 (*<code><mode>3): Likewise.
18928 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18929 (ia64_expand_atomic_op): Likewise.
18930 (ia64_arg_type): Likewise.
18931 (ia64_mode_to_int): Likewise.
18932 (ia64_scalar_mode_supported_p): Likewise.
18933 (ia64_vector_mode_supported_p): Likewise.
18934 (expand_vec_perm_broadcast): Likewise.
18935 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18936 (iq2000_function_arg_advance): Likewise.
18937 (iq2000_function_arg): Likewise.
18938 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18939 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18940 (m68k_libcall_value): Likewise.
18941 (m68k_function_value): Likewise.
18942 (sched_attr_op_type): Likewise.
18943 * config/mcore/mcore.c (mcore_output_move): Likewise.
18944 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18946 (microblaze_function_arg): Likewise.
18947 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18948 (mips_print_operand): Likewise.
18949 (mips_mode_ok_for_mov_fmt_p): Likewise.
18950 (mips_vector_mode_supported_p): Likewise.
18951 (mips_preferred_simd_mode): Likewise.
18952 (mips_expand_vpc_loongson_even_odd): Likewise.
18953 (mips_expand_vec_unpack): Likewise.
18954 (mips_expand_vi_broadcast): Likewise.
18955 (mips_expand_vector_init): Likewise.
18956 (mips_expand_vec_reduc): Likewise.
18957 (mips_expand_msa_cmp): Likewise.
18958 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18959 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18960 (cc_flags_for_mode): Likewise.
18961 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18962 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18963 (nds32_output_casesi_pc_relative): Likewise.
18964 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18965 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18966 (nvptx_gen_unpack): Likewise.
18967 (nvptx_gen_pack): Likewise.
18968 (nvptx_gen_shuffle): Likewise.
18969 (nvptx_gen_wcast): Likewise.
18970 (nvptx_preferred_simd_mode): Likewise.
18971 * config/pa/pa.c (pa_secondary_reload): Likewise.
18972 * config/pa/predicates.md (base14_operand): Likewise.
18973 * config/powerpcspe/powerpcspe-c.c
18974 (altivec_resolve_overloaded_builtin): Likewise.
18975 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18977 (rs6000_preferred_simd_mode): Likewise.
18978 (output_vec_const_move): Likewise.
18979 (rs6000_expand_vector_extract): Likewise.
18980 (rs6000_split_vec_extract_var): Likewise.
18981 (reg_offset_addressing_ok_p): Likewise.
18982 (rs6000_legitimate_offset_address_p): Likewise.
18983 (rs6000_legitimize_address): Likewise.
18984 (rs6000_emit_set_const): Likewise.
18985 (rs6000_const_vec): Likewise.
18986 (rs6000_emit_move): Likewise.
18987 (spe_build_register_parallel): Likewise.
18988 (rs6000_darwin64_record_arg_recurse): Likewise.
18989 (swap_selector_for_mode): Likewise.
18990 (spe_init_builtins): Likewise.
18991 (paired_init_builtins): Likewise.
18992 (altivec_init_builtins): Likewise.
18993 (do_load_for_compare): Likewise.
18994 (rs6000_generate_compare): Likewise.
18995 (rs6000_expand_float128_convert): Likewise.
18996 (emit_load_locked): Likewise.
18997 (emit_store_conditional): Likewise.
18998 (rs6000_output_function_epilogue): Likewise.
18999 (rs6000_handle_altivec_attribute): Likewise.
19000 (rs6000_function_value): Likewise.
19001 (emit_fusion_gpr_load): Likewise.
19002 (emit_fusion_p9_load): Likewise.
19003 (emit_fusion_p9_store): Likewise.
19004 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19005 (fusion_gpr_mem_load): Likewise.
19006 (fusion_addis_mem_combo_load): Likewise.
19007 (fusion_addis_mem_combo_store): Likewise.
19008 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19009 (fusion_gpr_mem_load): Likewise.
19010 (fusion_addis_mem_combo_load): Likewise.
19011 (fusion_addis_mem_combo_store): Likewise.
19012 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19014 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19015 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19016 (rs6000_preferred_simd_mode): Likewise.
19017 (output_vec_const_move): Likewise.
19018 (rs6000_expand_vector_extract): Likewise.
19019 (rs6000_split_vec_extract_var): Likewise.
19020 (reg_offset_addressing_ok_p): Likewise.
19021 (rs6000_legitimate_offset_address_p): Likewise.
19022 (rs6000_legitimize_address): Likewise.
19023 (rs6000_emit_set_const): Likewise.
19024 (rs6000_const_vec): Likewise.
19025 (rs6000_emit_move): Likewise.
19026 (rs6000_darwin64_record_arg_recurse): Likewise.
19027 (swap_selector_for_mode): Likewise.
19028 (paired_init_builtins): Likewise.
19029 (altivec_init_builtins): Likewise.
19030 (rs6000_expand_float128_convert): Likewise.
19031 (emit_load_locked): Likewise.
19032 (emit_store_conditional): Likewise.
19033 (rs6000_output_function_epilogue): Likewise.
19034 (rs6000_handle_altivec_attribute): Likewise.
19035 (rs6000_function_value): Likewise.
19036 (emit_fusion_gpr_load): Likewise.
19037 (emit_fusion_p9_load): Likewise.
19038 (emit_fusion_p9_store): Likewise.
19039 * config/rx/rx.c (rx_gen_move_template): Likewise.
19040 (flags_from_mode): Likewise.
19041 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19042 (s390_slb_comparison): Likewise.
19043 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19044 (s390_vector_mode_supported_p): Likewise.
19045 (s390_cc_modes_compatible): Likewise.
19046 (s390_match_ccmode_set): Likewise.
19047 (s390_canonicalize_comparison): Likewise.
19048 (s390_emit_compare_and_swap): Likewise.
19049 (s390_branch_condition_mask): Likewise.
19050 (s390_rtx_costs): Likewise.
19051 (s390_secondary_reload): Likewise.
19052 (__SECONDARY_RELOAD_CASE): Likewise.
19053 (s390_expand_cs): Likewise.
19054 (s390_preferred_simd_mode): Likewise.
19055 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19056 * config/sh/sh.c (sh_print_operand): Likewise.
19057 (dump_table): Likewise.
19058 (sh_secondary_reload): Likewise.
19059 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19060 * config/sh/sh.md (casesi_worker_1): Likewise.
19061 (casesi_worker_2): Likewise.
19062 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19063 (fcc_comparison_operator): Likewise.
19064 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19065 (emit_soft_tfmode_cvt): Likewise.
19066 (sparc_preferred_simd_mode): Likewise.
19067 (output_cbranch): Likewise.
19068 (sparc_print_operand): Likewise.
19069 (sparc_expand_vec_perm_bmask): Likewise.
19070 (vector_init_bshuffle): Likewise.
19071 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19072 (spu_vector_mode_supported_p): Likewise.
19073 (spu_expand_insv): Likewise.
19074 (spu_emit_branch_or_set): Likewise.
19075 (spu_handle_vector_attribute): Likewise.
19076 (spu_builtin_splats): Likewise.
19077 (spu_builtin_extract): Likewise.
19078 (spu_builtin_promote): Likewise.
19079 (spu_expand_sign_extend): Likewise.
19080 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19081 (tilegx_simd_int): Likewise.
19082 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19083 (tilepro_simd_int): Likewise.
19084 * config/v850/v850.c (const_double_split): Likewise.
19085 (v850_print_operand): Likewise.
19086 (ep_memory_offset): Likewise.
19087 * config/vax/vax.c (vax_rtx_costs): Likewise.
19088 (vax_output_int_move): Likewise.
19089 (vax_output_int_add): Likewise.
19090 (vax_output_int_subtract): Likewise.
19091 * config/visium/predicates.md (visium_branch_operator): Likewise.
19092 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19093 (visium_print_operand_address): Likewise.
19094 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19095 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19096 (xtensa_expand_conditional_branch): Likewise.
19097 (xtensa_copy_incoming_a7): Likewise.
19098 (xtensa_output_literal): Likewise.
19099 * dfp.c (decimal_real_maxval): Likewise.
19100 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19103 Alan Hayward <alan.hayward@arm.com>
19104 David Sherwood <david.sherwood@arm.com>
19106 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19107 (mode_nunits_inline): Likewise.
19108 (mode_inner_inline): Likewise.
19109 (mode_unit_size_inline): Likewise.
19110 (mode_unit_precision_inline): Likewise.
19111 (emit_insn_modes_h): Likewise. Also emit a #define of the
19113 (emit_mode_wider): Add an E_ prefix to mode names.
19114 (emit_mode_complex): Likewise.
19115 (emit_mode_inner): Likewise.
19116 (emit_mode_adjustments): Likewise.
19117 (emit_mode_int_n): Likewise.
19118 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19119 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19120 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19121 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19122 (CRC32_BUILTIN, ENTRY): Likewise.
19123 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19124 (aarch64_pop_regs): Likewise.
19125 (aarch64_process_components): Likewise.
19126 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19127 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19128 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19129 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19130 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19131 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19133 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19134 (pcmpistr): Likewise.
19135 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19136 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19137 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19139 * config/rl78/rl78.c (mduc_regs): Likewise.
19140 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19141 (htm_expand_builtin): Likewise.
19142 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19143 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19144 * config/xtensa/xtensa.c (print_operand): Likewise.
19145 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19146 (NUM_MODE_VECTOR_INT): Likewise.
19147 * genoutput.c (null_operand): Likewise.
19148 (output_operand_data): Likewise.
19149 * genrecog.c (print_parameter_value): Likewise.
19150 * lra.c (debug_operand_data): Likewise.
19152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19154 * dwarf2out.c (scompare_loc_descriptor_wide)
19155 (scompare_loc_descriptor_narrow): New functions, split out from...
19156 (scompare_loc_descriptor): ...here.
19157 * expmed.c (emit_store_flag_int): New function, split out from...
19158 (emit_store_flag): ...here.
19160 2017-08-30 Richard Biener <rguenther@suse.de>
19162 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19163 (dwarf2out_early_finish): Move setting of AT_pubnames from
19164 early debug output to early finish.
19166 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19168 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19169 and -mdata-region to the assembler.
19171 2017-08-30 Richard Biener <rguenther@suse.de>
19173 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19175 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19176 (dwarf2out_early_global_decl): Only generate a DIE for the
19177 abstract origin if it doesn't already exist or is a declaration DIE.
19178 (resolve_addr): Do not add the linkage name twice when
19179 generating a stub DIE for the DW_TAG_GNU_call_site target.
19181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19183 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19184 Use machine_mode rather than int for arg1_mode.
19186 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19189 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19190 that the second argument of the built-in functions to unpack
19191 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19192 switch statement instead a lot of if statements.
19193 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19194 Allow 64-bit values to be in Altivec registers as well as
19195 traditional floating point registers.
19196 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19198 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19200 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19201 MAX_REGS_PER_ADDRESS == 1.
19203 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19205 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19206 * config/i386/i386.c (ix86_option_override_internal): Simplify
19207 setting of opts->x_flag_entry.
19209 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19210 Jakub Jelinek <jakub@redhat.com>
19211 Richard Biener <rguenther@suse.de>
19213 PR tree-optimization/81503
19214 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19215 folded constant fits in the target type; reorder tests for clarity.
19217 2017-08-29 Martin Liska <mliska@suse.cz>
19219 * passes.def: Include pass_lower_switch.
19220 * stmt.c (dump_case_nodes): Remove and move to
19221 tree-switch-conversion.
19222 (case_values_threshold): Likewise.
19223 (expand_switch_as_decision_tree_p): Likewise.
19224 (emit_case_decision_tree): Likewise.
19225 (expand_case): Likewise.
19226 (balance_case_nodes): Likewise.
19227 (node_has_low_bound): Likewise.
19228 (node_has_high_bound): Likewise.
19229 (node_is_bounded): Likewise.
19230 (emit_case_nodes): Likewise.
19231 (struct simple_case_node): New struct.
19232 (add_case_node): Remove.
19233 (emit_case_dispatch_table): Use vector instead of case_list.
19234 (reset_out_edges_aux): Remove.
19235 (compute_cases_per_edge): Likewise.
19236 (expand_case): Build list of simple_case_node.
19237 (expand_sjlj_dispatch_table): Use it.
19238 * tree-switch-conversion.c (struct case_node): Moved from
19239 stmt.c and adjusted.
19240 (emit_case_nodes): Likewise.
19241 (node_has_low_bound): Likewise.
19242 (node_has_high_bound): Likewise.
19243 (node_is_bounded): Likewise.
19244 (case_values_threshold): Likewise.
19245 (reset_out_edges_aux): Likewise.
19246 (compute_cases_per_edge): Likewise.
19247 (add_case_node): Likewise.
19248 (dump_case_nodes): Likewise.
19249 (balance_case_nodes): Likewise.
19250 (expand_switch_as_decision_tree_p): Likewise.
19251 (emit_jump): Likewise.
19252 (emit_case_decision_tree): Likewise.
19253 (try_switch_expansion): Likewise.
19254 (do_jump_if_equal): Likewise.
19255 (emit_cmp_and_jump_insns): Likewise.
19256 (fix_phi_operands_for_edge): New function.
19257 (record_phi_operand_mapping): Likewise.
19258 (class pass_lower_switch): New pass.
19259 (pass_lower_switch::execute): New function.
19260 (make_pass_lower_switch): Likewise.
19261 (conditional_probability):
19262 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19263 * tree-pass.h: Add make_pass_lower_switch.
19265 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19268 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19271 2017-08-29 Richard Biener <rguenther@suse.de>
19273 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19274 we do not add a DW_AT_inline attribute twice.
19275 (gen_subprogram_die): Remove code setting DW_AT_inline on
19276 DECL_ABSTRACT_P nodes.
19278 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19280 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19281 calls to internal functions.
19282 (gimplify_modify_expr): Likewise.
19283 * tree-call-cdce.c (use_internal_fn): Likewise.
19284 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19285 (convert_to_divmod): Set the nothrow flag.
19286 * tree-if-conv.c (predicate_mem_writes): Likewise.
19287 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19288 (vectorizable_call): Likewise.
19289 (vectorizable_store): Likewise.
19290 (vectorizable_load): Likewise.
19291 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19292 (vect_recog_mask_conversion_pattern): Likewise.
19294 2017-08-29 Martin Liska <mliska@suse.cz>
19297 * gcc.c (driver_handle_option): Add new argument.
19298 * opts-common.c (handle_option): Pass
19299 target_option_override_hook.
19300 * opts-global.c (lang_handle_option): Add new option.
19301 (set_default_handlers): Add new argument.
19302 (decode_options): Likewise.
19303 * opts.c (target_handle_option): Likewise.
19304 (common_handle_option): Call target_option_override_hook.
19305 * opts.h (struct cl_option_handler_func): Add hook for
19306 target option override.
19307 (struct cl_option_handlers): Likewise.
19308 (set_default_handlers): Add new argument.
19309 (decode_options): Likewise.
19310 (common_handle_option): Likewise.
19311 (target_handle_option): Likewise.
19312 * toplev.c (toplev::main): Pass targetm.target_option.override
19315 2017-08-29 Richard Biener <rguenther@suse.de>
19316 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19318 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19319 life to the active subtree.
19321 2017-08-28 Jeff Law <law@redhat.com>
19323 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19324 derive_equivalences.
19325 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19326 Code moved into....
19327 (edge_info::derive_equivalences): New private member function
19329 * tree-ssa-dom.c (class edge_info): Changed from a struct
19330 to a class. Add ctor/dtor, methods and data members.
19331 (edge_info::edge_info): Renamed from allocate_edge_info.
19332 Initialize additional members.
19333 (edge_info::~edge_info): New.
19334 (free_dom_edge_info): Delete the edge info.
19335 (record_edge_info): Use new class & associated member functions.
19336 Tighten forms for testing for edge equivalences.
19337 (record_temporary_equivalences): Iterate over the simple
19338 equivalences rather than assuming there's only one per edge.
19339 (cprop_into_successor_phis): Iterate over the simple
19340 equivalences rather than assuming there's only one per edge.
19341 (optimize_stmt): Use operand_equal_p rather than pointer
19342 equality for mini-DSE code.
19344 2017-08-28 Nathan Sidwell <nathan@acm.org>
19346 * gcc.c (execute): Fold SIGPIPE handling into switch
19347 statement. Adjust internal error message.
19349 2017-08-28 Richard Biener <rguenther@suse.de>
19352 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19353 Do nothing for removed DIEs.
19355 2017-08-28 Richard Biener <rguenther@suse.de>
19357 PR tree-optimization/81977
19358 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19361 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19364 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19365 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19366 using targetm.gen_mem_thread_fence.
19368 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19371 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19372 predicate to register_operand. Reorder operands.
19373 (*btr<mode>): Ditto.
19374 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19375 (*btr<mode>_mask): Ditto.
19377 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19379 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19380 * config/rs6000/xmmintrin.h: New file.
19381 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19383 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19386 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19387 parameter and_insn and return it.
19388 (recombine_lvx_pattern): Insert a copy to ensure availability of
19389 the base register of the copied masking operation at the point of
19390 the instruction replacement.
19391 (recombine_stvx_pattern): Likewise.
19393 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19395 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19396 undocumented switches.
19397 (-mpower9-dform-vector): Likewise.
19398 (-mpower9-dform): Likewise.
19399 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19400 comments to delete references to -mpower9-dform* switches.
19401 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19402 Delete reference to -mpower9-dform* switches, test for
19403 -mpower9-vector instead.
19404 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19405 (OTHER_P9_VECTOR_MASKS): Likewise.
19406 (POWERPC_MASKS): Likewise.
19407 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19408 tests against -mpower9-dform* to -mpower9-vector. Delete code
19409 that checked for -mpower9-dform* consistancy with other options.
19410 Add test for -mpower9-misc to enable other power9 switches.
19411 (rs6000_init_hard_regno_mode_ok): Likewise.
19412 (rs6000_option_override_internal): Likewise.
19413 (rs6000_emit_prologue): Likewise.
19414 (rs6000_emit_epilogue): Likewise.
19415 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19416 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19417 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19419 (emit_fusion_p9_store): Likewise.
19420 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19421 resetting these macros if the assembler does not support ISA 3.0
19423 (TARGET_P9_DFORM_VECTOR): Likewise.
19424 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19425 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19427 2017-08-25 Alan Modra <amodra@gmail.com>
19429 PR rtl-optimization/81747
19430 * cse.c (cse_extended_basic_block): Don't attempt to record
19431 equivalences for degenerate conditional branches that branch
19432 to their fall-through.
19434 2017-08-24 Martin Sebor <msebor@redhat.com>
19436 PR middle-end/81908
19437 * gimple-fold.c (size_must_be_zero_p): New function.
19438 (gimple_fold_builtin_memory_op): Call it.
19440 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
19442 * config/rs6000/mm_malloc.h: New file.
19444 2017-08-24 Bin Cheng <bin.cheng@arm.com>
19446 PR tree-optimization/81913
19447 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
19448 analysis when either IVs in condition can wrap.
19450 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
19452 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
19453 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
19455 2017-08-24 Richard Biener <rguenther@suse.de>
19458 * targhooks.c (default_target_can_inline_p): Properly
19459 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
19460 is present and always compare.
19461 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
19462 infer -mfpmath=sse from TARGET_SSE_P.
19463 (ix86_can_inline_p): Properly use target_option_default_node when
19464 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
19466 2017-08-24 Richard Biener <rguenther@suse.de>
19469 * dwarf2out.c (output_die): Handle flag_generate_offload like
19471 (output_comp_unit): Likewise.
19472 (gen_array_type_die): Likewise.
19473 (dwarf2out_early_finish): Likewise.
19474 (note_variable_value_in_expr): Likewise.
19475 (dwarf2out_finish): Likewise. Adjust assert.
19476 * cgraphunit.c (symbol_table::compile): Move setting of
19477 flag_generate_offload earlier ...
19478 (symbol_table::finalize_compilation_unit): ... here, before
19479 early debug finalization.
19481 2017-08-24 Richard Biener <rguenther@suse.de>
19483 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
19484 and ipa-fnsummary.h.
19485 (ix86_can_inline_p): When ix86_fpmath flags do not match
19486 check whether the callee uses FP math at all.
19488 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
19490 PR middle-end/81931
19491 * tree-ssanames.c (get_nonzero_bits): Use element_precision
19492 instead of TYPE_PRECISION.
19494 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
19495 Alan Hayward <alan.hayward@arm.com>
19496 David Sherwood <david.sherwood@arm.com>
19498 * combine.c (make_extraction): Use subreg_offset_from_lsb.
19500 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
19502 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
19504 (ix86_frame::stack_realign_allocate): New field.
19505 (struct machine_frame_state): Modify comments.
19506 (machine_frame_state::sp_realigned_fp_end): New field.
19507 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
19508 layout calculation.
19509 (sp_valid_at): Add assertion to assure no attempt to access invalid
19510 offset of a realigned stack.
19511 (fp_valid_at): Likewise.
19512 (choose_baseaddr): Modify comments.
19513 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
19514 ix86_expand_prologue.
19515 (ix86_expand_prologue): Modify stack realignment and allocation.
19516 (ix86_expand_epilogue): Modify comments.
19517 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
19518 avx2_runtime, avx512f, and avx512f_runtime.
19520 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
19522 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
19523 (mstackrealign): Do not init to -1.
19524 * config/i386/i386.c (ix86_option_override_internal):
19525 Check opts_set, not opts when setting default value of
19526 opts->x_ix86_force_align_arg_pointer.
19528 2017-08-23 Richard Biener <rguenther@suse.de>
19530 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
19531 lang_hooks.decl_printable_name.
19532 * print-rtl-function.c (print_rtx_function): Likewise.
19533 * tree-pretty-print.c (dump_function_header): Likewise.
19535 2017-08-23 Richard Biener <rguenther@suse.de>
19538 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
19539 -g0 at compile-time.
19541 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19543 PR middle-end/19706
19544 * doc/sourcebuild.texi (Other hardware attributes):
19547 2017-08-23 Tamar Christina <tamar.christina@arm.com>
19549 PR middle-end/19706
19550 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
19551 Fix single-use check.
19553 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
19555 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
19557 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19559 * doc/install.texi: Modify to add more details on running selected
19562 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
19564 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
19565 is combined with -mabi=ms.
19566 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
19569 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19571 PR tree-optimization/81488
19572 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
19573 and cached_basis fields.
19574 (MAX_SPREAD): New constant.
19575 (alloc_cand_and_find_basis): Initialize new fields.
19576 (clear_visited): New function.
19577 (create_phi_basis_1): Rename from create_phi_basis, set visited
19578 and cached_basis fields.
19579 (create_phi_basis): New wrapper function.
19580 (phi_add_costs_1): Rename from phi_add_costs, add spread
19581 parameter, set visited field, short-circuit when limits reached.
19582 (phi_add_costs): New wrapper function.
19583 (record_phi_increments_1): Rename from record_phi_increments, set
19585 (record_phi_increments): New wrapper function.
19586 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
19587 (phi_incr_cost): New wrapper function.
19588 (all_phi_incrs_profitable_1): Rename from
19589 all_phi_incrs_profitable, set visited field.
19590 (all_phi_incrs_profitable): New wrapper function.
19592 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19593 Alan Hayward <alan.hayward@arm.com>
19594 David Sherwood <david.sherwood@arm.com>
19596 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
19597 that takes the outer and inner modes.
19598 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
19599 comparison as the canonical test for a paradoxical subreg.
19600 * combine.c (simplify_set): Use paradoxical_subreg_p.
19601 (make_extraction): Likewise.
19602 (force_to_mode): Likewise.
19603 (rtx_equal_for_field_assignment_p): Likewise.
19604 (gen_lowpart_for_combine): Likewise.
19605 (simplify_comparison): Likewise.
19606 * cse.c (equiv_constant): Likewise.
19607 * expmed.c (store_bit_field_1): Likewise.
19608 * final.c (alter_subreg): Likewise.
19609 * fwprop.c (propagate_rtx): Likewise.
19610 (forward_propagate_subreg): Likewise.
19611 * ira-conflicts.c (ira_build_conflicts): Likewise.
19612 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
19613 * lra-constraints.c (curr_insn_transform): Likewise.
19614 (split_reg): Likewise.
19615 * lra-eliminations.c (move_plus_up): Likewise.
19616 (lra_eliminate_regs_1): Likewise.
19617 * recog.c (general_operand): Likewise.
19618 * ree.c (combine_reaching_defs): Likewise.
19619 * reload.c (push_reload): Likewise.
19620 (find_reloads): Likewise.
19621 * reload1.c (elimination_effects): Likewise.
19622 (compute_reload_subreg_offset): Likewise.
19623 (choose_reload_regs): Likewise.
19624 * rtlanal.c (subreg_lsb_1): Likewise.
19625 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19626 (simplify_subreg): Likewise.
19627 * var-tracking.c (track_loc_p): Likewise.
19628 * emit-rtl.c (byte_lowpart_offset): Likewise.
19629 (paradoxical_subreg_p): Delete out-of-line definition.
19631 2017-08-22 Jeff Law <law@redhat.com>
19633 PR tree-optimization/81741
19634 PR tree-optimization/71947
19635 * tree-ssa-dom.c: Include tree-inline.h.
19636 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
19637 equivalences if one is more expensive to compute than the other.
19638 * tree-ssa-scopedtables.h (class const_or_copies): Make
19639 record_const_or_copy_raw method private.
19640 (class avail_exprs_stack): New method simplify_binary_operation.
19641 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
19642 avail_exprs_stack::simplify_binary_operation as needed.
19643 (avail_exprs_stack::simplify_binary_operation): New function.
19645 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
19647 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
19648 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
19649 (DOT_SYMBOLS): Likewise.
19650 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
19651 (RELOCATABLE_NEEDS_FIXUP): Likewise.
19652 (RS6000_ABI_NAME): Likewise.
19653 (TARGET_CMODEL): Likewise.
19654 (TOC_SECTION_ASM_OP): Likewise.
19655 (SET_CMODEL): New macro.
19656 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
19658 2017-08-22 Richard Biener <rguenther@suse.de>
19660 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
19661 to allow for free-lang-data replacements similar to verify_type_variant.
19663 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
19664 Alan Hayward <alan.hayward@arm.com>
19665 David Sherwood <david.sherwood@arm.com>
19667 * config/aarch64/aarch64.md (casesi): Use DImode rather than
19668 VOIDmode for the LABEL_REF.
19670 2017-08-22 Richard Biener <rguenther@suse.de>
19672 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
19674 2017-08-22 Richard Biener <rguenther@suse.de>
19676 * common.opt (feliminate-dwarf2-dups): Ignore.
19677 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
19678 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
19679 same_die_p_wrap, compute_section_prefix,
19680 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
19681 (comdat_symbol_id, comdat_symbol_number): Likewise.
19682 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
19684 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
19685 (output_die): Mark unreachable path unreachable.
19686 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
19687 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
19688 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
19689 (dwarf2out_early_finish): Likewise.
19691 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
19693 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
19695 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
19698 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
19699 not VAR_P. Filter attribute warnings with OPT_Wattributes.
19700 (avr_attribute_table) <io, io_low, address>: Initialize
19701 .decl_required with true.
19703 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
19705 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
19706 undocumented debugging options.
19707 (-mvsx-scalar-double): Likewise.
19708 (-mallow-df-permute): Likewise.
19709 (-mvectorize-builtins): Likewise.
19710 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19711 (rs6000_builtin_vectorized_function): Likewise.
19712 (rs6000_builtin_md_vectorized_function): Likewise.
19713 (rs6000_opt_vars): Likewise.
19715 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
19718 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
19719 (*btrq_imm): Rename from *btrq.
19720 (*btcq_imm): Rename from *btcq.
19721 (btsc): New code attribute.
19722 (*<btsc><mode>): New insn pattern.
19723 (*btr<mode>): Ditto.
19724 (*<btsc><mode>_mask): New insn_and_split pattern.
19725 (*btr<mode>_mask): Ditto.
19727 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19728 Alan Hayward <alan.hayward@arm.com>
19729 David Sherwood <david.sherwood@arm.com>
19731 * function.c (pad_below): Simplify padding calculation.
19733 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19734 Alan Hayward <alan.hayward@arm.com>
19735 David Sherwood <david.sherwood@arm.com>
19737 * target.def (function_prologue): Remove frame size argument.
19738 (function_epilogue): Likewise.
19739 * doc/tm.texi: Regenerate.
19740 * final.c (final_start_function): Update call to function_prologue.
19741 (final_end_function): Update call to function_epilogue.
19742 (default_function_pro_epilogue): Remove frame size argument.
19743 * output.h (default_function_pro_epilogue): Likewise.
19744 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
19745 (arm_output_function_prologue): Likewise.
19746 * config/frv/frv.c (frv_function_prologue): Likewise.
19747 (frv_function_epilogue): Likewise.
19748 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
19749 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
19750 (ia64_output_function_epilogue): Likewise.
19751 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
19752 (m32r_output_function_epilogue): Likewise.
19753 * config/microblaze/microblaze.c (microblaze_function_prologue)
19754 (microblaze_function_epilogue): Likewise.
19755 * config/mips/mips.c (mips_output_function_prologue): Likewise.
19756 (mips_output_function_epilogue): Likewise.
19757 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
19758 (mmix_target_asm_function_epilogue): Likewise.
19759 * config/msp430/msp430.c (msp430_start_function): Likewise.
19760 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
19761 (nds32_asm_function_epilogue): Likewise.
19762 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
19763 * config/pa/pa.c (pa_output_function_prologue): Likewise.
19764 (pa_output_function_epilogue): Likewise.
19765 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
19766 (rs6000_output_function_epilogue): Likewise.
19767 * config/rl78/rl78.c (rl78_start_function): Likewise.
19768 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19769 (rs6000_output_function_epilogue): Likewise.
19770 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19771 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19772 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19773 (sparc_asm_function_epilogue): Likewise.
19775 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19777 * tree.h (type_has_mode_precision_p): New function.
19778 * convert.c (convert_to_integer_1): Use it.
19779 * expr.c (expand_expr_real_2): Likewise.
19780 (expand_expr_real_1): Likewise.
19781 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19782 * match.pd: Likewise.
19783 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19784 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19785 * tree-tailcall.c (process_assignment): Likewise.
19786 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19787 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19788 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19789 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19790 (vectorizable_assignment): Likewise.
19791 (vectorizable_shift): Likewise.
19792 (vectorizable_operation): Likewise.
19793 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19795 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19797 * match.pd: Add pow (C, x) simplification.
19799 2017-08-21 Richard Biener <rguenther@suse.de>
19801 PR tree-optimization/81900
19802 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19803 for blocks with abnormal predecessors.
19804 (compute_antic): Do not set visited flag prematurely.
19806 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19809 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19811 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19813 * stor-layout.h (vector_type_mode): Move to...
19814 * tree.h (vector_type_mode): ...here.
19815 * stor-layout.c (vector_type_mode): Move to...
19816 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19818 2017-08-21 Richard Biener <rguenther@suse.de>
19820 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19821 register_external_die hooks.
19822 (debug_false_tree_charstarstar_uhwistar): Declare.
19823 (debug_nothing_tree_charstar_uhwi): Likewise.
19824 * debug.c (do_nothing_debug_hooks): Adjust.
19825 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19826 (debug_nothing_tree_charstar_uhwi): Likewise.
19827 * dbxout.c (dbx_debug_hooks): Adjust.
19828 (xcoff_debug_hooks): Likewise.
19829 * sdbout.c (sdb_debug_hooks): Likewise.
19830 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19831 * dwarf2out.c (macinfo_label_base): New global.
19832 (dwarf2out_register_external_die): New function for the
19833 register_external_die hook.
19834 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19835 (dwarf2_debug_hooks): Use them.
19836 (dwarf2_lineno_debug_hooks): Adjust.
19837 (struct die_struct): Add with_offset flag.
19838 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19839 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19840 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19841 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19842 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19843 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19844 defining section names for the early LTO debug variants.
19845 (reset_indirect_string): New helper.
19846 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19847 (print_dw_val): Add support for offsetted symbol references.
19848 (get_ultimate_context): Split out from is_cxx.
19849 (is_cxx): Use get_ultimate_context.
19850 (is_fortran): Add decl overload.
19851 (compute_comp_unit_symbol): Split out worker from
19852 compute_section_prefix.
19853 (compute_section_prefix): Call compute_comp_unit_symbol and
19854 set comdat_type_p here.
19855 (output_die): Skip DIE symbol output for the LTO added one.
19856 Handle DIE symbol references with offset.
19857 (output_comp_unit): Guard section name mangling properly.
19858 For LTO debug sections emit a symbol at the section beginning
19859 which we use to refer to its DIEs.
19860 (add_abstract_origin_attribute): For DIEs registered via
19861 dwarf2out_register_external_die directly refer to the early
19862 DIE rather than indirectly through the shadow one we created.
19863 Remove obsolete call to dwarf2out_abstract_function for
19864 non-function/block origins.
19865 (gen_array_type_die): When generating early LTO debug do
19866 not emit DW_AT_string_length.
19867 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19868 late when in LTO. As suggested place a gcc_unreachable for
19869 the DECL_ABSTRACT_P case.
19870 (gen_subprogram_die): Avoid another specification DIE
19871 for early built declarations/definitions for the late LTO case.
19872 (gen_variable_die): Add type references for late duplicated VLA dies
19874 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19875 we have the abstract instance already.
19876 (process_scope_var): Adjust decl DIE contexts in LTO which
19877 first puts them in limbo.
19878 (gen_decl_die): Do not generate type DIEs late apart from
19879 types for VLAs or for decls we do not yet have a DIE. Do not
19880 call dwarf2out_abstract_function late.
19881 (dwarf2out_early_global_decl): Make sure to create DIEs
19882 for abstract instances of a decl first.
19883 (dwarf2out_late_global_decl): Adjust comment.
19884 (output_macinfo_op): With multiple macro sections use
19885 macinfo_label_base to distinguish labels.
19886 (output_macinfo): Likewise. Update macinfo_label_base.
19887 Pass in the line info label.
19888 (note_variable_value_in_expr): When generating LTO resolve
19889 all variable values here by generating DIEs as needed.
19890 (init_sections_and_labels): Add early LTO debug flag parameter
19891 and generate different sections and names if set. Add generation
19892 counter for the labels so we can have multiple of them.
19893 (reset_dies): Helper to allow DIEs to be output multiple times.
19894 (dwarf2out_finish): When outputting DIEs to the fat part of an
19895 LTO object first reset DIEs.
19896 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19897 (modified_type_die): Check for decl_ultimate_origin being self
19899 (gen_type_die_with_usage): Likewise.
19900 (gen_typedef_die): Allow decl_ultimate_origin being self.
19901 (set_decl_abstract_flags): Remove.
19902 (set_block_abstract_flags): Likewise.
19903 (dwarf2out_abstract_function): Treat the early generated DIEs
19904 as the abstract copy and only add DW_AT_inline and
19905 DW_AT_artificial here and call set_decl_origin_self.
19906 If the DIE has an abstract origin don't do anything.
19907 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19908 if we have none yet (Go fails to build one, PR78628).
19909 (variably_modified_type_p): Prevent endless recursion for Ada
19910 cyclic pointer types.
19911 * lto-streamer-in.c: Include debug.h.
19912 (dref_queue): New global.
19913 (lto_read_tree_1): Stream in DIE references.
19914 (lto_input_tree): Register DIE references.
19915 (input_function): Stream DECL_DEBUG_ARGS.
19916 * lto-streamer-out.c: Include debug.h.
19917 (lto_write_tree_1): Output DIE references.
19918 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19919 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19920 (output_function): Stream DECL_DEBUG_ARGS.
19921 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19922 Stream DECL_ABSTRACT_ORIGIN.
19923 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19924 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19925 DECL_CONTEXT for file-scope decls.
19926 * lto-streamer.h (struct dref_entry): Declare.
19927 (dref_queue): Likewise.
19928 * cfgexpand.c (pass_expand::execute): Do not call the
19929 outlining_inline_function hook here.
19930 * lto-wrapper.c (debug_obj): New global.
19931 (tool_cleanup): Unlink it if required.
19932 (debug_objcopy): New function.
19933 (run_gcc): Handle early debug sections in the IL files by
19934 extracting them to separate files, partially linkin them and
19935 feeding the result back as result to the linker.
19936 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19937 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19938 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19939 sections into a separate segment.
19940 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19942 (darwin_asm_dwarf_section): Likewise.
19943 (darwin_asm_output_dwarf_offset): Likewise.
19944 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19946 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19947 Alan Hayward <alan.hayward@arm.com>
19948 David Sherwood <david.sherwood@arm.com>
19950 * read-md.h (md_reader::record_potential_iterator_use): Replace
19951 pointer argument with an rtx and an index.
19952 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19953 (apply_mode_iterator): Likewise.
19954 (apply_code_iterator): Likewise.
19955 (apply_int_iterator): Likewise.
19956 (apply_subst_iterator): Likewise.
19957 (record_iterator_use): Likewise.
19958 (record_attribute_use): Likewise.
19959 (md_reader::record_potential_iterator_use): Likewise. Update calls
19960 to record_iterator_use and apply_iterator.
19961 (iterator_use): Replace ptr with x and index.
19962 (attribute_use): Likewise.
19963 (apply_attribute_uses): Update calls to apply_iterator.
19964 (apply_iterators): Likewise. Update initialization of iterator_use.
19965 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19966 and record_potential_iterator_use.
19967 (rtx_reader::read_rtx_operand): Likewise.
19969 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19970 Alan Hayward <alan.hayward@arm.com>
19971 David Sherwood <david.sherwood@arm.com>
19973 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19976 2017-08-21 Richard Biener <rguenther@suse.de>
19978 PR middle-end/81884
19979 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19980 at struct end conservatively when comparing common bases.
19982 2017-08-21 Richard Biener <rguenther@suse.de>
19984 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19985 (mem_ref_in_stmt): Remove.
19986 (determine_max_movement): Use ref index to get at the reference.
19987 (invariantness_dom_walker::before_dom_children): Deal with
19988 lim data already initialized.
19989 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19991 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19993 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19994 (TARGET_ISA_ROUND): Ditto.
19995 (TARGET_ROUND): Ditto.
19996 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19997 * config/i386/i386.md: Ditto.
19998 * config/i386/sse.md: Ditto.
19999 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20000 with OPTION_MASK_ISA_SSE4_1.
20002 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20005 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20006 __builtin_ia32_lzcnt_u16.
20008 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20011 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20012 (rs6000_set_current_function): Rewrite function to use it.
20014 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20017 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20018 and TYPE_WARN_IF_NOT_ALIGN.
20019 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20020 (handle_warn_if_not_align): New.
20021 (place_union_field): Call handle_warn_if_not_align.
20022 (place_field): Call handle_warn_if_not_align.
20023 Copy TYPE_WARN_IF_NOT_ALIGN.
20024 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20025 (layout_type): Likewise.
20026 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20028 (tree_decl_common): Add warn_if_not_align.
20029 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20030 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20031 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20032 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20033 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20034 * doc/extend.texi: Document warn_if_not_aligned attribute.
20035 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20037 2017-08-17 Martin Liska <mliska@suse.cz>
20040 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20041 (get_data_dependence): Use it as pointer type.
20042 (distribute_loop): Likewise.
20044 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20046 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20047 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20048 (p8_vmrgow_<mode>_direct): New define_insn.
20049 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20050 handle endianness for vmrgew and vmrgow permute patterns.
20052 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20054 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20055 * config/rs6000/rs6000-cpus.def: Remove comment.
20056 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20057 (POWERPC_MASKS): Likewise.
20058 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20059 use of TARGET_VSX_TIMODE.
20060 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20061 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20062 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20063 (rs6000_option_override_internal): Remove dead code.
20064 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20065 (rs6000_legitimize_reload_address): Likewise.
20066 (rs6000_legitimate_address_p): Likewise.
20067 (rs6000_opt_masks): Delete "vsx-timode".
20068 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20069 from function comment.
20070 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20071 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20072 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20074 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20075 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20076 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20077 (VSX_TI): Likewise.
20079 (define_peephole2): Likewise.
20081 2017-08-17 Martin Sebor <msebor@redhat.com>
20084 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20085 past the end of an array.
20086 (test_pp_format): Add test cases.
20088 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20090 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20091 missing ECF_NOTHROW flags.
20093 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20096 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20097 operands residing in integer registers.
20098 (*vsx_le_perm_load_<mode>): Likewise.
20099 (*vsx_le_perm_store_<mode>): Likewise.
20100 (define_peephole2): Add peepholes to optimize the above.
20102 2017-08-17 Marek Polacek <polacek@redhat.com>
20104 PR middle-end/81814
20105 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20106 to mimic what shorten_compare did. Change the return type to bool.
20107 (fold_cond_expr_with_comparison): Update call to
20108 operand_equal_for_comparison_p.
20109 (fold_ternary_loc): Likewise.
20111 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20113 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20115 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20117 2017-08-17 Richard Biener <rguenther@suse.de>
20119 * tree-ssa-structalias.c (solve_graph): When propagating
20120 to successors update the graphs succ edges and avoid duplicate work.
20122 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20125 * config/i386/i386.c (ix86_option_override_internal): Save target
20126 specific options after ix86_stack_protector_guard_reg was changed.
20128 2017-08-17 Richard Biener <rguenther@suse.de>
20130 PR tree-optimization/81827
20131 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20132 (new_var_info): Initialize it conservatively.
20133 (get_call_vi): Mark register vars.
20134 (new_scalar_tmp_constraint_exp): Likewise.
20135 (handle_rhs_call): Likewise.
20136 (handle_const_call): Likewise.
20137 (create_function_info_for): Likewise.
20138 (solve_constraints): Sort varinfos to separate register from
20139 non-register vars to pack points-to solution bitmaps during
20142 2017-08-17 Marek Polacek <polacek@redhat.com>
20144 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20146 2017-08-17 Richard Biener <rguenther@suse.de>
20148 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20149 to true when overflow is undefined and we saturated the result.
20151 2017-08-17 Alan Modra <amodra@gmail.com>
20154 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20155 Don't use store multiple if only one reg needs saving.
20156 (interesting_frame_related_regno): New function.
20157 (rs6000_frame_related): Don't emit frame info for regs that
20159 (rs6000_emit_epilogue): Likewise.
20161 2017-08-16 Nathan Sidwell <nathan@acm.org>
20163 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20164 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20165 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20166 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20167 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20168 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20169 (verify_type): Adjust for TYPE_BINFO move.
20170 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20171 process TYPE_BINFO directly.
20172 (hash_tree): Likewise.
20173 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20175 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20178 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20180 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20182 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20185 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20186 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20187 Add ix86_binary_operator_ok to insn constraint.
20188 (*iordi_1_bts): Ditto.
20189 (*xordi_1_btc): Ditto.
20190 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20191 Update corresponding peephole2 pattern.
20195 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20197 PR tree-optimization/81832
20198 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20199 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20201 2017-08-16 Marek Polacek <polacek@redhat.com>
20204 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20205 regarding size_zero_node.
20207 2017-08-16 Martin Liska <mliska@suse.cz>
20210 * config.gcc: Respect previously set extra_objs in case
20213 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20215 PR tree-optimization/81835
20216 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20217 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20218 not depend on the phi.
20220 2017-08-16 Alan Modra <amodra@gmail.com>
20222 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20225 2017-08-16 Alan Modra <amodra@gmail.com>
20227 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20228 (save_reg_p): ..into this. Update all callers.
20229 (first_reg_to_save): Simplify.
20231 2017-08-16 Alan Modra <amodra@gmail.com>
20233 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20236 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20240 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20241 constant count if that count is less than 32.
20243 2017-08-15 Nathan Sidwell <nathan@acm.org>
20245 * gcc.c (execute): Emit friendlier message if inferior is killed
20246 by an external cause.
20248 2017-08-15 Richard Biener <rguenther@suse.de>
20250 PR tree-optimization/81790
20251 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20252 CONSTRUCTORs from simplifying and VN.
20254 2017-08-14 Martin Sebor <msebor@redhat.com>
20256 * builtin-attrs.def: Add comments.
20258 2017-08-14 Martin Sebor <msebor@redhat.com>
20261 * doc/extend.texi (attribute nonstring): Document new attribute.
20263 2017-08-14 Martin Sebor <msebor@redhat.com>
20266 * tree-diagnostic.c (default_tree_printer): Handle %G.
20267 * gimple-pretty-print.h (percent_G_format): Declare new function.
20268 * gimple-pretty-print.c (percent_G_format): Define.
20269 * tree-pretty-print.c (percent_K_format): Add argument.
20271 2017-08-14 Martin Sebor <msebor@redhat.com>
20273 PR translation/79998
20274 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20275 Remove a stray space.
20277 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20280 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20281 (*iordi_1_bts): Ditto.
20282 (*xordi_1_btc): Ditto.
20284 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20287 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20288 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20290 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20291 quoted strings, and make more translator-friendly.
20292 (darwin_rs6000_override_options): Likewise.
20293 (rs6000_option_override_internal): Likewise.
20294 (rs6000_return_in_memory): Fix overlong line.
20295 (init_cmulative_args): Use quoted strings, and make more
20296 translator-friendly.
20297 (rs6000_pass_by_reference): Fix overlong line.
20298 (def_builtin): Use quoted strings.
20299 (altivec_expand_predicate_builtin): Use quoted strings, and make
20300 more translator-friendly.
20301 (htm_expand_builtin): Use quoted strings.
20302 (cpu_expand_builtin): Use quoted strings, and make more
20303 translator-friendly.
20304 (altivec_expand_builtin): Likewise.
20305 (paired_expand_predicate_builtin): Likewise.
20306 (rs6000_invalid_builtin): Likewise.
20307 (builtin_function_type): Use quoted strings.
20308 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20309 more translator-friendly.
20310 (rs6000_trampoline_init): Likewise.
20311 (rs6000_handle_altivec_attribute): Likewise.
20312 (rs6000_inner_target_options): Use quoted strings.
20313 (rs6000_disable_incompatible_switches): Likewise.
20314 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20315 strings, and make more translator-friendly.
20316 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20318 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20320 PR tree-optimization/81799
20321 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20322 cond_expr to simple gimple operand.
20324 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20326 PR middle-end/46932
20327 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20329 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20333 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20334 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20335 TARGET_GASISR_PROLOGUES.
20336 * config/avr/avr.c (avr_option_override): Same.
20337 (avr_pass_pre_proep::execute): Same.
20339 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20342 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20343 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20345 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20347 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20348 define_insn_and_split. Split to a memory load from 0 in
20349 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20350 using PTR mode iterator.
20351 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20352 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20353 (*add_tp_<mode>): Redefine as define_insn_and_split.
20354 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20355 address space. Merge with *add_tp_x32 using PTR mode iterator.
20356 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20357 Split to an add with a memory load from 0 in
20358 DEFAULT_TLS_SEG_REG address space.
20360 2017-08-12 Andrew Pinski <apinski@cavium.com>
20362 * config/aarch64/aarch64-option-extensions.def (rdma):
20363 Fix feature string to what Linux prints out in /proc/cpuinfo.
20365 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20368 * dwarf2out.c (modified_type_die): For C typedef types that have
20369 an ultimate origin, process the ultimate origin instead of the
20371 (gen_typedef_die): Assert that input DECLs have no ultimate
20373 (gen_type_die_with_usage): For typedef variants that have an
20374 ultimate origin, just call gen_decl_die on the original DECL.
20375 (process_scope_var): Avoid creating DIEs for local typedefs and
20376 concrete static variables.
20378 2017-08-12 Alan Modra <amodra@gmail.com>
20382 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20383 match gnu-user.h startfile.
20384 (ENDFILE_LINUX_SPEC): Similarly.
20386 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20389 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20391 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20393 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20394 * config/aarch64/aarch64.md (mov<mode>): Change.
20395 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20396 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20397 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20399 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20401 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20402 for storage order barriers.
20404 2017-08-11 Martin Liska <mliska@suse.cz>
20407 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20408 variables of void type.
20410 2017-08-11 Martin Liska <mliska@suse.cz>
20412 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20413 TARGET_SUPPORTS_ALIASES.
20414 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20415 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20416 (optimize_weakref): Likewise.
20417 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20418 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20419 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20421 2017-08-11 Martin Liska <mliska@suse.cz>
20424 * config/i386/i386.c (make_resolver_func): Do complete
20425 refactoring of the function.
20427 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20430 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20431 * config/i386/i386.c (ix86_stack_protect_guard): Use
20432 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20433 * doc/invoke.texi (x86 Options): Document
20434 -mstack-protector-guard-symbol= option.
20436 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20438 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
20439 * config/i386/i386.c (ix86_split_stack_guard): New function.
20440 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
20441 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
20442 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
20443 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
20444 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
20445 (split_stack_space_check): Call ix86_split_stack_guard.
20447 2017-08-10 Martin Sebor <msebor@redhat.com>
20449 * print-tree.c (print_node): Print location using the established
20451 Replace spaces with colons.
20452 (debug_raw, debug): Ditto.
20454 2017-08-10 Martin Sebor <msebor@redhat.com>
20457 * pretty-print.c (pp_format): Correct the handling of %s precision.
20459 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
20462 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20464 (ix86_finalize_stack_frame_flags): This. Also clear
20465 frame_pointer_needed if -fno-omit-frame-pointer is used without
20467 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20468 with ix86_finalize_stack_frame_flags.
20469 (ix86_expand_epilogue): Likewise.
20470 (ix86_expand_split_stack_prologue): Likewise.
20471 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
20473 2017-08-10 Martin Liska <mliska@suse.cz>
20476 * c-attribs.c (handle_target_attribute):
20477 Report warning for an empty string argument of target attribute.
20479 2017-08-09 Jakub Jelinek <jakub@redhat.com>
20482 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
20484 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
20485 or DECL_NONLOCAL labels.
20486 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
20487 or DECL_NONLOCAL labels here.
20489 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
20491 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
20492 to indicate when early gimple folding has been disabled.
20493 (rs6000_gimple_fold_builtin): Add debug content.
20494 (rs6000_invalid_builtin): Fix whitespace.
20495 (rs6000_expand_builtin): Fix whitespace.
20496 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
20498 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
20501 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
20502 SAVE_MULTIPLE if not all the registers that saves, should be saved.
20504 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
20506 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
20507 (qdf24xx): Likewise.
20508 * config/aarch64/aarch64.md: Include falkor.md.
20509 * config/aarch64/falkor.md: New.
20511 2017-08-09 Marek Polacek <polacek@redhat.com>
20514 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
20515 * diagnostic.c (emit_diagnostic): Add a comment.
20516 (emit_diagnostic_valist): New function.
20518 2017-08-09 Marek Polacek <polacek@redhat.com>
20521 * input.c (make_location): New overload.
20522 * input.h (make_location): Declare.
20524 2017-08-08 Alan Modra <amodra@gmail.com>
20525 H.J. Lu <hongjiu.lu@intel.com>
20528 * gcc.c (NO_PIE_SPEC): Delete.
20529 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
20531 (LINK_PIE_SPEC): ..to here.
20532 (LINK_COMMAND_SPEC): Support -no-pie.
20533 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
20534 chain of crtbegin*.o selection, update for PIE_SPEC changes and
20536 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
20537 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
20538 (ENDFILE_CRTEND_SPEC): Similarly.
20540 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
20543 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
20544 (mstack-protector-guard-offset=): Ditto.
20545 * config/i386/i386.c (ix86_option_override): Handle
20546 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
20548 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
20549 ix86_stack_protect_guard_offset variables.
20550 (TARGET_STACK_PROTECT_GUARD): Always define.
20551 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
20552 and -mstack-protector-guard-offset= options.
20554 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20556 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
20557 boundary case for the last candidate.
20559 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20561 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
20562 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
20564 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20566 PR middle-end/19706
20567 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
20568 * config/aarch64/aarch64-builtins.c
20569 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
20570 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
20571 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
20573 2017-08-08 Tamar Christina <tamar.christina@arm.com>
20574 Andrew Pinski <pinskia@gmail.com>
20576 PR middle-end/19706
20577 * internal-fn.def (XORSIGN): New.
20578 * optabs.def (xorsign_optab): New.
20579 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
20580 (convert_expand_mult_copysign): New.
20581 (pass_optimize_widening_mul::execute): Call
20582 convert_expand_mult_copysign.
20584 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20586 PR tree-optimization/81354
20587 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
20588 Insert on edges rather than explicitly creating landing pads.
20589 (analyze_candidates_and_replace): Commit edge inserts.
20591 2017-08-08 Richard Biener <rguenther@suse.de>
20593 PR middle-end/81719
20594 * tree-ssa-loop-niter.c: Include tree-dfa.h.
20595 (expand_simple_operations): Also look through ADDR_EXPRs with
20596 MEM_REF bases treating them as POINTER_PLUS_EXPR.
20598 2017-08-08 Richard Biener <rguenther@suse.de>
20600 PR tree-optimization/81723
20601 * tree-vect-slp.c (struct bst_traits): New hash traits.
20602 (bst_fail): New global.
20603 (vect_build_slp_tree_2): New worker, split out from ...
20604 (vect_build_slp_tree): ... this now wrapping it with using
20605 bst_fail set to cache SLP tree build fails. Properly handle
20607 (vect_analyze_slp_instance): Allocate and free bst_fail.
20609 2017-08-08 Martin Liska <mliska@suse.cz>
20612 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
20613 LABEL_DECLs that can be from a different function.
20615 2017-08-08 Bin Cheng <bin.cheng@arm.com>
20617 PR tree-optimization/81744
20618 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
20619 loop's number of iterations.
20621 2017-08-08 Martin Liska <mliska@suse.cz>
20623 * asan.c: Include header files.
20624 * attribs.c (build_decl_attribute_variant): New function moved
20626 (build_type_attribute_qual_variant): Likewise.
20627 (cmp_attrib_identifiers): Likewise.
20628 (simple_cst_list_equal): Likewise.
20629 (omp_declare_simd_clauses_equal): Likewise.
20630 (attribute_value_equal): Likewise.
20631 (comp_type_attributes): Likewise.
20632 (build_type_attribute_variant): Likewise.
20633 (lookup_ident_attribute): Likewise.
20634 (remove_attribute): Likewise.
20635 (merge_attributes): Likewise.
20636 (merge_type_attributes): Likewise.
20637 (merge_decl_attributes): Likewise.
20638 (merge_dllimport_decl_attributes): Likewise.
20639 (handle_dll_attribute): Likewise.
20640 (attribute_list_equal): Likewise.
20641 (attribute_list_contained): Likewise.
20642 * attribs.h (lookup_attribute): New function moved from tree.[ch].
20643 (lookup_attribute_by_prefix): Likewise.
20644 * bb-reorder.c: Include header files.
20645 * builtins.c: Likewise.
20646 * calls.c: Likewise.
20647 * cfgexpand.c: Likewise.
20648 * cgraph.c: Likewise.
20649 * cgraphunit.c: Likewise.
20650 * convert.c: Likewise.
20651 * dwarf2out.c: Likewise.
20652 * final.c: Likewise.
20653 * fold-const.c: Likewise.
20654 * function.c: Likewise.
20655 * gimple-expr.c: Likewise.
20656 * gimple-fold.c: Likewise.
20657 * gimple-pretty-print.c: Likewise.
20658 * gimple.c: Likewise.
20659 * gimplify.c: Likewise.
20660 * hsa-common.c: Likewise.
20661 * hsa-gen.c: Likewise.
20662 * internal-fn.c: Likewise.
20663 * ipa-chkp.c: Likewise.
20664 * ipa-cp.c: Likewise.
20665 * ipa-devirt.c: Likewise.
20666 * ipa-fnsummary.c: Likewise.
20667 * ipa-inline.c: Likewise.
20668 * ipa-visibility.c: Likewise.
20670 * lto-cgraph.c: Likewise.
20671 * omp-expand.c: Likewise.
20672 * omp-general.c: Likewise.
20673 * omp-low.c: Likewise.
20674 * omp-offload.c: Likewise.
20675 * omp-simd-clone.c: Likewise.
20676 * opts-global.c: Likewise.
20677 * passes.c: Likewise.
20678 * predict.c: Likewise.
20679 * sancov.c: Likewise.
20680 * sanopt.c: Likewise.
20681 * symtab.c: Likewise.
20682 * toplev.c: Likewise.
20683 * trans-mem.c: Likewise.
20684 * tree-chkp.c: Likewise.
20685 * tree-eh.c: Likewise.
20686 * tree-into-ssa.c: Likewise.
20687 * tree-object-size.c: Likewise.
20688 * tree-parloops.c: Likewise.
20689 * tree-profile.c: Likewise.
20690 * tree-ssa-ccp.c: Likewise.
20691 * tree-ssa-live.c: Likewise.
20692 * tree-ssa-loop.c: Likewise.
20693 * tree-ssa-sccvn.c: Likewise.
20694 * tree-ssa-structalias.c: Likewise.
20695 * tree-ssa.c: Likewise.
20696 * tree-streamer-in.c: Likewise.
20697 * tree-vectorizer.c: Likewise.
20698 * tree-vrp.c: Likewise.
20699 * tsan.c: Likewise.
20700 * ubsan.c: Likewise.
20701 * varasm.c: Likewise.
20702 * varpool.c: Likewise.
20703 * tree.c: Remove functions moved to attribs.[ch].
20704 * tree.h: Likewise.
20705 * config/aarch64/aarch64.c: Add attrs.h header file.
20706 * config/alpha/alpha.c: Likewise.
20707 * config/arc/arc.c: Likewise.
20708 * config/arm/arm.c: Likewise.
20709 * config/avr/avr.c: Likewise.
20710 * config/bfin/bfin.c: Likewise.
20711 * config/c6x/c6x.c: Likewise.
20712 * config/cr16/cr16.c: Likewise.
20713 * config/cris/cris.c: Likewise.
20714 * config/darwin.c: Likewise.
20715 * config/epiphany/epiphany.c: Likewise.
20716 * config/fr30/fr30.c: Likewise.
20717 * config/frv/frv.c: Likewise.
20718 * config/ft32/ft32.c: Likewise.
20719 * config/h8300/h8300.c: Likewise.
20720 * config/i386/winnt.c: Likewise.
20721 * config/ia64/ia64.c: Likewise.
20722 * config/iq2000/iq2000.c: Likewise.
20723 * config/lm32/lm32.c: Likewise.
20724 * config/m32c/m32c.c: Likewise.
20725 * config/m32r/m32r.c: Likewise.
20726 * config/m68k/m68k.c: Likewise.
20727 * config/mcore/mcore.c: Likewise.
20728 * config/microblaze/microblaze.c: Likewise.
20729 * config/mips/mips.c: Likewise.
20730 * config/mmix/mmix.c: Likewise.
20731 * config/mn10300/mn10300.c: Likewise.
20732 * config/moxie/moxie.c: Likewise.
20733 * config/msp430/msp430.c: Likewise.
20734 * config/nds32/nds32-isr.c: Likewise.
20735 * config/nds32/nds32.c: Likewise.
20736 * config/nios2/nios2.c: Likewise.
20737 * config/nvptx/nvptx.c: Likewise.
20738 * config/pa/pa.c: Likewise.
20739 * config/pdp11/pdp11.c: Likewise.
20740 * config/powerpcspe/powerpcspe.c: Likewise.
20741 * config/riscv/riscv.c: Likewise.
20742 * config/rl78/rl78.c: Likewise.
20743 * config/rx/rx.c: Likewise.
20744 * config/s390/s390.c: Likewise.
20745 * config/sh/sh.c: Likewise.
20746 * config/sol2.c: Likewise.
20747 * config/sparc/sparc.c: Likewise.
20748 * config/spu/spu.c: Likewise.
20749 * config/stormy16/stormy16.c: Likewise.
20750 * config/tilegx/tilegx.c: Likewise.
20751 * config/tilepro/tilepro.c: Likewise.
20752 * config/v850/v850.c: Likewise.
20753 * config/vax/vax.c: Likewise.
20754 * config/visium/visium.c: Likewise.
20755 * config/xtensa/xtensa.c: Likewise.
20757 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20760 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
20761 constraints since the -mupper-regs-* switches have been
20763 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
20764 into a vector from a double word element that was extracted from
20765 another vector, and eliminate extra XXPERMDI instructions.
20766 (vsx_concat_<mode>_2): Likewise.
20767 (vsx_concat_<mode>_3): Likewise.
20768 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20769 concat to allow optimizing inserts from previous extracts.
20771 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20773 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20774 memory reference to a SSP offset in TLS address space.
20775 (ix86_print_operand) <case '@'>: Remove.
20776 (ix86_print_operand_punct_valid_p): Remove '@' code.
20777 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20778 UNSPEC_SP_TLS_TEST.
20779 (stack_tls_protect_set_<mode>): Remove.
20780 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20781 (stack_tls_protect_test_<mode>): Remove.
20782 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20784 2017-08-07 Olivier Hainque <hainque@adacore.com>
20787 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20789 2017-08-07 Douglas Rupp <rupp@adacore.com>
20791 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20792 variable was referenced as multidir in command.
20794 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20797 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20800 2017-08-07 Martin Liska <mliska@suse.cz>
20802 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20803 * config/rl78/rl78.c: Add include of attribs.h.
20804 * config/sh/sh.c: Likewise.
20805 * config/v850/v850.c: Likewise.
20807 2017-08-07 Tom de Vries <tom@codesourcery.com>
20809 PR middle-end/78266
20810 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20812 2017-08-07 Martin Liska <mliska@suse.cz>
20814 * config/mips/mips.c: Include attribs.h.
20816 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20819 * doc/invoke.texi: Document change in behvaior for -Ofast for
20822 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20824 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20826 (aarch64_pop_regs): Likewise.
20827 (aarch64_gen_load_pair): Likewise.
20828 (aarch64_save_callee_saves): Likewise.
20829 (aarch64_restore_callee_saves): Likewise.
20831 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20833 * config/i386/i386.c: Revert the last change.
20835 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20838 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20840 (ix86_finalize_stack_frame_flags): This. Also clear
20841 frame_pointer_needed if -fno-omit-frame-pointer is used without
20843 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20844 with ix86_finalize_stack_frame_flags.
20845 (ix86_expand_epilogue): Likewise.
20846 (ix86_expand_split_stack_prologue): Likewise.
20848 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20851 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20852 to P_AES for Westmere.
20854 2017-08-07 Jonathan Yong <10walls@gmail.com>
20856 * config/i386/mingw.opt (fset-stack-executable): Removed.
20857 * config/i386/cygming.opt (fset-stack-executable): Moved
20859 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20861 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20863 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20865 2017-08-07 Marek Polacek <polacek@redhat.com>
20867 PR middle-end/81737
20868 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20870 2017-08-07 Martin Liska <mliska@suse.cz>
20872 * attribs.h (canonicalize_attr_name): New function.
20873 (cmp_attribs): Move from c-format.c and adjusted.
20874 (is_attribute_p): Moved from tree.h.
20875 * tree-inline.c: Add new includes.
20876 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20877 (private_is_attribute_p): Remove.
20878 (private_lookup_attribute): Likewise.
20879 (private_lookup_attribute_by_prefix): Simplify.
20880 (remove_attribute): Use is_attribute_p.
20881 * tree.h: Remove removed declarations.
20883 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20885 PR middle-end/81698
20886 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20887 instead of computing it in the function. Formatting fix.
20888 (expand_case): Don't rely on default_edge being the first edge,
20889 clear it if removing it, pass default_edge to
20890 emit_case_dispatch_table.
20891 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20894 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20896 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20897 insn in the function, emit NOP after the insn.
20899 2017-08-06 Tom de Vries <tom@codesourcery.com>
20901 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20904 2017-08-06 Tom de Vries <tom@codesourcery.com>
20906 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20909 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20911 PR tree-optimization/57371
20912 * match.pd: New pattern.
20914 2017-08-04 Marek Polacek <polacek@redhat.com>
20916 PR middle-end/81695
20917 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20918 perform the computation in offset_int.
20920 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20922 PR tree-optimization/81136
20923 * tree-vectorizer.h: Include tree-hash-traits.h.
20924 (vec_base_alignments): New typedef.
20925 (vec_info): Add a base_alignments field.
20926 (vect_record_base_alignments): Declare.
20927 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20929 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20930 (create_data_ref): Add an is_conditional_in_stmt argument.
20931 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20932 the is_conditional_in_stmt field.
20933 (data_ref_loc): Add an is_conditional_in_stmt field.
20934 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20935 (find_data_references_in_stmt): Update call to create_data_ref.
20936 (graphite_find_data_references_in_stmt): Likewise.
20937 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20938 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20939 (vect_record_base_alignment): New function.
20940 (vect_record_base_alignments): Likewise.
20941 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20942 for nested statements even if we fail to compute a misalignment.
20943 Use pooled base alignments for unconditional references.
20944 (vect_find_same_alignment_drs): Compare base addresses instead
20946 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20947 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20949 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20951 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20952 Add an explicit name for the enum. Use auto_vec for slp_instances
20953 and grouped_stores.
20954 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20956 (_bb_vec_info): Add a constructor and destructor.
20957 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20958 (destroy_loop_vec_info): Delete.
20959 (vect_destroy_datarefs): Likewise.
20960 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20961 (vec_info::vec_info): New function.
20962 (vec_info::~vec_info): Likewise.
20963 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20964 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20965 destroy_loop_vec_info.
20966 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20967 (_loop_vec_info::_loop_vec_info): ...this.
20968 (destroy_loop_vec_info): Replace with...
20969 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20970 the stmt_vec_infos. Leave handling of vec_info information to its
20971 destructor. Remove explicit vector releases.
20972 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20973 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20974 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20975 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20976 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20977 (destroy_bb_vec_info): Replace with...
20978 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20979 information to its destructor.
20980 (vect_slp_analyze_bb_1): Use new and delete instead of
20981 new_bb_vec_info and destroy_bb_vec_info.
20982 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20985 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20987 * tree-data-ref.h (subscript): Add access_fn field.
20988 (data_dependence_relation): Add could_be_independent_p.
20989 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20990 (same_access_functions): Move to tree-data-ref.c.
20991 * tree-data-ref.c (ref_contains_union_access_p): New function.
20992 (access_fn_component_p): Likewise.
20993 (access_fn_components_comparable_p): Likewise.
20994 (dr_analyze_indices): Add a reference to access_fn_component_p.
20995 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20997 (constant_access_functions): Likewise.
20998 (add_other_self_distances): Likewise.
20999 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21000 (initialize_data_dependence_relation): Use XCNEW and remove
21001 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21002 of access functions that have the same type. Allow the
21003 subsequence to end with different bases in some circumstances.
21004 Record the chosen access functions in SUB_ACCESS_FN.
21005 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21006 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21007 (subscript_dependence_tester_1): Likewise dra and drb.
21008 (build_classic_dist_vector): Update calls accordingly.
21009 (subscript_dependence_tester): Likewise.
21010 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21011 DDR_COULD_BE_INDEPENDENT_P.
21012 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21013 comp_alias_ddrs instead of may_alias_ddrs.
21014 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21016 (vect_analyze_data_ref_dependence): Use it if
21017 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21018 distance vectors if that fails.
21019 (dependence_distance_ge_vf): New function.
21020 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21021 LOOP_VINFO_MAY_ALIAS_DDRS.
21023 2017-08-04 Richard Biener <rguenther@suse.de>
21025 PR middle-end/81705
21026 * fold-const.c (fold_binary_loc): Properly restrict
21027 minus_var0 && minus_var1 case when associating undefined overflow
21030 2017-08-04 Tom de Vries <tom@codesourcery.com>
21032 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21034 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21036 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21037 Don't start diagnostic messages with a capital letter.
21038 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21040 (rs6000_invalid_builtin): Likewise.
21041 (rs6000_trampoline_init): Likewise.
21043 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21046 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21047 after setting changeable df flags.
21049 2017-08-03 Richard Biener <rguenther@suse.de>
21051 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21052 up if the use is in USE - X.
21054 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21056 * toplev.c (dumpfile.h): New include.
21057 (internal_error_reentered): New static function. Use it...
21058 (internal_error_function): ...here to handle reentered internal_error.
21060 2017-08-03 Richard Biener <rguenther@suse.de>
21062 PR middle-end/81148
21063 * fold-const.c (split_tree): Add minus_var and minus_con
21064 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21065 here but always use minus_*.
21066 (associate_trees): Assert we never associate with MINUS_EXPR
21067 and NULL first operand. Do not recurse for PLUS_EXPR operands
21068 when associating as MINUS_EXPR either.
21069 (fold_binary_loc): Track minus_var and minus_con.
21071 2017-08-03 Tom de Vries <tom@codesourcery.com>
21074 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21075 ACCEL_COMPILER, apply finish_options on
21076 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21078 2017-08-03 Tom de Vries <tom@codesourcery.com>
21081 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21082 function_entry_patch_area_size > 0.
21084 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21087 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21088 instead of 10??LU, perform unit multiplication in wide_int,
21089 don't change alloc_object_size_limit if the limit is larger
21092 PR tree-optimization/81655
21093 PR tree-optimization/81588
21094 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21095 the case when ranges[i].low and high are 1 for unsigned type with
21098 PR middle-end/81052
21099 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21100 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21102 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21104 * tree-vrp.h: Add include guard.
21106 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21109 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21110 (ud2): New insn pattern.
21111 * config/i386/i386.c (ix86_expand_epilogue):
21112 For naked functions, generate ud2 instead of trap insn.
21114 2017-08-02 Marek Polacek <polacek@redhat.com>
21117 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21119 2017-08-02 Tom de Vries <tom@codesourcery.com>
21120 Cesar Philippidis <cesar@codesourcery.com>
21122 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21123 Add missing edge probabilities.
21125 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21127 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21128 Correct endianness.
21130 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21132 PR middle-end/79499
21133 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21134 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21135 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21136 prologue_seq sequences - if any.
21138 2017-08-02 Richard Biener <rguenther@suse.de>
21140 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21141 via vectors if supported, integer extracts via punning if supported
21142 or otherwise vector extracts.
21144 2017-08-02 Richard Biener <rguenther@suse.de>
21146 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21148 (bitmap_insert_into_set): ... this.
21150 2017-08-02 Richard Biener <rguenther@suse.de>
21152 PR tree-optimization/81633
21154 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21156 PR tree-optimization/71752
21157 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21159 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21161 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21162 (machine_function::call_ms2sysv_pad_out): Remove field.
21163 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21164 (ix86_compute_frame_layout): Likewise.
21166 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21169 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21170 attribute with interrupt attribute.
21172 2017-08-01 Andrew Pinski <apinski@cavium.com>
21174 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21175 BIT_INSERT_EXPR's operand 1
21176 to see if the types precision matches.
21178 2017-08-01 Martin Liska <mliska@suse.cz>
21180 PR middle-end/70140
21181 * builtins.c (expand_builtin_memcpy_args): Remove.
21182 (expand_builtin_memcpy): Call newly added function
21183 expand_builtin_memory_copy_args.
21184 (expand_builtin_memcpy_with_bounds): Likewise.
21185 (expand_builtin_mempcpy): Remove last argument.
21186 (expand_builtin_mempcpy_with_bounds): Likewise.
21187 (expand_builtin_memory_copy_args): New function created from
21188 expand_builtin_mempcpy_args with small modifications.
21189 (expand_builtin_mempcpy_args): Remove.
21190 (expand_builtin_stpcpy): Remove unused argument.
21191 (expand_builtin): Likewise.
21192 (expand_builtin_with_bounds): Likewise.
21194 2017-08-01 Martin Liska <mliska@suse.cz>
21197 Make mempcpy more optimal (PR middle-end/70140).
21199 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21202 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21203 __builtin_vec_cmpne verify both arguments are compatible vectors
21204 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21205 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21206 move computation of aligned to after checking the argument types.
21210 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21211 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21214 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21215 Doug Rupp <rupp@adacore.com>
21216 Olivier Hainque <hainque@adacore.com>
21218 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21219 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21221 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21222 for TARGET_OS_CPP_BUILTIN.
21223 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21224 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21226 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21227 passing required abi options to the assembler for EABI configurations.
21228 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21229 of .text.hot and .text.unlikely sections for kernel modules when
21230 using ARM style exceptions.
21231 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21232 options. Add EXTRA_CC1_SPEC.
21233 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21235 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21237 (ARM_TARGET2_DWARF_FORMAT): Define.
21238 * config/arm/t-vxworks: Adjust multilib control to removal of the
21239 Diab command line options.
21241 2017-08-01 Martin Liska <mliska@suse.cz>
21243 PR gcov-profile/81561
21244 * gcov.c (unblock): Make unblocking safe as we need to preserve
21245 index correspondence of blocks and block_lists.
21247 2017-08-01 Richard Biener <rguenther@suse.de>
21249 PR tree-optimization/81181
21250 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21251 (compute_antic): ... end of iteration here.
21253 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21255 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21256 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21257 (ftree-slp-vectorize): Likewise.
21258 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21259 can no longer be set independent of flag_tree_loop_vectorize.
21260 * omp-general.c (emp_max_vf): Likewise.
21261 * opts.c (enable_fdo_optimizations): Remove references to
21262 flag_tree_vectorize, these are now implicit.
21263 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21264 and leave it for the options machinery.
21266 2017-08-01 Martin Liska <mliska@suse.cz>
21268 PR middle-end/70140
21269 * builtins.c (expand_builtin_memcpy_args): Remove.
21270 (expand_builtin_memcpy): Call newly added function
21271 expand_builtin_memory_copy_args.
21272 (expand_builtin_memcpy_with_bounds): Likewise.
21273 (expand_builtin_mempcpy): Remove last argument.
21274 (expand_builtin_mempcpy_with_bounds): Likewise.
21275 (expand_builtin_memory_copy_args): New function created from
21276 expand_builtin_mempcpy_args with small modifications.
21277 (expand_builtin_mempcpy_args): Remove.
21278 (expand_builtin_stpcpy): Remove unused argument.
21279 (expand_builtin): Likewise.
21280 (expand_builtin_with_bounds): Likewise.
21282 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21285 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21286 print "ds:" only for immediates in generic address space.
21288 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21291 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21292 (ix86_function_ok_for_sibcall): Return false for naked functions.
21294 2017-08-01 Richard Biener <rguenther@suse.de>
21296 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21297 (compute_antic): Seed worklist with exit block predecessors.
21298 * cfganal.c (dfs_find_deadend): For a cycle return the source
21299 of the edge closing it.
21301 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21303 * config/aarch64/aarch64.c
21304 (aarch64_can_const_movi_rtx_p): Move 0 check.
21306 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21308 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21309 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21311 * match.pd: Ditto in address comparison pattern.
21313 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21315 PR tree-optimization/81627
21316 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21317 closed ssa form for store-store chain.
21319 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21321 PR tree-optimization/81620
21322 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21323 for store-store chain.
21325 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21327 PR tree-optimization/81588
21328 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21329 ranges[i].in_p, invert comparison code ccode. For >/>=,
21330 swap rhs1 and rhs2 and comparison code unconditionally,
21331 for </<= don't do that. Don't swap rhs1/rhs2 again if
21332 ranges[i].in_p, instead invert comparison code ccode if
21333 opcode or oe->rank is BIT_IOR_EXPR.
21336 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21337 a direct optab to conversion optab.
21338 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21339 with GET_MODE_INNER as last argument instead of optab_handler.
21340 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21341 vector extraction if possible and optab is available.
21342 * expr.c (store_constructor): Use convert_optab_handler instead
21343 of optab_handler. Use vector initialization from smaller
21344 vectors if possible and optab is available.
21345 * tree-vect-stmts.c (vectorizable_load): Likewise.
21346 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21347 now have two modes.
21348 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21349 of vec_init from half-sized vectors with the same element mode.
21350 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21351 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21352 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21353 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21354 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21355 after mode in gen_vec_extract* calls.
21356 (vec_extract<mode>): Renamed to ...
21357 (vec_extract<mode><ssescalarmodelower>): ... this.
21358 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21359 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21360 element mode after mode in gen_vec_init* calls.
21361 (VEC_INIT_HALF_MODE): New mode iterator.
21362 (vec_init<mode>): Renamed to ...
21363 (vec_init<mode><ssescalarmodelower>): ... this.
21364 (vec_init<mode><ssehalfvecmodelower>): New expander.
21365 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21366 (vec_extractv2sfsf): ... this.
21367 (vec_initv2sf): Renamed to ...
21368 (vec_initv2sfsf): ... this.
21369 (vec_extractv2si): Renamed to ...
21370 (vec_extractv2sisi): ... this.
21371 (vec_initv2si): Renamed to ...
21372 (vec_initv2sisi): ... this.
21373 (vec_extractv4hi): Renamed to ...
21374 (vec_extractv4hihi): ... this.
21375 (vec_initv4hi): Renamed to ...
21376 (vec_initv4hihi): ... this.
21377 (vec_extractv8qi): Renamed to ...
21378 (vec_extractv8qiqi): ... this.
21379 (vec_initv8qi): Renamed to ...
21380 (vec_initv8qiqi): ... this.
21381 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21382 (vec_init<mode>): Renamed to ...
21383 (vec_init<mode><VEC_base_l>): ... this.
21384 (vec_extract<mode>): Renamed to ...
21385 (vec_extract<mode><VEC_base_l>): ... this.
21386 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21387 (vec_initv2sfsf): ... this.
21388 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21389 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21390 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21391 element mode after mode in gen_vec_init* calls.
21392 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21393 (vec_init<mode><Vel>): ... this.
21394 (vec_extract<mode>): Renamed to ...
21395 (vec_extract<mode><Vel>): ... this.
21396 * config/aarch64/iterators.md (Vel): New mode attribute.
21397 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21398 Add element mode after mode in gen_vec_extract* calls.
21399 * config/s390/vector.md (non_vec_l): New mode attribute.
21400 (vec_extract<mode>): Renamed to ...
21401 (vec_extract<mode><non_vec_l>): ... this.
21402 (vec_init<mode>): Renamed to ...
21403 (vec_init<mode><non_vec_l>): ... this.
21404 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21405 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21407 * config/arm/iterators.md (V_elem_l): New mode attribute.
21408 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21409 (vec_extract<mode><V_elem_l>): ... this.
21410 (vec_extractv2di): Renamed to ...
21411 (vec_extractv2didi): ... this.
21412 (vec_init<mode>): Renamed to ...
21413 (vec_init<mode><V_elem_l>): ... this.
21414 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21415 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21416 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21417 Add element mode after gen_vec_extract* calls.
21418 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21419 (vec_init<mode><unitmode>): ... this.
21420 (vec_extract<mode>): Renamed to ...
21421 (vec_extract<mode><unitmode>): ... this.
21422 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21423 (vec_init<mode><unitmode>): ... this.
21424 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21425 (vec_initv2sfsf): ... this.
21426 (vec_extractv2sf): Renamed to ...
21427 (vec_extractv2sfsf): ... this.
21428 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21429 Add element mode after gen_vec_extract* calls.
21430 * config/mips/mips.md (unitmode): New mode iterator.
21431 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21432 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21433 * config/spu/spu.md (inner_l): New mode attribute.
21434 (vec_init<mode>): Renamed to ...
21435 (vec_init<mode><inner_l>): ... this.
21436 (vec_extract<mode>): Renamed to ...
21437 (vec_extract<mode><inner_l>): ... this.
21438 * config/sparc/sparc.md (veltmode): New mode iterator.
21439 (vec_init<VMALL:mode>): Renamed to ...
21440 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
21441 * config/ia64/vect.md (vec_initv2si): Renamed to ...
21442 (vec_initv2sisi): ... this.
21443 (vec_initv2sf): Renamed to ...
21444 (vec_initv2sfsf): ... this.
21445 (vec_extractv2sf): Renamed to ...
21446 (vec_extractv2sfsf): ... this.
21447 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
21448 (vec_init<mode>): Renamed to ...
21449 (vec_init<mode><VEC_base_l>): ... this.
21450 (vec_extract<mode>): Renamed to ...
21451 (vec_extract<mode><VEC_base_l>): ... this.
21452 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
21453 (vec_initv2sfsf): ... this.
21454 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
21455 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21456 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
21457 gen_vec_init* calls.
21459 2017-08-01 Richard Biener <rguenther@suse.de>
21461 PR tree-optimization/81297
21462 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
21463 TREE_OVERFLOW from INTEGER_CSTs.
21465 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
21467 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
21469 2017-07-31 Carl Love <cel@us.ibm.com>
21471 * config/rs6000/rs6000-c: Add support for built-in functions
21472 vector signed char vec_xl_be (signed long long, signed char *);
21473 vector unsigned char vec_xl_be (signed long long, unsigned char *);
21474 vector signed int vec_xl_be (signed long long, signed int *);
21475 vector unsigned int vec_xl_be (signed long long, unsigned int *);
21476 vector signed long long vec_xl_be (signed long long, signed long long *);
21477 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
21478 vector signed short vec_xl_be (signed long long, signed short *);
21479 vector unsigned short vec_xl_be (signed long long, unsigned short *);
21480 vector double vec_xl_be (signed long long, double *);
21481 vector float vec_xl_be (signed long long, float *);
21482 * config/rs6000/altivec.h (vec_xl_be): Add #define.
21483 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
21484 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
21486 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
21487 (altivec_expand_builtin): Add switch statement to call
21488 altivec_expand_xl_be for each builtin.
21489 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
21490 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
21491 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
21492 __builtin_vsx_le_be_v16qi.
21493 * doc/extend.texi: Update the built-in documentation file for the
21494 new built-in functions.
21496 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21499 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
21501 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21503 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21505 * config.gcc: Add z14.
21506 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
21507 CPU model numbers for z13s and z14.
21508 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
21510 * config/s390/s390-opts.h (enum processor_type): Rename
21511 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21512 * config/s390/s390.c (processor_table): Add field for CPU name to
21513 be passed to Binutils.
21514 (s390_asm_output_machine_for_arch): Use the new field in
21515 processor_table for Binutils.
21516 (s390_expand_builtin): Replace arch12 with z14.
21517 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
21518 (s390_get_sched_attrmask): Likewise.
21519 (s390_get_unit_mask): Likewise.
21520 * config/s390/s390.opt: Add z14 to processor_type enum.
21522 2017-07-31 Martin Jambor <mjambor@suse.cz>
21525 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
21526 regardless of optimization level.
21528 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
21529 Martin Liska <mliska@suse.cz>
21531 * predict.def: Remove old comment and adjust probability.
21532 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
21533 PREDICT statements.
21535 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
21538 * config/i386/i386.c (ix86_function_naked): New function.
21539 (ix86_can_use_return_insn_p): Return false for naked functions.
21540 (ix86_expand_prologue): Skip prologue for naked functions.
21541 (ix86_expand_epilogue): Skip epilogue for naked functions
21542 and emit trap instruction.
21543 (ix86_warn_func_return): New function.
21544 (ix86_attribute_table): Add "naked" attribute specification.
21545 (TARGET_WARN_FUNC_RETURN): Define.
21546 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
21548 2017-07-31 Martin Liska <mliska@suse.cz>
21550 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
21551 (dump_gimple_bb_header): Always dump BB info.
21552 (pp_cfg_jump): Do not append info about BB when dumping a jump.
21554 2017-07-31 Martin Liska <mliska@suse.cz>
21557 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
21558 also with current_function_decl non-null equality.
21560 2017-07-31 Jakub Jelinek <jakub@redhat.com>
21563 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
21564 change type to the element type, instead add eltype variable and
21565 use it where we are interested in the element type.
21567 PR tree-optimization/81603
21568 * ipa-polymorphic-call.c
21569 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
21570 offset arithmetic in offset_int, bail out if the resulting bit offset
21571 doesn't fit into shwi.
21573 2017-07-31 Martin Liska <mliska@suse.cz>
21575 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
21576 (gimplify_save_expr): Fix comment.
21578 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
21581 * config/i386/i386.c (ix86_function_arg): Update arguments for
21583 (ix86_compute_frame_layout): Set the initial stack offset to
21584 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
21585 INCOMING_FRAME_SP_OFFSET.
21586 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
21587 stack before exception handler returns.
21588 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
21589 the 'ERROR_CODE' for exception handler.
21591 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
21593 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
21594 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
21595 (ASM_OUTPUT_REG_POP): Ditto.
21596 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
21597 instead of asm_fprintf to output pure string.
21599 2017-07-29 Jakub Jelinek <jakub@redhat.com>
21601 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
21602 to imported_module_or_decl hook.
21603 (debug_nothing_tree_tree_tree_bool): Remove.
21604 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
21605 * debug.c (do_nothing_debug_hooks): Use
21606 debug_nothing_tree_tree_tree_bool_bool instead of
21607 debug_nothing_tree_tree_tree_bool.
21608 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21609 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
21610 * sdbout.c (sdb_debug_hooks): Likewise.
21611 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
21612 (gen_namespace_die): Add DW_AT_export_symbols attribute if
21614 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
21615 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
21616 attribute, don't add anything.
21618 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21620 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
21621 (fold_build2_stat_loc): Likewise.
21622 (fold_build3_stat_loc): Likewise.
21623 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
21624 (fold_build1_loc): Remove macro.
21625 (fold_build2_loc): Likewise.
21626 (fold_build3_loc): Likewise.
21628 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21630 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
21631 (gimple_build_debug_bind_source_stat): Likewise.
21632 * gimple.h (gimple_build_debug_bind): Remove macro.
21633 (gimple_build_debug_bind_source): Likewise.
21635 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21637 * bitmap.c (bitmap_alloc): Adjust.
21638 (bitmap_gc_alloc): Likewise.
21639 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
21641 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21643 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
21644 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
21645 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
21646 (bitmap_gc_alloc_stat): Likewise.
21647 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
21649 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21651 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
21652 * rtl.h (shallow_copy_rtx): Remove macro.
21654 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21656 * emit-rtl.c (gen_raw_REG): Adjust.
21657 * gengenrtl.c (gendef): Likewise.
21658 * rtl.c (rtx_alloc_stat): Remove _stat from name.
21659 * rtl.h (rtx_alloc): Remove macro.
21661 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21663 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
21664 (build_tree_list_stat): Likewise.
21665 * tree.h (build_tree_list): Remove macro.
21666 (build_tree_list_vec): Likewise.
21668 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21670 * tree.c (make_vector_stat): Remove _stat from name.
21671 (build_vector_stat): Likewise.
21672 * tree.h (make_vector_stat): Remove macro.
21673 (build_vector_stat): Likewise.
21675 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21677 * tree.h (build_var_debug_value): Remove prototype.
21679 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21681 * tree.c (tree_cons_stat): Remove _stat from name.
21682 * tree.h (tree_cons): Remove macro.
21684 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21686 * tree.c (build_vl_exp_stat): Remove _stat from name.
21687 * tree.h (build_vl_exp): Remove macro.
21689 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21691 * tree.c (build_decl_stat): Remove _stat from name.
21692 * tree.h (build_decl): Remove macro.
21694 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21696 * gimple.c (gimple_build_with_ops_stat): Adjust.
21697 (gimple_alloc_stat): Remove _stat from name.
21698 * gimple.h (gimple_alloc): Remove macro.
21700 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21702 * tree.c (make_tree_vec_stat): Remove _stat from name.
21703 (grow_tree_vec_stat): Likewise.
21704 * tree.h (make_tree_vec_stat): Adjust prototype.
21705 (grow_tree_vec_stat): Likewise.
21706 (make_tree_vec): Remove macro.
21707 (grow_tree_vec): Likewise.
21709 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21711 * fold-const.c (fold_build1_stat_loc): Adjust.
21712 (fold_build2_stat_loc): Likewise.
21713 (fold_build3_stat_loc): Likewise.
21714 * tree.c (build0_stat): Remove _stat from name.
21715 (build1_stat): Likewise.
21716 (build2_stat): Likewise.
21717 (build3_stat): Likewise.
21718 (build4_stat): Likewise.
21719 (build5_stat): Likewise.
21720 * tree.h (build1_loc): Remove macro, and rename _stat function
21722 (build2_loc): Likewise.
21723 (build3_loc): Likewise.
21724 (build4_loc): Likewise.
21725 (build5_loc): Likewise.
21727 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21729 * tree.c (make_int_cst_stat): Remove _stat from name.
21730 * tree.h (make_int_cst_stat): Adjust prototype.
21731 (make_int_cst): Remove macro.
21733 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21735 * tree.c (make_tre_binfo_stat): Remove _stat from name.
21736 * tree.h (make_tree_binfo_stat): Adjust prototype.
21737 (make_tree_binfo): Remove.
21739 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21741 * tree.c (copy_node_stat): Rename to copy_node.
21742 (build_distinct_type_copy): Adjust.
21743 * tree.h (copy_node_stat): Adjust prototype.
21744 (copy_node): Remove macro.
21746 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21748 * tree.c (make_node_stat): rename to make_node.
21749 (build_tree_list_stat): Adjust.
21750 (build0_stat): Likewise.
21751 (build2_stat): Likewise.
21752 (build3_stat): Likewise.
21753 (build4_stat): Likewise.
21754 (build5_stat): Likewise.
21755 (build_decl_stat): Likewise.
21756 * tree.h (make_node_stat): Adjust prototype.
21757 (make_node): remove macro.
21759 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
21761 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
21762 (PPC_FEATURE2_SCV): Likewise.
21763 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
21765 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21767 * config/aarch64/aarch64.c
21768 (aarch64_internal_mov_immediate): Add new special pattern.
21769 * config/aarch64/aarch64.md (*movdi_aarch64):
21770 Add reg/32bit const mov case.
21772 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21773 Richard Sandiford <richard.sandiford@linaro.org>
21775 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21776 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21777 Add integer and movi cases.
21778 (movi-split-hf-df-sf split, fp16): New.
21779 (enabled): Added TARGET_FP_F16INST.
21780 * config/aarch64/iterators.md (GPF_HF): New.
21781 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21783 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21785 * config/aarch64/aarch64.c
21786 (aarch64_simd_container_mode): Add prototype.
21787 (aarch64_expand_mov_immediate): Add HI support.
21788 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21789 (aarch64_can_const_movi_rtx_p): New.
21790 (aarch64_preferred_reload_class):
21791 Remove restrictions of using FP registers for certain SIMD operations.
21792 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21793 (aarch64_valid_floating_const): Add integer move validation.
21794 (aarch64_simd_imm_scalar_p): Remove.
21795 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21796 (aarch64_legitimate_constant_p): Expand list of supported cases.
21797 * config/aarch64/aarch64-protos.h
21798 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21799 (aarch64_reinterpret_float_as_int): New.
21800 (aarch64_simd_imm_scalar_p): Remove.
21801 * config/aarch64/constraints.md (Uvi): New.
21802 (Dd): Split into Ds and new Dd.
21803 * config/aarch64/aarch64.md (*movsi_aarch64):
21805 (*movdi_aarch64): Add SIMD mov case.
21807 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21809 * tree-predcom.c: (struct chain): Handle store-store chain in which
21810 stores for elimination only store loop invariant values.
21811 (execute_pred_commoning_chain): Ditto.
21812 (prepare_initializers_chain_store_elim): Ditto.
21813 (prepare_finalizers): Ditto.
21814 (is_inv_store_elimination_chain): New function.
21815 (initialize_root_vars_store_elim_1): New function.
21817 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21819 * tree-predcom.c: Revise general description of the pass.
21820 (enum chain_type): New enum type for store elimination.
21821 (struct chain): New field supporting store elimination.
21822 (struct component): Ditto.
21823 (dump_chain): Dump store-stores chain.
21824 (release_chain): Release resources.
21825 (split_data_refs_to_components): Compute and create component
21826 contains only stores for elimination.
21827 (get_chain_last_ref_at): New function.
21828 (make_invariant_chain): Initialization.
21829 (make_rooted_chain): Specify chain type in parameter and record it.
21830 (add_looparound_copies): Skip for store-stores chain.
21831 (determine_roots_comp): Compute type of chain and pass it to
21833 (initialize_root_vars_store_elim_2): New function.
21834 (finalize_eliminated_stores): New function.
21835 (remove_stmt): Handle store for elimination.
21836 (execute_pred_commoning_chain): Execute predictive commoning on
21837 store-store chains.
21838 (determine_unroll_factor): Skip unroll for store-stores chain.
21839 (prepare_initializers_chain_store_elim): New function.
21840 (prepare_initializers_chain): Hanlde store-store chain.
21841 (prepare_finalizers_chain, prepare_finalizers): New function.
21842 (tree_predictive_commoning_loop): Return integer value indicating
21843 if loop is unrolled or lcssa form is corrupted.
21844 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21846 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21848 * tree-predcom.c (initialize_root): Delete.
21849 (execute_pred_commoning_chain): Initialize root vars and replace
21850 reference of non-combined chain directly, rather than call above
21853 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21855 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21856 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21858 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21860 * tree-predcom.c (struct chain): New field init_seq.
21861 (release_chain): Release init_seq.
21862 (prepare_initializers_chain): Record intialization stmts in above
21864 (insert_init_seqs): New function.
21865 (tree_predictive_commoning_loop): Call insert_init_seqs.
21867 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21869 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21871 2017-07-28 Richard Biener <rguenther@suse.de>
21873 * match.pd: Remove superfluous :c.
21874 * genmatch.c (simplify::id): Add member.
21875 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21877 (current_id): New global.
21878 (dt_node::parent): Move from ...
21879 (dt_operand::parent): ... here. Add for_id member.
21880 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21881 (decision_tree::find_node): Relax order requirement when
21882 merging DT_TRUE nodes to ones inbetween the current simplify
21883 and the one we try to merge with. Add diagnostic whenever
21884 we need to enforce pattern order by not merging.
21885 (decision_tree::insert): Set current_id.
21886 (decision_tree::print_node): Dump parent node and for_id.
21887 (parser::last_id): Add member.
21888 (parser::push_simplify): Assign unique id.
21889 (parser::parser): Initialize last_id.
21891 2017-07-28 Martin Liska <mliska@suse.cz>
21894 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21895 gimple_build_debug_bind.
21897 2017-07-28 Richard Biener <rguenther@suse.de>
21899 PR tree-optimization/81502
21900 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21902 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21904 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21905 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21906 for BIT_FIELD_REF args.
21907 * fold-const.c (make_bit_field_ref): Likewise.
21908 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21910 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21913 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21914 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21915 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21916 Or it into SANITIZER_UNDEFINED.
21917 * ubsan.c: Include gimple-fold.h and varasm.h.
21918 (ubsan_expand_ptr_ifn): New function.
21919 (instrument_pointer_overflow): New function.
21920 (maybe_instrument_pointer_overflow): New function.
21921 (instrument_object_size): Formatting fix.
21922 (pass_ubsan::execute): Call instrument_pointer_overflow
21923 and maybe_instrument_pointer_overflow.
21924 * internal-fn.c (expand_UBSAN_PTR): New function.
21925 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21926 * sanitizer.def (__ubsan_handle_pointer_overflow,
21927 __ubsan_handle_pointer_overflow_abort): New builtins.
21928 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21929 * internal-fn.def (UBSAN_PTR): New internal function.
21930 * opts.c (sanitizer_opts): Add pointer-overflow.
21931 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21932 * fold-const.c (build_range_check): Compute pointer range check in
21933 integral type if pointer arithmetics would be needed. Formatting
21936 2017-07-28 Martin Liska <mliska@suse.cz>
21939 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21940 parameters that are of a variable-length.
21942 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21944 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21946 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21947 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21948 (CRT_CALL_STATIC_FUNCTION): Likewise.
21949 (ASM_DEFAULT_SPEC): New define.
21950 (ASM_SPEC32): Likewise.
21951 (ASM_SPEC64): Likewise.
21952 (ASM_SPEC_COMMON): Likewise.
21953 (ASM_SPEC): Likewise.
21954 (INVALID_64BIT): Likewise.
21955 (LINK_OS_DEFAULT_SPEC): Likewise.
21956 (LINK_OS_SPEC32): Likewise.
21957 (LINK_OS_SPEC64): Likewise.
21958 (POWERPC_LINUX): Likewise.
21959 (PTRDIFF_TYPE): Likewise.
21960 (RESTORE_FP_PREFIX): Likewise.
21961 (RESTORE_FP_SUFFIX): Likewise.
21962 (SAVE_FP_PREFIX): Likewise.
21963 (SAVE_FP_SUFFIX): Likewise.
21964 (SIZE_TYPE): Likewise.
21965 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21966 (TARGET_64BIT): Likewise.
21967 (TARGET_64BIT): Likewise.
21968 (TARGET_AIX): Likewise.
21969 (WCHAR_TYPE_SIZE): Likewise.
21970 (WCHAR_TYPE): Undefine.
21971 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21972 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21973 (CPP_OS_RTEMS_SPEC): Delete.
21974 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21975 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21977 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21979 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21981 PR tree-optimization/81578
21982 * tree-parloops.c (build_new_reduction): Bail out if
21983 reduction_code isn't one of the standard OpenMP reductions.
21984 Move the details printing after that decision.
21986 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21988 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21989 related to reload_in_progress.
21990 (splat_input_operand): Likewise.
21991 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21993 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21995 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21996 (TARGET_INSTANTIATE_DECLS): Likewise.
21997 (legitimate_indexed_address_p): Delete reload_in_progress code.
21998 (rs6000_debug_legitimate_address_p): Likewise.
21999 (rs6000_eliminate_indexed_memrefs): Likewise.
22000 (rs6000_emit_le_vsx_store): Likewise.
22001 (rs6000_emit_move_si_sf_subreg): Likewise.
22002 (rs6000_emit_move): Likewise.
22003 (register_to_reg_type): Likewise.
22004 (rs6000_pre_atomic_barrier): Likewise.
22005 (rs6000_machopic_legitimize_pic_address): Likewise.
22006 (rs6000_allocate_stack_temp): Likewise.
22007 (rs6000_address_for_fpconvert): Likewise.
22008 (rs6000_address_for_altivec): Likewise.
22009 (rs6000_secondary_memory_needed_rtx): Delete function.
22010 (rs6000_check_sdmode): Likewise.
22011 (rs6000_alloc_sdmode_stack_slot): Likewise.
22012 (rs6000_instantiate_decls): Likewise.
22013 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22014 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22015 Delete reload_in_progress.
22016 (*vec_reload_and_plus_<mptrsize>): Likewise.
22017 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22018 (vsx_div_v2di): Likewise.
22019 (vsx_udiv_v2di): Likewise.
22021 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22023 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22024 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22025 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22026 (rs6000_debug_reg_global): Delete print of LRA status.
22027 (rs6000_option_override_internal): Delete dead LRA related code.
22028 (rs6000_lra_p): Delete function.
22029 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22031 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22033 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22034 * config/riscv/rtems.h: New file.
22036 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22037 Sudakshina Das <sudi.das@arm.com>
22039 * config/aarch64/aarch64.md
22040 (define_split for and<mode>3nr_compare): Move
22041 non aarch64_logical_operand to a register.
22042 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22043 register immediate operand to a register.
22044 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22046 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22048 PR middle-end/81564
22049 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22051 2017-07-27 Richard Biener <rguenther@suse.de>
22053 PR tree-optimization/81573
22054 PR tree-optimization/81494
22055 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22056 multi defuse cycle case.
22058 2017-07-27 Richard Biener <rguenther@suse.de>
22060 PR tree-optimization/81571
22061 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22064 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22066 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22067 earlier and only if MASK_FPU is set. Adjust formatting.
22069 2017-07-27 Martin Liska <mliska@suse.cz>
22071 * opt-functions.awk: Add validation of value of Init.
22072 * optc-gen.awk: Pass new argument.
22074 2017-07-27 Martin Liska <mliska@suse.cz>
22076 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22077 Fix wrong condition.
22079 2017-07-27 Martin Liska <mliska@suse.cz>
22081 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22082 BBs and edges seen by autoFDO.
22084 2017-07-27 Richard Biener <rguenther@suse.de>
22086 PR tree-optimization/81502
22087 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22088 with incompatible but same sized type.
22089 (execute_update_addresses_taken): Likewise.
22091 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22093 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22094 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22096 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22099 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22100 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22101 Change s_operand to memory_operand.
22103 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22105 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22106 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22107 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22108 Emit instructions rather than returning an expression. Handle TFmode
22109 and KFmode by casting to TImode.
22110 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22111 (rs6000_emit_le_vsx_store): Likewise.
22112 * config/rs6000/vsx.md (VSX_TI): New iterator.
22113 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22114 (*vsx_le_undo_permute_<mode>): Likewise.
22115 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22116 emit the split sequence.
22117 (*vsx_le_perm_store_<mode>): Likewise.
22119 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22121 PR tree-optimization/81555
22122 PR tree-optimization/81556
22123 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22124 if true, force CHANGED for the recursive invocation.
22125 (reassociate_bb): Remember original length of ops array, pass
22126 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22128 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22129 attributes for noipa attribute. For naked attribute use
22130 lookup_attribute first before lookup_attribute_spec.
22131 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22133 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22134 for functions with noipa attribute.
22135 (cgraph_externally_visible_p): Return true for functions with noipa
22137 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22138 for functions with noipa attribute.
22139 * doc/extend.texi: Document noipa function attribute.
22140 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22141 also for functions with noipa attribute.
22142 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22144 2017-07-26 Andrew Pinski <apinski@cavium.com>
22146 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22147 vec_unalign_load_cost and vec_unalign_store_cost.
22149 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22151 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22152 -mvsx-small-integer option.
22153 (ISA_3_0_MASKS_IEEE): Likewise.
22154 (OTHER_VSX_VECTOR_MASKS): Likewise.
22155 (POWERPC_MASKS): Likewise.
22156 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22157 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22158 code, only testing for DImode being allowed in non-VSX floating
22160 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22161 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22163 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22164 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22165 TARGET_P8_VECTOR test.
22166 (rs6000_secondary_reload_simple_move): Likewise.
22167 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22168 since TARGET_P9_VECTOR was already tested.
22169 (rs6000_opt_masks): Remove -mvsx-small-integer.
22170 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22171 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22173 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22174 test for TARGET_VEXTRACTUB was used, and that uses
22176 (p9 extract splitter): Likewise.
22177 (vsx_extract_<mode>_di_p9): Likewise.
22178 (vsx_extract_<mode>_store_p9): Likewise.
22179 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22180 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22181 the elimination of TARGET_VSX_SMALL_INTEGER.
22182 (vsx_extract_<mode>_p8): Likewise.
22183 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22184 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22185 (vsx_set_<mode>_p9): Likewise.
22186 (vsx_set_v4sf_p9): Likewise.
22187 (vsx_set_v4sf_p9_zero): Likewise.
22188 (vsx_insert_extract_v4sf_p9): Likewise.
22189 (vsx_insert_extract_v4sf_p9_2): Likewise.
22190 * config/rs6000/rs6000.md (sign extend splitter): Change
22191 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22192 (floatsi<mode>2_lfiwax_mem): Likewise.
22193 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22194 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22195 since a test for TARGET_P9_VECTOR was used.
22196 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22197 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22198 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22199 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22200 TARGET_P8_VECTOR test.
22201 (fix_trunc<mode>si2_stfiwx): Likewise.
22202 (fix_trunc<mode>si2_internal): Likewise.
22203 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22204 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22206 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22207 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22208 TARGET_P8_VECTOR test.
22209 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22210 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22211 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22213 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22214 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22215 since a test for TARGET_P9_VECTOR was used.
22216 (splitter for loading small constants): Likewise.
22218 2017-07-26 Andrew Pinski <apinski@cavium.com>
22220 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22223 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22226 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22227 (fp_valid_at): Likewise.
22229 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22231 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22232 (qdf24xx_addrcost_table): Likewise.
22233 (cortexa57_tunings): Update to use generic_branch_cost.
22234 (cortexa72_tunings): Likewise.
22235 (cortexa73_tunings): Likewise.
22236 (qdf24xx_tunings): Likewise.
22238 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22240 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22241 (thunderx2t99_branch_cost): Likewise.
22242 (cortexa35_tunings): Update to use generic_branch_cost.
22243 (cortexa53_tunings): Likewise.
22244 (cortexa57_tunings): Likewise.
22245 (cortexa72_tunings): Likewise.
22246 (cortexa73_tunings): Likewise.
22247 (thunderx2t99_tunings): Likewise.
22249 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22251 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22252 (sparc_option_override): Honour MASK_FSMULD.
22253 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22254 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22255 * config/sparc/sparc.opt (mfsmuld): New option.
22256 * doc/invoke.texi (mfsmuld): Document option.
22258 2017-07-26 Marek Polacek <polacek@redhat.com>
22260 PR middle-end/70992
22261 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22263 2017-07-26 Richard Biener <rguenther@suse.de>
22265 * gimple-match-head.c (do_valueize): Return OP if valueize
22267 (get_def): New helper to get at the def stmt of a SSA name
22268 if valueize allows.
22269 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22270 do_valueize to get at the def stmt.
22271 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22273 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22275 PR middle-end/46932
22276 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22278 2017-07-26 Martin Liska <mliska@suse.cz>
22281 * function.c (expand_function_start): Make expansion of
22282 nonlocal_goto_save_area after parm_birth_insn.
22284 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22286 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22287 from all CPU target flags enable members.
22289 2017-07-26 Richard Biener <rguenther@suse.de>
22291 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22292 (decision_tree::gen): Make 'type' const.
22293 (write_predicate): Likewise.
22295 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22297 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22298 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22299 (rs6000_option_override_internal): Likewise.
22300 (rs6000_expand_vector_set): Likewise.
22301 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22302 (TARGET_UPPER_REGS_SF): Likewise.
22303 (TARGET_UPPER_REGS_DI): Likewise.
22304 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22305 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22306 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22307 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22308 (Splitters for DI constants in Altivec registers): Likewise.
22309 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22310 (vsx_set_v4sf_p9): Likewise.
22311 (vsx_set_v4sf_p9_zero): Likewise.
22312 (vsx_insert_extract_v4sf_p9): Likewise.
22313 (vsx_insert_extract_v4sf_p9_2): Likewise.
22315 2017-07-25 Carl Love <cel@us.ibm.com>
22317 * doc/extend.texi: Update the built-in documentation file for the
22318 existing built-in functions
22319 vector signed char vec_cnttz (vector signed char);
22320 vector unsigned char vec_cnttz (vector unsigned char);
22321 vector signed short vec_cnttz (vector signed short);
22322 vector unsigned short vec_cnttz (vector unsigned short);
22323 vector signed int vec_cnttz (vector signed int);
22324 vector unsigned int vec_cnttz (vector unsigned int);
22325 vector signed long long vec_cnttz (vector signed long long);
22326 vector unsigned long long vec_cnttz (vector unsigned long long);
22328 2017-07-25 Andrew Pinski <apinski@cavium.com>
22330 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22331 accesses where the use is for the first operand of a BIT_INSERT.
22333 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22336 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22337 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22339 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22341 * config/i386/gstabs.h: Delete.
22342 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22344 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22346 * config/i386/i386.c (ix86_decompose_address): Do not check for
22347 register RTX when looking at index_reg or base_reg.
22348 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22350 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22352 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22353 to update EH info here.
22355 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22357 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22359 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22361 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22363 2017-07-25 Torsten Duwe <duwe@suse.de>
22365 * common.opt: Introduce -fpatchable-function-entry
22366 command line option, and its variables function_entry_patch_area_size
22367 and function_entry_patch_area_start.
22368 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22369 including a two-value parser.
22370 * target.def (print_patchable_function_entry): New target hook.
22371 * targhooks.h (default_print_patchable_function_entry): New function.
22372 * targhooks.c (default_print_patchable_function_entry): Likewise.
22373 * toplev.c (process_options): Switch off IPA-RA if
22374 patchable function entries are being generated.
22375 * varasm.c (assemble_start_function): Look at the
22376 patchable-function-entry command line switch and current
22377 function attributes and maybe generate NOP instructions by
22378 calling the print_patchable_function_entry hook.
22379 * doc/extend.texi: Document patchable_function_entry attribute.
22380 * doc/invoke.texi: Document -fpatchable_function_entry
22381 command line option.
22382 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22384 * doc/tm.texi: Re-generate.
22386 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22389 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22390 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22392 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22394 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22396 2017-07-25 Richard Biener <rguenther@suse.de>
22398 PR tree-optimization/81455
22399 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22400 not walk in cycles when looking for guards.
22402 2017-07-25 Richard Biener <rguenther@suse.de>
22404 PR tree-optimization/81529
22405 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22406 when optimizing backedge uses.
22408 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22410 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22412 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22413 to dl_section_ref. On AIX, append an expression to subtract
22414 the size of the section length to dl_section_ref.
22416 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22418 * configure.ac: If any of the config.* scripts fail, exit 1.
22419 * configure: Regenerate.
22421 2017-07-25 Richard Biener <rguenther@suse.de>
22423 PR middle-end/81546
22424 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22425 of immediate uses, be more verbose on errors.
22427 2017-07-25 Richard Biener <rguenther@suse.de>
22429 PR tree-optimization/81510
22430 * tree-vect-loop.c (vect_is_simple_reduction): When the
22431 reduction stmt is not inside the loop bail out.
22433 2017-07-25 Richard Biener <rguenther@suse.de>
22435 PR tree-optimization/81303
22436 * tree-vect-loop-manip.c (vect_loop_versioning): Build
22437 profitability check against LOOP_VINFO_NITERSM1.
22439 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22441 * domwalk.c (cmp_bb_postorder): Simplify.
22442 (sort_bbs_postorder): New function. Use it...
22443 (dom_walker::walk): ...here to optimize common cases.
22445 2017-07-25 Martin Liska <mliska@suse.cz>
22448 * ipa-visibility.c (function_and_variable_visibility): Make the
22449 redirection just on target that supports aliasing.
22450 Fix GNU coding style.
22452 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
22455 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
22457 * config/arm/rtems.h: Likewise.
22458 * config/bfin/rtems.h: Likewise.
22459 * config/i386/rtemself.h: Likewise.
22460 * config/lm32/rtems.h: Likewise.
22461 * config/m32c/rtems.h: Likewise.
22462 * config/m68k/rtemself.h: Likewise.
22463 * config/microblaze/rtems.h: Likewise.
22464 * config/mips/rtems.h: Likewise.
22465 * config/moxie/rtems.h: Likewise.
22466 * config/nios2/rtems.h: Likewise.
22467 * config/powerpcspe/rtems.h: Likewise.
22468 * config/rs6000/rtems.h: Likewise.
22469 * config/rtems.h: Likewise.
22470 * config/sh/rtems.h: Likewise.
22471 * config/sh/rtemself.h: Likewise.
22472 * config/sparc/rtemself.h: Likewise.
22474 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22477 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
22478 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
22479 * tree-ssa-structalias.c (alias_get_name): Same.
22481 2017-07-25 Bin Cheng <bin.cheng@arm.com>
22484 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
22485 instructions if no du chain is found.
22487 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
22489 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
22491 2017-07-25 Richard Biener <rguenther@suse.de>
22493 PR middle-end/81505
22494 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
22497 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22499 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
22500 upper-regs options.
22501 (ISA_2_7_MASKS_SERVER): Likewise.
22502 (ISA_3_0_MASKS_IEEE): Likewise.
22503 (OTHER_P8_VECTOR_MASKS): Likewise.
22504 (OTHER_VSX_VECTOR_MASKS): Likewise.
22505 (POWERPC_MASKS): Likewise.
22506 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
22507 duplicate list of options.
22508 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
22509 explicit -mupper-regs options.
22510 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
22511 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
22512 alias for -mupper-regs-df.
22513 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
22514 (rs6000_init_hard_regno_mode_ok): Likewise.
22515 (rs6000_option_override_internal): Likewise.
22516 (rs6000_opt_masks): Likewise.
22517 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
22518 options in terms of whether -mvsx or -mpower8-vector was used.
22519 (TARGET_UPPER_REGS_DI): Likewise.
22520 (TARGET_UPPER_REGS_SF): Likewise.
22521 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
22522 -mupper-regs-* options.
22524 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22526 * passes.c (emergency_dump_function): Print some empty lines and a
22527 header before the RTL dump.
22529 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
22531 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
22533 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
22536 * config/aarch64/aarch64.c (aarch64_classify_symbol):
22537 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
22539 2017-07-24 Carl Love <cel@us.ibm.com>
22541 * config/rs6000/rs6000-c.c: Add support for built-in functions
22542 vector float vec_extract_fp32_from_shorth (vector unsigned short);
22543 vector float vec_extract_fp32_from_shortl (vector unsigned short);
22544 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
22545 vec_extract_fp_from_shortl): Add defines for the two builtins.
22546 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
22547 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
22549 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
22550 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
22551 * doc/extend.texi: Update the built-in documentation file for the
22552 new built-in function.
22554 2017-07-24 Jakub Jelinek <jakub@redhat.com>
22557 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
22559 * doc/generic.texi: Likewise.
22560 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22561 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22563 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
22565 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
22566 (aarch64_mls_elt_merge<mode>): Likewise.
22568 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
22570 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22571 having __cxa_atexit.
22573 2017-07-23 Michael Collison <michael.collison@arm.com>
22575 * config/arm/arm.c (arm_option_override): Deprecate
22576 use of -mstructure-size-boundary.
22577 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
22578 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
22580 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22583 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22584 Reduce cost estimate for direct moves.
22586 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
22589 * config/i386/i386.c (ix86_option_override_internal): Disable
22590 BMI, BMI2 and TBM instructions for -m16.
22592 2017-07-21 Carl Love <cel@us.ibm.com>
22594 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22595 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22596 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22597 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22598 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22599 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22600 VMULOSW): New enum "unspec" values.
22601 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22602 altivec_vmulosw): New patterns.
22603 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22604 VMULOSW): Add definitions.
22606 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
22608 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
22609 (qdf24xx): Likewise.
22610 * config/aarch64/aarch64-options-extensions.def (rdma); New.
22611 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
22612 (AARCH64_FL_V8_1): Renumber.
22613 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
22614 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
22615 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
22616 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
22617 rdma to feature modifiers list.
22619 2017-07-21 Yury Gribov <tetra2005@gmail.com>
22621 PR middle-end/56727
22622 * ipa-visibility (function_and_variable_visibility): Convert
22623 recursive PLT call to direct call if appropriate.
22625 2017-07-21 Andrew Pinski <apinski@cavium.com>
22627 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
22628 operand 1 to see if the types precision matches.
22629 * fold-const.c (operand_equal_p): Likewise.
22631 2017-07-21 Richard Biener <rguenther@suse.de>
22633 PR tree-optimization/81303
22634 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
22635 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
22636 (vect_peeling_hash_get_lowest_cost): Adjust.
22637 (vect_enhance_data_refs_alignment): Likewise. Use
22638 vect_get_peeling_costs_all_drs to compute the penalty for no
22639 peeling to match up costs.
22641 2017-07-21 Richard Biener <rguenther@suse.de>
22643 PR tree-optimization/81500
22644 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
22645 we didn't identify a reduction path.
22647 2017-07-21 Tom de Vries <tom@codesourcery.com>
22648 Cesar Philippidis <cesar@codesourcery.com>
22650 PR gcov-profile/81442
22651 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
22654 2017-07-21 Tom de Vries <tom@codesourcery.com>
22657 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
22659 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
22660 nvptx_override_options_after_change.
22662 2017-07-21 Ulrich Drepper <drepper@redhat.com>
22664 * dwarf2out.c (output_file_names): Avoid double testing for
22665 dwarf_version >= 5.
22667 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
22669 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
22671 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22673 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
22675 (try_crossjump_to_edge): Do not punt on partitioned functions.
22677 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
22679 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
22680 Put all BBs reachable only via paths crossing cold region to cold
22682 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
22684 2016-07-21 Richard Biener <rguenther@suse.de>
22686 PR tree-optimization/81303
22687 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
22688 into account prologue and epilogue iterations when raising
22689 min_profitable_iters to sth at least covering one vector iteration.
22691 2017-07-21 Tamar Christina <tamar.christina@arm.com>
22693 * config/arm/arm.c (arm_test_cpu_arch_dat):
22696 2017-07-20 Nathan Sidwell <nathan@acm.org>
22698 Remove TYPE_METHODS.
22699 * tree.h (TYPE_METHODS): Delete.
22700 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
22701 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
22702 (dbxout_type_methods): Scan TYPE_FIELDS.
22703 (dbxout_type): Don't check TYPE_METHODS here.
22704 * function.c (use_register_for_decl): Always ignore register for
22705 class types when not optimizing.
22706 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
22707 * tree.c (free_lang_data_in_type): Stitch out member functions and
22708 templates from TYPE_FIELDS.
22709 (build_distinct_type_copy, verify_type_variant,
22710 verify_type): Member fns are on TYPE_FIELDS.
22711 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
22712 * tree-pretty-print.c (dump_generic_node): Likewise.
22714 2017-07-20 Jakub Jelinek <jakub@redhat.com>
22717 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
22718 V2TImode and V4TImode.
22719 (ix86_expand_vector_extract): Likewise.
22720 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
22721 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
22722 (ssescalarmode): Handle V4TImode and V2TImode.
22723 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
22724 (*vec_extractv2ti, *vec_extractv4ti): New insns.
22725 (VEXTRACTI128_MODE): New mode iterator.
22726 (splitter for *vec_extractv?ti first element): New.
22727 (VEC_INIT_MODE): New mode iterator.
22728 (vec_init<mode>): Consolidate 3 expanders into one using
22729 VEC_INIT_MODE mode iterator.
22731 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22733 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
22734 non_spilled_static_chain_regno_p.
22736 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
22738 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
22740 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
22742 * bb-reorder.c (connect_traces): Allow copying of blocks within
22745 2017-07-20 Richard Biener <rguenther@suse.de>
22747 * gimple.h (gimple_phi_result): Add gphi * overload.
22748 (gimple_phi_result_ptr): Likewise.
22749 (gimple_phi_arg): Likewise. Adjust index assert to only
22750 allow actual argument accesses rather than all slots available
22752 (gimple_phi_arg_def): Add gphi * overload.
22753 * tree-phinodes.c (make_phi_node): Initialize only actual
22755 (resize_phi_node): Clear memory not covered by old node,
22756 do not initialize excess argument slots.
22757 (reserve_phi_args_for_new_edge): Initialize new argument slot
22760 2017-07-20 Bin Cheng <bin.cheng@arm.com>
22762 PR tree-optimization/81388
22764 2016-07-21 Bin Cheng <bin.cheng@arm.com>
22766 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
22767 by removing computation of may_be_zero.
22769 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22770 Tom de Vries <tom@codesourcery.com>
22772 PR middle-end/81030
22773 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22774 when gimple level profile disagrees with what RTL expander did.
22776 2017-07-20 Richard Biener <rguenther@suse.de>
22778 PR tree-optimization/61171
22779 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22780 (vect_analyze_stmt): Add slp instance parameter.
22781 (vectorizable_reduction): Likewise.
22782 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22783 (vect_is_simple_reduction): Deal with chains not detected
22784 as SLP reduction chain, specifically not properly associated
22785 chains containing a mix of plus/minus.
22786 (get_reduction_op): Remove.
22787 (get_initial_defs_for_reduction): Simplify, pass in whether
22788 this is a reduction chain, pass in the SLP node for the PHIs.
22789 (vect_create_epilog_for_reduction): Get the SLP instance as
22791 (vectorizable_reduction): Get the SLP instance as arg.
22792 During analysis remember the SLP node with the PHIs in the
22793 instance. Simplify getting at the vectorized reduction PHIs.
22794 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22795 through SLP instance.
22796 (vect_slp_analyze_operations): Likewise.
22797 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22798 (vect_transform_stmt): Likewise.
22800 2017-07-20 Tom de Vries <tom@codesourcery.com>
22802 PR tree-optimization/81489
22803 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22804 read of phi arg location to before loop that modifies phi.
22806 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22808 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22811 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22813 PR middle-end/81331
22814 * except.c (execute): Fix ordering issue.
22816 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22818 PR rtl-optimization/81423
22819 * combine.c (make_compound_operation_int): Don't try to optimize
22820 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22822 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22824 PR rtl-optimization/81423
22825 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22826 with a constant that is -1 in the truncated to mode.
22828 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22830 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22831 (determine_unlikely_bbs): ... here.
22832 * predict.h (propagate_unlikely_bbs_forward): Declare.
22833 * cfgexpand.c (pass_expand::execute): Use it.
22834 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22836 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22837 propagate_unlikely_bbs_forward.
22839 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22841 PR middle-end/81331
22842 * except.c (maybe_add_nop_after_section_switch): New function.
22845 2017-07-19 Tom de Vries <tom@codesourcery.com>
22847 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22849 2017-07-19 Tom de Vries <tom@codesourcery.com>
22851 * gimple.h (gimple_phi_arg): Make assert more strict.
22853 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22855 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22856 * config/rs6000/mmintrin.h: New file.
22857 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22859 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22861 PR tree-optimization/81346
22862 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22864 2017-07-19 Tom de Vries <tom@codesourcery.com>
22866 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22868 2017-07-19 Tom de Vries <tom@codesourcery.com>
22870 * config/nvptx/nvptx-modes.def: Add V2DImode.
22871 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22872 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22873 (nvptx_output_mov_insn): Handle lack of mov.b128.
22874 (nvptx_print_operand): Handle 'H' and 'L' codes.
22875 (nvptx_vector_mode_supported): Allow V2DImode.
22876 (nvptx_preferred_simd_mode): New function.
22877 (nvptx_data_alignment): New function.
22878 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22879 nvptx_preferred_simd_mode.
22880 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22882 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22884 2017-07-19 Tom de Vries <tom@codesourcery.com>
22886 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22887 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22888 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22889 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22890 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22891 (mov<VECIM>_insn): New define_insn.
22892 (define_expand "mov<VECIM>): New define_expand.
22894 2017-07-19 Tom de Vries <tom@codesourcery.com>
22896 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22898 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22900 PR tree-optimization/81346
22901 * fold-const.h (fold_div_compare, range_check_type): Declare.
22902 * fold-const.c (range_check_type): New function.
22903 (build_range_check): Use range_check_type.
22904 (fold_div_compare): No longer static, rewritten into
22905 a match.pd helper function.
22906 (fold_comparison): Don't call fold_div_compare here.
22907 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22908 as helper function.
22910 2017-07-19 Nathan Sidwell <nathan@acm.org>
22912 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22913 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22914 * tree.c (find_decls_types_r, verify_type): Use
22915 TYPE_{MIN,MAX}_VALUE_RAW.
22916 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22917 (hash_tree): Likewise.
22918 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22920 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22923 2017-07-18 Tom de Vries <tom@codesourcery.com>
22925 PR middle-end/81464
22926 * omp-expand.c (expand_omp_for_static_chunk): Handle
22927 equal-argument loop exit phi.
22929 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22932 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22933 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22934 operand 2 predicate.
22935 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22936 operand 2 predicate.
22937 (ror,rol -> rorx splitters): Use const_int_operand as
22938 operand 2 predicate.
22940 2017-06-18 Richard Biener <rguenther@suse.de>
22942 PR tree-optimization/81410
22943 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22944 the gap in the ! slp_perm SLP case after each group.
22946 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22948 PR middle-end/81463
22949 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22952 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22954 PR middle-end/81462
22955 * predict.c (set_even_probabilities): Cleanup; do not affect
22956 probabilities that are already known.
22957 (combine_predictions_for_bb): Call even when count is set.
22959 2017-07-18 Nathan Sidwell <nathan@acm.org>
22961 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22964 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22967 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22968 optimization for loop niter analysis.
22970 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22973 * config/avr/avr.c (avr_optimize_casesi): Don't use
22974 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22976 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22978 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22979 body_cost_vec from _vect_peel_extended_info.
22980 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22981 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22984 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22986 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22988 2017-07-18 Richard Biener <rguenther@suse.de>
22990 PR tree-optimization/80620
22991 PR tree-optimization/81403
22992 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22993 info when re-using a VN table entry.
22995 2017-07-18 Richard Biener <rguenther@suse.de>
22997 PR tree-optimization/81418
22998 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22999 vectype_in. Verify that with lane-reducing reduction operations
23000 we have a single def-use cycle.
23002 2017-07-17 Carl Love <cel@us.ibm.com>
23004 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23006 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23007 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23008 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23009 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23010 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23011 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23012 VMULOSW): New enum "unspec" values.
23013 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23014 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23015 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23016 altivec_vmulosw): New patterns.
23017 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23018 VMULOSW): Add definitions.
23020 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23022 * config/alpha/alpha.c: Include predict.h.
23024 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23026 * tree-vrp.c (compare_assert_loc): Fix comparison function
23027 to return predictable results.
23029 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23031 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23033 (subdi3): Likewise.
23034 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23035 * doc/invoke.texi (mexpand-adddi): Update text.
23037 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23039 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23040 that also clobbers the CC register. The old expand code is moved
23042 (*arc_clzsi2): ... here.
23043 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23044 the CC register. The old expand code is moved to ...
23045 (arc_ctzsi2): ... here.
23047 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23049 * config/arc/arc.opt (mindexed-loads): Use initial value
23050 TARGET_INDEXED_LOADS_DEFAULT.
23051 (mauto-modify-reg): Use initial value
23052 TARGET_AUTO_MODIFY_REG_DEFAULT.
23053 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23054 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23055 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23056 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23058 2017-07-17 Martin Liska <mliska@suse.cz>
23061 * opts.c (finish_options): Do not allow -fgnu-tm
23062 w/ -fsanitize={kernel-,}address. Say sorry.
23064 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23067 * tree-loop-distribution.c (classify_partition): Only assert on
23068 numer of iterations.
23069 (merge_dep_scc_partitions): Delete prameter. Update function call.
23070 (distribute_loop): Remove code handling loop with unknown niters.
23071 (pass_loop_distribution::execute): Skip loop with unknown niters.
23073 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23076 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23077 function sort_partitions_by_post_order.
23079 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23081 PR tree-optimization/81374
23082 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23083 the max index of basic blocks, rather than number of basic blocks.
23085 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23087 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23089 (arc_legitimate_pic_operand_p): Likewise.
23090 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23092 (arc_needs_pcl_p): Likewise.
23093 (arc_legitimate_pc_offset_p): Likewise.
23094 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23095 function is also used in constrains.md.
23096 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23097 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23098 PLUS. Only return true/false in known cases, otherwise assert.
23099 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23100 is already called in arc_legitimate_constant_p.
23101 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23103 (LEGITIMATE_PIC_OPERAND_P): Use
23104 arc_raw_symbolic_reference_mentioned_p function.
23105 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23110 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23111 Andrew Burgess <andrew.burgess@embecosm.com>
23113 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23114 (arc_return_address_register): New function.
23115 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23116 (arc_handle_fndecl_attribute): Add naked attribute.
23117 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23118 (TARGET_WARN_FUNC_RETURN): Likewise.
23119 (arc_allocate_stack_slots_for_args): New function.
23120 (arc_warn_func_return): Likewise.
23121 (machine_function): Change type fn_type.
23122 (arc_compute_function_type): Consider new naked function type,
23123 change function return type.
23124 (arc_must_save_register): Adapt to handle new
23125 arc_compute_function_type's return type.
23126 (arc_expand_prologue): Likewise.
23127 (arc_expand_epilogue): Likewise.
23128 (arc_return_address_regs): Delete.
23129 (arc_return_address_register): New function.
23130 (arc_epilogue_uses): Use above function.
23131 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23132 (arc_function_type): Change encoding, add naked type.
23133 (ARC_INTERRUPT_P): Change to handle the new encoding.
23134 (ARC_FAST_INTERRUPT_P): Likewise.
23135 (ARC_NORMAL_P): Define.
23136 (ARC_NAKED_P): Likewise.
23137 (arc_compute_function_type): Delete prototype.
23138 * config/arc/arc.md (in_ret_delay_slot): Use
23139 arc_return_address_register function.
23140 (simple_return): Likewise.
23141 (p_return_i): Likewise.
23143 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23145 PR tree-optimization/81428
23146 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23147 can't be built for those types.
23149 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23151 Remove stuff dead since r239246.
23153 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23154 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23155 (avr_inform_devices): Remove dead stuff.
23157 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23159 * config/arm/arm_neon.h: Fix softp typo.
23161 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23163 PR tree-optimization/81365
23164 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23165 aggregate moves onto bb predecessor edges, make sure there are no
23166 loads that could alias the lhs in between the start of bb and the
23169 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23172 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23173 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23174 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23176 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23178 PR tree-optimization/81396
23179 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23180 (init_symbolic_number): Initialize it to 1.
23181 (perform_symbolic_merge): Add n_ops from both operands into the new
23183 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23184 without base_addr as useless if they need more than one operation.
23185 (bswap_replace): Handle !bswap case for NULL base_addr.
23187 2017-07-17 Tom de Vries <tom@codesourcery.com>
23190 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23193 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23195 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23196 conditional builtin define __FIX_LEON3FT_B2BST.
23198 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23200 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23201 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23204 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23206 PR rtl-optimization/81424
23207 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23208 to remove potential trapping from operands if -fnon-call-exceptions.
23210 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23212 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23213 profile_proability for scalling.
23214 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23216 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23218 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23220 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23222 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23223 fixpoint arithmetics.
23225 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23227 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23228 fixpoint arithmetics.
23230 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23232 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23233 fixpoint arithmetics.
23235 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23237 * profile-count.h (profile_probability::from_reg_br_prob_note,
23238 profile_probability::to_reg_br_prob_note): New functions.
23239 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23240 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23241 * predict.c (probability_reliable_p): Update.
23242 (edge_probability_reliable_p): Update.
23243 (br_prob_note_reliable_p): Update.
23244 (invert_br_probabilities): Update.
23245 (add_reg_br_prob_note): New function.
23246 (combine_predictions_for_insn): Update.
23247 * asan.c (asan_clear_shadow): Update.
23248 * cfgbuild.c (compute_outgoing_frequencies): Update.
23249 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23250 (update_br_prob_note): Update.
23251 (rtl_verify_edges): Update.
23252 (purge_dead_edges): Update.
23253 (fixup_reorder_chain): Update.
23254 * emit-rtl.c (try_split): Update.
23255 * ifcvt.c (cond_exec_process_insns): Update.
23256 (cond_exec_process_if_block): Update.
23257 (dead_or_predicable): Update.
23258 * internal-fn.c (expand_addsub_overflow): Update.
23259 (expand_neg_overflow): Update.
23260 (expand_mul_overflow): Update.
23261 * loop-doloop.c (doloop_modify): Update.
23262 * loop-unroll.c (compare_and_jump_seq): Update.
23263 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23264 * predict.h: Update.
23265 * reorg.c (mostly_true_jump): Update.
23267 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23268 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23269 * config/arc/arc.c: (emit_unlikely_jump): Update.
23270 * config/arm/arm.c: (emit_unlikely_jump): Update.
23271 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23272 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23273 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23274 (ix86_print_operand): Update.
23275 (ix86_split_fp_branch): Update.
23276 (predict_jump): Update.
23277 * config/ia64/ia64.c (ia64_print_operand): Update.
23278 * config/mmix/mmix.c (mmix_print_operand): Update.
23279 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23280 (rs6000_expand_split_stack_prologue): Update.
23281 * config/rs6000/rs6000.c: Update.
23282 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23283 (s390_expand_vec_movstr): Update.
23284 (s390_expand_cs_tdsi): Update.
23285 (s390_expand_split_stack_prologue): Update.
23286 * config/sh/sh.c (sh_print_operand): Update.
23287 (expand_cbranchsi4): Update.
23288 (expand_cbranchdi4): Update.
23289 * config/sparc/sparc.c (output_v9branch): Update.
23290 * config/spu/spu.c (get_branch_target): Update.
23291 (ea_load_store_inline): Update.
23292 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23293 * config/tilepro/tilepro.c: Update.
23295 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23297 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23298 (gimplify_save_expr): Likewise.
23300 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23302 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23304 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23306 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23308 * timevar.def (TV_IPA_FNSUMMARY): Define.
23310 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23312 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23313 to back store errata sensitive sequence from being generated.
23314 (sqrtdf2_fix): Likewise.
23316 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23318 * tree-ssa-threadupdate.c (compute_path_counts,
23319 update_joiner_offpath_counts): Use profile_probability.
23321 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23324 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23326 * config/arm/arm-c.c (arm_cpu_builtins): Define
23327 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23329 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23331 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23332 array entries to represent __ieee128 versions of the
23333 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23334 scalar_extract_sig, and scalar_insert_exp built-in functions.
23335 (altivec_resolve_overloaded_builtin): Add special case handling
23336 for the __builtin_scalar_insert_exp function, as represented by
23337 the P9V_BUILTIN_VEC_VSIEDP constant.
23338 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23339 exponent support for __ieee128 argument.
23340 (VSESQP): Add scalar extract signature support for __ieee128
23342 (VSTDCNQP): Add scalar test negative support for __ieee128
23344 (VSIEQP): Add scalar insert exponent support for __int128 argument
23345 with __ieee128 result.
23346 (VSIEQPF): Add scalar insert exponent support for __ieee128
23347 argument with __ieee128 result.
23348 (VSTDCQP): Add scalar test data class support for __ieee128
23350 (VSTDCNQP): Add overload support for scalar test negative with
23351 __ieee128 argument.
23352 (VSTDCQP): Add overload support for scalar test data class
23353 __ieee128 argument.
23354 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23355 UNSPEC_VSX_SXSIGDP.
23356 (UNSPEC_VSX_SIEXPQP): New constant.
23357 (xsxexpqp): New insn for VSX scalar extract exponent quad
23359 (xsxsigqp): New insn for VSX scalar extract significand quad
23361 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23362 precision with floating point argument.
23363 (xststdcqp): New expand for VSX scalar test data class quad
23365 (xststdcnegqp): New expand for VSX scalar test negative quad
23367 (xststdcqp): New insn to match expansions for VSX scalar test data
23368 class quad precision and VSX scalar test negative quad precision.
23369 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23370 special case operand checking to enforce that second operand of
23371 VSX scalar test data class with quad precision argument is a 7-bit
23373 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23374 prototypes and descriptions of __ieee128 versions of
23375 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23376 scalar_test_data_class, and scalar_test_neg built-in functions.
23378 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23380 PR tree-optimization/81162
23381 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23382 replace a negate with an add.
23384 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23386 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23388 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23390 * config/arm/arm-c.c (arm_cpu_builtins): Define
23391 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23393 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23395 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23396 (armv8-r): Set ARM Cortex-R52 as default CPU.
23397 * config/arm/arm-tables.opt: Regenerate.
23398 * config/arm/arm-tune.md: Regenerate.
23399 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23401 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23402 extension for -mcpu=cortex-r52.
23404 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23406 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23407 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23408 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23409 (fp-armv8): Define it as FP_ARMv8 only.
23410 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23411 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23413 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23414 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23415 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23416 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23417 than TARGET_FPU_ARMV8.
23418 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23419 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23420 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23422 * config/arm/neon.md (neon_vrint): Likewise.
23423 (neon_vcvt): Likewise.
23424 (neon_<fmaxmin_op><mode>): Likewise.
23425 (<fmaxmin><mode>3): Likewise.
23426 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23427 * config/arm/predicates.md (arm_cond_move_operator): Check against
23428 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23430 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23432 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23433 to top of function.
23435 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23437 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
23438 loop in comment with memset.
23440 2017-07-14 Martin Liska <mliska@suse.cz>
23442 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
23443 * dwarf2out.c (is_java): Remove the function.
23444 (output_pubname): Remove usage of the function.
23445 (lower_bound_default): Remove usage of DW_LANG_Java.
23446 (gen_compile_unit_die): Likewise.
23447 * gcc.c: Remove compiler defaults for .java and .zip files.
23448 * gimple-expr.c (remove_suffix): Change as there's no longer
23449 extension than 4-letter one.
23450 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
23451 (gimplify_save_expr): Likewise.
23452 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
23453 as it's possible even for other languages than Java.
23454 * langhooks.h (struct lang_hooks): Remove Java from a comment.
23455 * lto-opts.c (lto_write_options): Remove reference to Java.
23456 * opts.c (strip_off_ending): Update file extension handling.
23457 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
23458 * tree-eh.c (lower_resx): Likewise.
23459 * tree.c (free_lang_data_in_type): Remove dead code.
23460 (find_decls_types_r): Likewise.
23461 (build_common_builtin_nodes): Remove Java from a comment.
23462 (verify_type): Remove dead code.
23463 * varasm.c (assemble_external): Remove Java from a comment.
23465 2017-07-14 Martin Liska <mliska@suse.cz>
23467 * opts.c (finish_options): Add quotes.
23468 (common_handle_option): Likewise.
23470 2017-07-14 Martin Liska <mliska@suse.cz>
23472 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
23473 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
23474 Remove N_SO_PASCAL.
23475 * dwarf2out.c (lower_bound_default): Do not handle
23477 (gen_compile_unit_die): Likewise.
23478 * gcc.c: Remove default extension binding for GNU Pascal.
23479 * stmt.c: Remove Pascal language from a comment.
23480 * xcoffout.c: Likewise.
23482 2017-07-13 David Malcolm <dmalcolm@redhat.com>
23485 * diagnostic-show-locus.c (fixit_cmp): New function.
23486 (layout::layout): Sort m_fixit_hints.
23487 (column_range::column_range): Assert that the values are valid.
23488 (struct char_span): New struct.
23489 (correction::overwrite): New method.
23490 (struct source_line): New struct.
23491 (line_corrections::add_hint): Add assertions. Reimplement memcpy
23492 calls in terms of classes source_line and char_span, and
23493 correction::overwrite.
23494 (selftest::test_overlapped_fixit_printing_2): New function.
23495 (selftest::diagnostic_show_locus_c_tests): Call it.
23497 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
23499 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
23500 early if there is no lhs.
23502 2017-07-13 Martin Liska <mliska@suse.cz>
23504 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
23505 (gen_reference_type_die): Likewise.
23506 * stor-layout.c: Remove Pascal-related comment.
23508 2017-07-13 Martin Liska <mliska@suse.cz>
23510 * opts.c (finish_options): Add quotes to error messages.
23511 (parse_sanitizer_options): Likewise.
23513 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
23515 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
23517 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
23519 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
23521 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23523 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
23525 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
23527 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23530 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
23531 provides the hardware capability bits, define the macro
23532 __BUILTIN_CPU_SUPPORTS__.
23533 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
23534 if GLIBC does not provide the hardware capability bits. Add a
23535 gcc_unreachable call if the built-in cpu function is neither
23536 __builtin_cpu_is nor __builtin_cpu_supports.
23537 (rs6000_get_function_versions_dispatcher): Change the warning
23538 that an old GLIBC is used which does not export the capability
23539 bits to be an error.
23540 * doc/extend.texi (target_clones attribute): Document the
23541 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
23542 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
23543 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
23544 the macros defined by GCC if the newer GLIBC is available.
23546 2017-07-12 Jeff Law <law@redhat.com>
23548 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
23549 remaining includes slightly.
23550 * config/riscv/riscv-builtins.c: Include profile-count.h.
23552 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23555 * config/avr/avr.c (avr_set_current_function): In diagnostic
23556 messages: Quote keywords and (parts of) identifiers.
23557 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
23560 2017-07-12 Carl Love <cel@us.ibm.com>
23562 * config/rs6000/rs6000-c.c: Add support for built-in functions
23563 vector bool char vec_revb (vector bool char);
23564 vector bool short vec_revb (vector short char);
23565 vector bool int vec_revb (vector bool int);
23566 vector bool long long vec_revb (vector bool long long);
23567 * doc/extend.texi: Update the built-in documentation file for the
23568 new built-in functions.
23570 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23572 * config/s390/s390.md: Remove movcc splitter.
23574 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23576 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
23577 load/store on condition.
23579 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
23582 * config/avr/avr.c (avr_encode_section_info)
23583 [progmem && !TREE_READONLY]: Error if progmem object needs
23586 2017-07-11 Michael Collison <michael.collison@arm.com>
23588 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
23591 2017-07-11 Carl Love <cel@us.ibm.com>
23593 * config/rs6000/rs6000-c.c: Add support for builtins
23594 vector unsigned int vec_parity_lsbb (vector signed int);
23595 vector unsigned int vec_parity_lsbb (vector unsigned int);
23596 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
23597 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
23598 vector unsigned long long vec_parity_lsbb (vector signed long long);
23599 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
23600 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
23601 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
23602 * doc/extend.texi: Update the built-in documentation file for the
23603 new built-in functions.
23605 2017-07-11 David Malcolm <dmalcolm@redhat.com>
23607 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23608 (layout::m_primary_loc): New field.
23609 (layout::layout): Initialize new field. Move location filtering
23610 logic from here to...
23611 (layout::maybe_add_location_range): ...this new method. Add
23612 support for filtering to just the lines already specified by other
23614 (layout::will_show_line_p): New method.
23615 (gcc_rich_location::add_location_if_nearby): New method.
23616 (selftest::test_add_location_if_nearby): New test function.
23617 (selftest::diagnostic_show_locus_c_tests): Call it.
23618 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
23621 2017-07-11 Tom de Vries <tom@codesourcery.com>
23623 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
23624 (bb_first_real_insn): New function.
23625 (nvptx_single): Add extra initialization of broadcasted condition
23628 2017-07-11 Nathan Sidwell <nathan@acm.org>
23630 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
23632 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
23634 * doc/extend.texi (AVR Function Attributes): Remove weblink to
23635 Binutils doc as TEXI will mess them up.
23636 * doc/invoke.texi (AVR Options): Same here.
23638 2017-07-11 Daniel Cederman <cederman@gaisler.com>
23640 * config/sparc/sparc.opt (mfix-ut700): New option.
23641 (mfix-gr712rc): Likewise.
23642 (sparc_fix_b2bst): New variable.
23643 * doc/invoke.texi (SPARC options): Document them.
23644 (ARM options): Fix warnings.
23645 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
23646 instructions to prevent sequences that can trigger the store-store
23647 errata for certain LEON3FT processors.
23648 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
23649 (sparc_option_override): Set sparc_fix_b2bst appropriately.
23650 * config/sparc/sparc.md (fix_b2bst): New attribute.
23651 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
23653 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
23656 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
23658 (*rsqrtsf2_sse): Ditto.
23660 (div<mode>3): Macroize insn from divdf3 and divsf3
23661 using MODEF mode iterator.
23663 2017-07-10 Martin Sebor <msebor@redhat.com>
23665 PR tree-optimization/80397
23666 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
23667 instead of testing for equality to INTEGER_TYPE.
23669 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
23671 * config.gcc: Remove uclibc from arc target spec.
23673 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
23675 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
23677 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23680 * lto-wrapper.c (remove_option): New function.
23681 (merge_and_complain): Merge PIC/PIE options more realistically.
23683 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23685 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
23689 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
23690 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
23691 * config.in: Regenerate.
23692 * configure: Regenerate.
23693 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
23694 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
23695 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
23696 (TARGET_GASISR_PROLOGUES): ...target mask.
23697 * common/config/avr/avr-common.c
23698 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
23699 Set -mgas-isr-prologues.
23700 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
23701 INSERT_PASS_BEFORE for it.
23702 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
23703 * config/avr/avr.c (avr_option_override)
23704 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
23705 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
23706 (avr_attribute_table) <no_gccisr>: Add new function attribute.
23707 (avr_set_current_function) <is_no_gccisr>: Init machine field.
23708 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
23710 (make_avr_pass_pre_proep): New function.
23711 (emit_push_sfr) <treg>: Add argument to function and use it
23712 instead of TMP_REG.
23713 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
23714 and set machine->gasisr.yes.
23715 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
23716 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
23717 __gcc_isr.n_pushed to .L__stack_usage.
23718 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
23719 (avr_asm_final_postscan_insn): ...this new static function.
23720 * config/avr/avr.h (machine_function)
23721 <is_no_gccisr, use_L__stack_usage>: New fields.
23722 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
23723 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
23724 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
23725 (gasisr, *gasisr): New expander and insn.
23726 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
23727 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
23728 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
23730 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
23732 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
23735 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
23737 Move jump-tables out of .text again.
23740 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
23741 (ASM_OUTPUT_ADDR_VEC): New function.
23742 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
23743 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
23744 INSN_ADDRESSes as asm comment.
23745 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
23746 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
23747 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
23748 * config/avr/avr.md (*tablejump): Adjust comment.
23749 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
23750 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
23752 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
23753 (avr_output_addr_vec): New proto.
23754 (avr_log_t) <insn_addresses>: New field.
23756 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
23759 * config/i386/i386.c (ix86_function_arg_advance): Set
23760 outgoing_args_on_stack to true if there are outgoing arguments
23762 (ix86_function_arg): Likewise.
23763 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
23764 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
23765 * config/i386/i386.h (machine_function): Add
23766 outgoing_args_on_stack.
23768 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23770 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23771 supporting pthreds.
23772 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23774 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23776 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23777 (REAL_H): Remove $(MACHMODE_H).
23778 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23780 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23781 $(MACHMODE_H) and double-int.h.
23782 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23784 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23787 2017-07-07 Andrew Pinski <apinski@cavium.com>
23789 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23790 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23792 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23794 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23795 Add warning if GCC was not configured to link against a GLIBC that
23796 exports the hardware capability bits.
23797 (make_resolver_func): Make resolver function private and not a
23798 COMDAT function. Create the name with clone_function_name instead
23799 of make_unique_name.
23802 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23803 correct operand in doing the split.
23805 2017-07-07 Carl Love <cel@us.ibm.com>
23807 * config/rs6000/rs6000-c: Add support for built-in function
23808 vector unsigned short vec_pack_to_short_fp32 (vector float,
23810 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23811 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23812 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23813 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23814 (convert_4f32_8i16): Add define_expand.
23815 * doc/extend.texi: Update the built-in documentation file for the
23816 new built-in function.
23818 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23820 * config/sparc/m8.md: New file.
23821 * config/sparc/sparc.md: Include m8.md.
23823 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23825 * config/sparc/sparc.opt: New option -mvis4b.
23826 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23827 (sparc_option_override): Handle VIS4B.
23828 (enum sparc_builtins): Define
23829 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23830 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23831 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23832 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23833 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23834 (check_constant_argument): New function.
23835 (sparc_vis_init_builtins): Define builtins
23836 __builtin_vis_dictunpack{8,16,32},
23837 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23838 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23839 __builtin_vis_fpcmpde{8,16,32}shl and
23840 __builtin_vis_fpcmpur{8,16,32}shl.
23841 (sparc_expand_builtin): Check that the constant operands to
23842 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23843 constant and in range.
23844 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23846 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23847 (SPARC_IMM5_P): Likewise.
23848 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23849 (enabled): Handle vis4b.
23850 (UNSPEC_DICTUNPACK): New unspec.
23851 (UNSPEC_FPCMPSHL): Likewise.
23852 (UNSPEC_FPUCMPSHL): Likewise.
23853 (UNSPEC_FPCMPDESHL): Likewise.
23854 (UNSPEC_FPCMPURSHL): Likewise.
23855 (cpu_feature): New CPU feature `vis4b'.
23856 (dictunpack{8,16,32}): New insns.
23857 (FPCSMODE): New mode iterator.
23858 (fpcscond): New code iterator.
23859 (fpcsucond): Likewise.
23860 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23861 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23862 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23863 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23864 * config/sparc/constraints.md: Define constraints `q' for unsigned
23865 2-bit integer constants and `t' for unsigned 5-bit integer
23867 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23869 (imm5_operand_dictunpack16): Likewise.
23870 (imm5_operand_dictunpack32): Likewise.
23871 (imm2_operand): Likewise.
23872 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23873 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23874 ditunpack* and fpcmp*shl builtins.
23876 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23878 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23879 * config.in: Add HAVE_AS_SPARC6 define.
23880 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23882 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23884 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23885 (CPP_CPU_SPEC): Handle m8.
23886 (ASM_CPU_SPEC): Likewise.
23887 * config/sparc/sparc-opts.h (enum processor_type): Add
23889 * config/sparc/sparc.c (m8_costs): New struct.
23890 (sparc_option_override): Handle TARGET_CPU_m8.
23891 (sparc32_initialize_trampoline): Likewise.
23892 (sparc64_initialize_trampoline): Likewise.
23893 (sparc_issue_rate): Likewise.
23894 (sparc_register_move_cost): Likewise.
23895 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23896 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23897 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23898 (CPP_CPU_SPEC): Handle M8.
23899 (ASM_CPU_SPEC): Likewise.
23900 (AS_M8_FLAG): Define.
23901 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23902 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23903 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23905 * configure: Regenerate.
23906 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23909 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23911 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23913 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23914 ("*movdi_insn_sp32"): Do not set v3pipe.
23915 ("*movsi_insn"): Likewise.
23916 ("*movdi_insn_sp64"): Likewise.
23917 ("*movsf_insn"): Likewise.
23918 ("*movdf_insn_sp32"): Likewise.
23919 ("*movdf_insn_sp64"): Likewise.
23920 ("*zero_extendsidi2_insn_sp64"): Likewise.
23921 ("*sign_extendsidi2_insn"): Likewise.
23922 ("*mov<VM32:mode>_insn"): Likewise.
23923 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23924 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23925 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23926 ("<vlop:code><VL:mode>3"): Likewise.
23927 ("*not_<vlop:code><VL:mode>3"): Likewise.
23928 ("*nand<VL:mode>_vis"): Likewise.
23929 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23930 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23931 ("one_cmpl<VL:mode>2"): Likewise.
23932 ("faligndata<VM64:mode>_vis"): Likewise.
23933 ("alignaddrsi_vis"): Likewise.
23934 ("alignaddrdi_vis"): Likweise.
23935 ("alignaddrlsi_vis"): Likewise.
23936 ("alignaddrldi_vis"): Likewise.
23937 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23938 ("bmaskdi_vis"): Likewise.
23939 ("bmasksi_vis"): Likewise.
23940 ("bshuffle<VM64:mode>_vis"): Likewise.
23941 ("cmask8<P:mode>_vis"): Likewise.
23942 ("cmask16<P:mode>_vis"): Likewise.
23943 ("cmask32<P:mode>_vis"): Likewise.
23944 ("pdistn<P:mode>_vis"): Likewise.
23945 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23947 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23949 * config/sparc/sparc.md ("subtype"): New insn attribute.
23950 ("*wrgsr_sp64"): Set insn subtype.
23951 ("*rdgsr_sp64"): Likewise.
23952 ("alignaddrsi_vis"): Likewise.
23953 ("alignaddrdi_vis"): Likewise.
23954 ("alignaddrlsi_vis"): Likewise.
23955 ("alignaddrldi_vis"): Likewise.
23956 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23957 ("fexpand_vis"): Likewise.
23958 ("fpmerge_vis"): Likewise.
23959 ("faligndata<VM64:mode>_vis"): Likewise.
23960 ("bshuffle<VM64:mode>_vis"): Likewise.
23961 ("cmask8<P:mode>_vis"): Likewise.
23962 ("cmask16<P:mode>_vis"): Likewise.
23963 ("cmask32<P:mode>_vis"): Likewise.
23964 ("fchksm16_vis"): Likewise.
23965 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23966 ("fmean16_vis"): Likewise.
23967 ("fp<plusminus_insn>64_vis"): Likewise.
23968 ("<plusminus_insn>v8qi3"): Likewise.
23969 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23970 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23971 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23972 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23973 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23974 ("*movqi_insn"): Likewise.
23975 ("*movhi_insn"): Likewise.
23976 ("*movsi_insn"): Likewise.
23977 ("movsi_pic_gotdata_op"): Likewise.
23978 ("*movdi_insn_sp32"): Likewise.
23979 ("*movdi_insn_sp64"): Likewise.
23980 ("movdi_pic_gotdata_op"): Likewise.
23981 ("*movsf_insn"): Likewise.
23982 ("*movdf_insn_sp32"): Likewise.
23983 ("*movdf_insn_sp64"): Likewise.
23984 ("*zero_extendhisi2_insn"): Likewise.
23985 ("*zero_extendqihi2_insn"): Likewise.
23986 ("*zero_extendqisi2_insn"): Likewise.
23987 ("*zero_extendqidi2_insn"): Likewise.
23988 ("*zero_extendhidi2_insn"): Likewise.
23989 ("*zero_extendsidi2_insn_sp64"): Likewise.
23990 ("ldfsr"): Likewise.
23991 ("prefetch_64"): Likewise.
23992 ("prefetch_32"): Likewise.
23993 ("tie_ld32"): Likewise.
23994 ("tie_ld64"): Likewise.
23995 ("*tldo_ldub_sp32"): Likewise.
23996 ("*tldo_ldub1_sp32"): Likewise.
23997 ("*tldo_ldub2_sp32"): Likewise.
23998 ("*tldo_ldub_sp64"): Likewise.
23999 ("*tldo_ldub1_sp64"): Likewise.
24000 ("*tldo_ldub2_sp64"): Likewise.
24001 ("*tldo_ldub3_sp64"): Likewise.
24002 ("*tldo_lduh_sp32"): Likewise.
24003 ("*tldo_lduh1_sp32"): Likewise.
24004 ("*tldo_lduh_sp64"): Likewise.
24005 ("*tldo_lduh1_sp64"): Likewise.
24006 ("*tldo_lduh2_sp64"): Likewise.
24007 ("*tldo_lduw_sp32"): Likewise.
24008 ("*tldo_lduw_sp64"): Likewise.
24009 ("*tldo_lduw1_sp64"): Likewise.
24010 ("*tldo_ldx_sp64"): Likewise.
24011 ("*mov<VM32:mode>_insn"): Likewise.
24012 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24013 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24015 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24017 * config/sparc/sparc.md ("type"): New insn type viscmp.
24018 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24020 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24021 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24022 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24023 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24025 ("n7_vis_logical_11cycle"): Likewise.
24026 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24027 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24028 * config/sparc/niagara.md ("niag_vis"): Likewise.
24029 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24030 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24032 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24034 * config/sparc/sparc.md: New instruction type `bmask'.
24035 (bmaskdi_vis): Use the `bmask' type.
24036 (bmasksi_vis): Likewise.
24037 * config/sparc/ultra3.md (us3_array): Likewise.
24038 * config/sparc/niagara7.md (n7_array): Likewise.
24039 * config/sparc/niagara4.md (n4_array): Likewise.
24040 * config/sparc/niagara2.md (niag2_vis): Likewise.
24041 (niag3_vis): Likewise.
24042 * config/sparc/niagara.md (niag_vis): Likewise.
24044 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24046 * ipa-comdats.c: Remove optimize check from gate.
24047 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24048 for functions not optimized.
24049 (ipa_fn_summary_read): Skip optimize check.
24050 (ipa_fn_summary_write): Likewise.
24051 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24053 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24055 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24056 (check_callers): Give up on caller which is not optimized.
24057 (inline_small_functions): Likewise.
24058 (ipa_inline): Do not give up when not optimizing.
24059 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24060 away unoptimizes cdtors.
24061 (whole_program_function_and_variable_visibility): Do
24062 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24063 * ipa.c (process_references): Do not check optimize.
24064 (symbol_table::remove_unreachable_nodes): Update optimize check.
24065 (set_writeonly_bit): Update optimize check.
24066 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24067 (pass_ipa_single_use::gate): Remove.
24069 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24071 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24072 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24073 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24074 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24075 permute_load, permute_store, adjust_extract, adjust_splat,
24076 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24077 replace_swap_with_copy, dump_swap_insn_table,
24078 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24079 recombine_lvx_pattern, recombine_stvx_pattern,
24080 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24081 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24082 to file rs6000-p8swap.c.
24083 * config/rs6000/rs6000-p8swap.c: New file.
24084 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24085 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24086 and rs6000*-*-* targets.
24088 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24090 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24092 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24094 * lto-wrapper.c (merge_and_complain): Do not merge
24095 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24096 fsigned_zeros, ftrapping_math, fwrapv.
24097 (append_compiler_options): Do not track these options.
24098 (append_linker_options): Likewie
24100 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24102 * cgraphunit.c (cgraph_node::finalize_function): When
24103 !flag_toplevel_reorde set no_reorder flag.
24104 (varpool_node::finalize_decl): Likewise.
24105 (symbol_table::compile): Drop no toplevel reorder path.
24107 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24109 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24110 edges; zero probability is not better than uninitialized.
24112 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24114 * asan.h (asan_sanitize_allocas_p): Declare.
24115 * asan.c (asan_sanitize_allocas_p): New function.
24116 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24117 (handle_builtin_alloca): Likewise.
24118 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24119 if !asan_sanitize_allocas_p.
24120 * params.def (asan-instrument-allocas): Add new option.
24121 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24122 * opts.c (common_handle_option): Disable allocas sanitization for
24125 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24127 * asan.c: Include gimple-fold.h.
24128 (get_last_alloca_addr): New function.
24129 (handle_builtin_stackrestore): Likewise.
24130 (handle_builtin_alloca): Likewise.
24131 (asan_emit_allocas_unpoison): Likewise.
24132 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24133 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24134 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24135 (instrument_builtin_call): Pass gimple iterator to
24136 get_mem_refs_of_builtin_call.
24137 (last_alloca_addr): New global.
24138 * asan.h (asan_emit_allocas_unpoison): Declare.
24139 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24140 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24141 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24142 if function calls alloca.
24143 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24144 * gimple-fold.h (replace_call_with_value): Declare.
24145 * internal-fn.c: Include asan.h.
24146 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24147 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24149 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24151 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24152 (C_SELFTEST_FLAGS): New.
24153 (CPP_SELFTEST_FLAGS): New.
24154 (SELFTEST_DEPS): New, from deps of s-selftest.
24155 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24156 (CPP_SELFTEST_DEPS): New.
24157 (selftest): Add dependency on s-selftest-c++.
24158 (s-selftest): Rename to...
24159 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24160 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24161 than SELFTEST_FLAGS.
24162 (selftest-gdb): Rename to...
24163 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24165 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24166 (selftest-valgrind): Rename to...
24167 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24169 (selftest-valgrind): Reintroduce as an alias for
24170 selftest-c-valgrind.
24171 (s-selftest-c++): New.
24172 (selftest-c++-gdb): New.
24173 (selftest-c++-valgrind): New.
24175 2017-07-06 Olivier Hainque <hainque@adacore.com>
24177 * gcc.c (process_command): When deciding if undefined variables
24178 should be ignored when processing specs, accept "gcc -v" as well.
24180 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24182 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24183 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24185 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24187 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24188 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24189 * config/arm/arm-tables.opt: Regenerate.
24190 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24192 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24194 2017-07-06 Carl Love <cel@us.ibm.com>
24196 * ChangeLog: Clean up from mid air collision
24198 2017-07-06 Carl Love <cel@us.ibm.com>
24200 * config/rs6000/rs6000-c.c: Add support for built-in functions
24201 vector signed int vec_subc (vector signed int, vector signed int);
24202 vector signed __int128 vec_subc (vector signed __int128,
24203 vector signed __int128);
24204 vector unsigned __int128 vec_subc (vector unsigned __int128,
24205 vector unsigned __int128);
24206 vector signed int vec_sube (vector signed int, vector signed int,
24207 vector signed int);
24208 vector unsigned int vec_sube (vector unsigned int,
24209 vector unsigned int,
24210 vector unsigned int);
24211 vector signed __int128 vec_sube (vector signed __int128,
24212 vector signed __int128,
24213 vector signed__int128);
24214 vector unsigned __int128 vec_sube (vector unsigned __int128,
24215 vector unsigned __int128,
24216 vector unsigned __int128);
24217 vector signed int vec_subec (vector signed int, vector signed int,
24218 vector signed int);
24219 vector unsigned int vec_subec (vector unsigned int,
24220 vector unsigned int,
24221 vector unsigned int);
24222 vector signed __int128 vec_subec (vector signed __int128,
24223 vector signed __int128,
24224 vector signed__int128);
24225 vector unsigned __int128 vec_subec (vector unsigned __int128,
24226 vector unsigned __int128,
24227 vector unsigned __int128);
24228 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24229 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24230 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24231 BU_ALTIVEC_OVERLOAD_X definitions.
24232 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24233 * doc/extend.texi: Update the built-in documentation file for the new
24234 built-in functions.
24236 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24239 * diagnostic-show-locus.c (layout::layout): Use start and finish
24240 spelling location for the start and finish of each range.
24241 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24242 Add unused aspect param.
24243 * input.c (expand_location_1): Add "aspect" param, and use it
24244 to access the correct part of the location.
24245 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24247 (expand_location_to_spelling_point): Likewise.
24248 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24249 param, and pass it to expand_location_1.
24251 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24253 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24254 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24255 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24256 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24257 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24258 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24259 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24260 _mm_maskz_getmant_ss): New intrinsics.
24261 (__builtin_ia32_getexpss128_mask): Changed to ...
24262 __builtin_ia32_getexpss128_round ... this.
24263 (__builtin_ia32_getexpsd128_mask): Changed to ...
24264 __builtin_ia32_getexpsd128_round ... this.
24265 * config/i386/i386-builtin-types.def
24266 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24267 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24268 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24269 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24270 __builtin_ia32_getmantss_mask_round): New builtins.
24271 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24272 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24273 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24274 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24275 * config/i386/sse.md
24276 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24277 avx512f_sgetexp<mode><mask_scalar_name>
24278 <round_saeonly_scalar_name> ... this.
24279 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24280 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24281 vgetexp<ssescalarmodesuffix>
24282 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24283 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24284 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24285 avx512f_vgetmant<mode><mask_scalar_name>
24286 <round_saeonly_scalar_name> ... this.
24287 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24288 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24289 vgetmant<ssescalarmodesuffix>
24290 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24291 %0<mask_scalar_operand4>, %1, %2
24292 <round_saeonly_scalar_mask_op4>, %3} ... this.
24293 * config/i386/subst.md (mask_scalar_operand4,
24294 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24295 round_saeonly_scalar_nimm_predicate): New subst attributes.
24297 2017-07-06 Julia Koval <julia.koval@intel.com>
24299 * config/i386/i386.c (ix86_erase_embedded_rounding):
24300 Remove code for old rounding pattern.
24302 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24304 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24306 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24308 * doc/sourcebuild.texi (Test Directives, Variants of
24309 dg-require-support): Add documentation for dg-require-stack-check.
24311 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24313 * config/i386/subst.md (mask_scalar, round_scalar,
24314 round_saeonly_scalar): New meta-templates.
24315 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24316 round_scalar_mask_operand3, round_scalar_mask_op3,
24317 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24318 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24319 round_saeonly_scalar_constraint,
24320 round_saeonly_scalar_prefix): New subst attribute.
24321 * config/i386/sse.md
24322 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24323 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24324 <round_scalar_name> ... this.
24325 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24326 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24327 <round_scalar_name> ... this.
24328 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24329 <sse>_vm<code><mode>3<mask_scalar_name>
24330 <round_saeonly_scalar_name> ... this.
24331 (v<plusminus_mnemonic><ssescalarmodesuffix>
24332 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24333 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24334 v<plusminus_mnemonic><ssescalarmodesuffix>
24335 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24336 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24337 (v<multdiv_mnemonic><ssescalarmodesuffix>
24338 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24339 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24340 v<multdiv_mnemonic><ssescalarmodesuffix>
24341 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24342 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24343 (v<maxmin_float><ssescalarmodesuffix>
24344 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24345 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24346 v<maxmin_float><ssescalarmodesuffix>
24347 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24348 %0<mask_scalar_operand3>, %1, %<iptr>2
24349 <round_saeonly_scalar_mask_op3>} ... this.
24351 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24353 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24354 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24356 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24357 Alan Hayward <alan.hayward@arm.com>
24358 David Sherwood <david.sherwood@arm.com>
24360 * combine.c (simplify_if_then_else): Remove "enum" before
24362 * compare-elim.c (can_eliminate_compare): Likewise.
24363 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24365 (aarch64_lookup_simd_builtin_type): Likewise.
24366 (aarch64_simd_builtin_type): Likewise.
24367 (aarch64_init_simd_builtin_types): Likewise.
24368 (aarch64_simd_expand_args): Likewise.
24369 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24371 (aarch64_reverse_mask): Likewise.
24372 (aarch64_simd_emit_reg_reg_move): Likewise.
24373 (aarch64_gen_adjusted_ldpstp): Likewise.
24374 (aarch64_ccmp_mode_to_code): Likewise.
24375 (aarch64_operands_ok_for_ldpstp): Likewise.
24376 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24377 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24379 (aarch64_min_divisions_for_recip_mul): Likewise.
24380 (aarch64_reassociation_width): Likewise.
24381 (aarch64_get_condition_code_1): Likewise.
24382 (aarch64_simd_emit_reg_reg_move): Likewise.
24383 (aarch64_simd_attr_length_rglist): Likewise.
24384 (aarch64_reverse_mask): Likewise.
24385 (aarch64_operands_ok_for_ldpstp): Likewise.
24386 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24387 (aarch64_gen_adjusted_ldpstp): Likewise.
24388 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24390 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24391 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24392 (arm_lookup_simd_builtin_type): Likewise.
24393 (arm_simd_builtin_type): Likewise.
24394 (arm_init_simd_builtin_types): Likewise.
24395 (arm_expand_builtin_args): Likewise.
24396 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24397 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24398 (ft32_setup_incoming_varargs): Likewise.
24399 (ft32_function_arg): Likewise.
24400 (ft32_function_arg_advance): Likewise.
24401 (ft32_pass_by_reference): Likewise.
24402 (ft32_arg_partial_bytes): Likewise.
24403 (ft32_valid_pointer_mode): Likewise.
24404 (ft32_addr_space_pointer_mode): Likewise.
24405 (ft32_addr_space_legitimate_address_p): Likewise.
24406 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24408 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24409 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24410 (iamcu_alignment): Likewise.
24411 (canonicalize_vector_int_perm): Likewise.
24412 (ix86_noce_conversion_profitable_p): Likewise.
24413 (ix86_mpx_bound_mode): Likewise.
24414 (ix86_operands_ok_for_move_multiple): Likewise.
24415 * config/microblaze/microblaze-protos.h
24416 (microblaze_expand_conditional_branch_reg): Likewise.
24417 * config/microblaze/microblaze.c
24418 (microblaze_expand_conditional_branch_reg): Likewise.
24419 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24421 (rs6000_reassociation_width): Likewise.
24422 (rs6000_invalid_binary_op): Likewise.
24423 (fusion_p9_p): Likewise.
24424 (emit_fusion_p9_load): Likewise.
24425 (emit_fusion_p9_store): Likewise.
24426 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24428 (riscv_hard_regno_mode_ok_p): Likewise.
24429 (riscv_address_insns): Likewise.
24430 (riscv_split_symbol): Likewise.
24431 (riscv_legitimize_move): Likewise.
24432 (riscv_function_value): Likewise.
24433 (riscv_hard_regno_nregs): Likewise.
24434 (riscv_expand_builtin): Likewise.
24435 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
24436 (riscv_build_integer): Likewise.
24437 (riscv_split_integer): Likewise.
24438 (riscv_legitimate_constant_p): Likewise.
24439 (riscv_cannot_force_const_mem): Likewise.
24440 (riscv_regno_mode_ok_for_base_p): Likewise.
24441 (riscv_valid_base_register_p): Likewise.
24442 (riscv_valid_offset_p): Likewise.
24443 (riscv_valid_lo_sum_p): Likewise.
24444 (riscv_classify_address): Likewise.
24445 (riscv_legitimate_address_p): Likewise.
24446 (riscv_address_insns): Likewise.
24447 (riscv_load_store_insns): Likewise.
24448 (riscv_force_binary): Likewise.
24449 (riscv_split_symbol): Likewise.
24450 (riscv_force_address): Likewise.
24451 (riscv_legitimize_address): Likewise.
24452 (riscv_move_integer): Likewise.
24453 (riscv_legitimize_const_move): Likewise.
24454 (riscv_legitimize_move): Likewise.
24455 (riscv_address_cost): Likewise.
24456 (riscv_subword): Likewise.
24457 (riscv_output_move): Likewise.
24458 (riscv_canonicalize_int_order_test): Likewise.
24459 (riscv_emit_int_order_test): Likewise.
24460 (riscv_function_arg_boundary): Likewise.
24461 (riscv_pass_mode_in_fpr_p): Likewise.
24462 (riscv_pass_fpr_single): Likewise.
24463 (riscv_pass_fpr_pair): Likewise.
24464 (riscv_get_arg_info): Likewise.
24465 (riscv_function_arg): Likewise.
24466 (riscv_function_arg_advance): Likewise.
24467 (riscv_arg_partial_bytes): Likewise.
24468 (riscv_function_value): Likewise.
24469 (riscv_pass_by_reference): Likewise.
24470 (riscv_setup_incoming_varargs): Likewise.
24471 (riscv_print_operand): Likewise.
24472 (riscv_elf_select_rtx_section): Likewise.
24473 (riscv_save_restore_reg): Likewise.
24474 (riscv_for_each_saved_reg): Likewise.
24475 (riscv_register_move_cost): Likewise.
24476 (riscv_hard_regno_mode_ok_p): Likewise.
24477 (riscv_hard_regno_nregs): Likewise.
24478 (riscv_class_max_nregs): Likewise.
24479 (riscv_memory_move_cost): Likewise.
24480 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
24481 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
24482 (rl78_addr_space_address_mode): Likewise.
24483 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
24485 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
24486 (rs6000_reassociation_width): Likewise.
24487 (rs6000_invalid_binary_op): Likewise.
24488 (fusion_p9_p): Likewise.
24489 (emit_fusion_p9_load): Likewise.
24490 (emit_fusion_p9_store): Likewise.
24491 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
24492 (ok_for_simple_move_operands): Likewise.
24493 (ok_for_simple_move_strict_operands): Likewise.
24494 (ok_for_simple_arith_logic_operands): Likewise.
24495 (visium_legitimize_reload_address): Likewise.
24496 (visium_select_cc_mode): Likewise.
24497 (output_cbranch): Likewise.
24498 (visium_split_double_move): Likewise.
24499 (visium_expand_copysign): Likewise.
24500 (visium_expand_int_cstore): Likewise.
24501 (visium_expand_fp_cstore): Likewise.
24502 * config/visium/visium.c (visium_pass_by_reference): Likewise.
24503 (visium_function_arg): Likewise.
24504 (visium_function_arg_advance): Likewise.
24505 (visium_libcall_value): Likewise.
24506 (visium_setup_incoming_varargs): Likewise.
24507 (visium_legitimate_constant_p): Likewise.
24508 (visium_legitimate_address_p): Likewise.
24509 (visium_legitimize_address): Likewise.
24510 (visium_secondary_reload): Likewise.
24511 (visium_register_move_cost): Likewise.
24512 (visium_memory_move_cost): Likewise.
24513 (prepare_move_operands): Likewise.
24514 (ok_for_simple_move_operands): Likewise.
24515 (ok_for_simple_move_strict_operands): Likewise.
24516 (ok_for_simple_arith_logic_operands): Likewise.
24517 (visium_function_value_1): Likewise.
24518 (rtx_ok_for_offset_p): Likewise.
24519 (visium_legitimize_reload_address): Likewise.
24520 (visium_split_double_move): Likewise.
24521 (visium_expand_copysign): Likewise.
24522 (visium_expand_int_cstore): Likewise.
24523 (visium_expand_fp_cstore): Likewise.
24524 (visium_split_cstore): Likewise.
24525 (visium_select_cc_mode): Likewise.
24526 (visium_split_cbranch): Likewise.
24527 (output_cbranch): Likewise.
24528 (visium_print_operand_address): Likewise.
24529 * expmed.c (flip_storage_order): Likewise.
24530 * expmed.h (emit_cstore): Likewise.
24531 (flip_storage_order): Likewise.
24532 * genrecog.c (validate_pattern): Likewise.
24533 * hsa-gen.c (gen_hsa_addr): Likewise.
24534 * internal-fn.c (expand_arith_overflow): Likewise.
24535 * ira-color.c (allocno_copy_cost_saving): Likewise.
24536 * lra-assigns.c (find_hard_regno_for_1): Likewise.
24537 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
24538 (process_invariant_for_inheritance): Likewise.
24539 * lra-eliminations.c (move_plus_up): Likewise.
24540 * omp-low.c (lower_oacc_reductions): Likewise.
24541 * simplify-rtx.c (simplify_subreg): Likewise.
24542 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
24543 (TARGET_CHKP_BOUND_MODE): Likewise..
24544 * targhooks.c (default_chkp_bound_mode): Likewise.
24545 (default_setup_incoming_vararg_bounds): Likewise.
24546 * targhooks.h (default_chkp_bound_mode): Likewise.
24547 (default_setup_incoming_vararg_bounds): Likewise.
24548 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
24549 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
24550 (have_whole_vector_shift): Likewise.
24551 * tree-vect-stmts.c (vectorizable_load): Likewise.
24552 * doc/tm.texi: Regenerate.
24554 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24556 Graceful degrade if Binutils PR21472 is not available.
24559 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
24560 .rodata in flash test fails.
24561 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
24562 * confgure: Regenerate.
24563 * config.in: Regenerate.
24564 * config/avr/avr.c (avr_asm_named_section)
24565 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
24566 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
24567 (avr_asm_init_sections): Same.
24569 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24571 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
24572 (fma<VH:mode>4_intrinsic): Likewise.
24573 (*fmsub<VCVTF:mode>4): Likewise.
24574 (*fmsub<VH:mode>4_intrinsic): Likewise.
24576 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
24579 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
24580 Don't depend on "optimize > 0".
24581 (out_movhi_r_mr, out_movqi_mr_r): Same.
24582 (out_movhi_mr_r, out_movqi_r_mr): Same.
24583 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
24584 io_address_operand on "optimize > 0".
24586 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24588 * tree-loop-distribution.c: Add general explanantion on the pass.
24589 (generate_loops_for_partition): Mark distributed loop.
24590 (pg_add_dependence_edges): New parameter. Handle alias data
24591 dependence specially and record it in the parameter if asked.
24592 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
24593 (init_partition_graph_vertices, add_partition_graph_edge): New.
24594 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
24595 (free_partition_graph_vdata, build_partition_graph): New.
24596 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
24597 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
24598 (data_ref_segment_size, latch_dominated_by_data_ref): New.
24599 (compute_alias_check_pairs, version_loop_by_alias_check): New.
24600 (version_for_distribution_p, finalize_partitions): New.
24601 (distribute_loop): Handle alias data dependence specially. Factor
24602 out loop fusion code as functions and call these functions.
24604 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24606 * tree-loop-distribution.c (classify_partition): New parameter and
24607 better handle reduction statement.
24608 (rdg_build_partitions): Revise comment.
24609 (distribute_loop): Compute statements in all partitions and pass it
24610 to classify_partition.
24612 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24614 * tree-loop-distribution.c (enum partition_type): New.
24615 (struct partition): New field type.
24616 (partition_merge_into): Add parameter. Update partition type.
24617 (data_dep_in_cycle_p, update_type_for_merge): New functions.
24618 (build_rdg_partition_for_vertex): Compute partition type.
24619 (rdg_build_partitions): Dump partition type.
24620 (distribute_loop): Update calls to partition_merge_into.
24622 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24624 * tree-loop-distribution.c (struct ddr_hasher): New.
24625 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
24627 (classify_partition): Call get_data_dependence.
24628 (pg_add_dependence_edges): Ditto.
24629 (distribute_loop): Release data dependence hash table.
24631 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24633 * tree-loop-distribution.c (ref_base_address): Delete.
24634 (similar_memory_accesses): Rename ...
24635 (share_memory_accesses): ... to this. Check if partitions access
24636 the same memory reference.
24637 (distribute_loop): Call share_memory_accesses.
24639 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24641 * tree-loop-distribution.c (struct partition): New field recording
24642 its data reference.
24643 (partition_alloc, partition_free): Init and release data refs.
24644 (partition_merge_into): Merge data refs.
24645 (build_rdg_partition_for_vertex): Collect data refs for partition.
24646 (pg_add_dependence_edges): Change parameters from vector to bitmap.
24648 (distribute_loop): Remve data refs from vertice data of partition
24651 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24653 * tree-loop-distribution.c (params.h): Include header file.
24654 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
24655 (datarefs_vec): New global var.
24656 (create_rdg_vertices): Use datarefs_vec directly.
24657 (free_rdg): Don't free data references.
24658 (build_rdg): Update use. Don't free data references.
24659 (distribute_loop): Compute global variable for data references.
24660 Bail out if there are too many data references.
24662 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24664 * tree-loop-distribution.c (loop_nest): New global var.
24665 (build_rdg): Use loop directly, rather than loop nest.
24666 (pg_add_dependence_edges): Remove loop nest parameter. Use global
24668 (distribute_loop): Compute global variable loop nest. Update use.
24670 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24672 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
24673 (partition_merge_into): New parameter. Dump reason for fusion.
24674 (distribute_loop): Update use of partition_merge_into.
24676 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24678 * tree-loop-distribution.c (bb_top_order_index): New.
24679 (bb_top_order_index_size, bb_top_order_cmp): New.
24680 (stmts_from_loop): Use topological order.
24681 (pass_loop_distribution::execute): Compute and release topological
24682 order for basic blocks.
24684 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24686 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
24689 2017-07-05 Bin Cheng <bin.cheng@arm.com>
24691 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
24692 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
24693 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
24694 * internal-fn.def (LOOP_DIST_ALIAS): New.
24695 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
24696 (fold_loop_internal_call): ... this.
24697 (vect_loop_dist_alias_call): New function.
24698 (set_uid_loop_bbs): Call fold_loop_internal_call.
24699 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
24702 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24705 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
24706 Require dead FLAGS_REG at the beginning of a peephole.
24708 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
24711 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
24712 arguments in the call to __builtin_ia32_sbb_u32.
24713 (_subborrow_u64): Swap _X and _Y arguments in the call to
24714 __builtin_ia32_sbb_u64.
24716 2017-07-04 Jakub Jelinek <jakub@redhat.com>
24719 * tree-vrp.c (compare_assert_loc): Turn into a function template
24720 with stable template parameter. Only test if a->e is NULL,
24721 !a->e == !b->e has been verified already. Use e == NULL or
24722 e != NULL instead of e or ! e tests. If stable is true, don't use
24723 iterative_hash_expr, on the other side allow a or b or both NULL
24724 and sort the NULLs last.
24725 (process_assert_insertions): Sort using compare_assert_loc<false>
24726 instead of compare_assert_loc, later sort using
24727 compare_assert_loc<true> before calling process_assert_insertions_for
24728 in a loop. Use break instead of continue once seen NULL pointer.
24730 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24732 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24733 Cortex-R7 and Cortex-R8 processors.
24735 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24737 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
24738 uninitialized while src is not.
24740 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
24742 * common/config/arm/arm-common.c: Adjust include path for
24744 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
24745 (arm-cpu.h): Create in build directory. Adjust dependency rules.
24746 (arm-cpu-data.h): Likewise.
24747 (arm-cpu-cdata.h): Likewise.
24748 * config/arm/arm-cpu.h: Delete.
24749 * config/arm/arm-cpu-cdata.h: Delete.
24750 * config/arm/arm-cpu-data.h: Delete.
24752 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
24754 * config/arm/arm-cpus.in (cortex-a55): New.
24755 (cortex-a75): Likewise.
24756 (cortex-a75.cortex-a55): Likewise.
24757 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
24759 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
24760 * config/arm/arm-cpu-cdata.h: Regenerate.
24761 * config/arm/arm-cpu-data.h: Regenerate.
24762 * config/arm/arm-cpu.h: Regenerate.
24763 * config/arm/arm-tables.opt: Regenerate.
24764 * config/arm/arm-tune.md: Regenerate.
24766 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24768 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24770 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24772 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24775 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24777 PR tree-optimization/81292
24778 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24779 full_string_p, also call adjust_related_strinfos if the adjustment
24780 is simple, otherwise invalidate related strinfos.
24782 2017-07-04 Martin Liska <mliska@suse.cz>
24785 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24786 newly created variable as DECL_IGNORED_P.
24788 2017-07-04 Martin Liska <mliska@suse.cz>
24791 * ipa-inline.c (inline_small_functions):
24792 Use xstrdup_for_dump.
24794 2017-07-04 Tom de Vries <tom@codesourcery.com>
24796 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24798 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24801 * config/darwin.c (darwin_function_switched_text_sections):
24804 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24806 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24808 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24810 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24812 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24814 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24815 min_profitable_iters, and th as inclusive lower bounds.
24816 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24817 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24818 for min_profitable_iters and min_profitable_estimate.
24819 (vect_transform_loop): Treat th as an inclusive lower bound.
24820 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24822 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24825 * config/darwin.c (darwin_function_switched_text_sections):
24826 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24827 in two pieces, and suppress the use of buf.
24829 2017-07-03 Nathan Sidwell <nathan@acm.org>
24831 * hash-table.h (hash_table_mod1): Fix indentation.
24833 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24835 PR middle-end/81290
24836 * predict.c (force_edge_cold): Be more careful about propagation
24838 * profile-count.h (profile_probability::guessed,
24839 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24841 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24843 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24845 * doc/invoke.texi (rcpc architecture extension): Document it.
24847 2017-07-03 Richard Biener <rguenther@suse.de>
24849 PR tree-optimization/60510
24850 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24851 the scalar reduction PHI and use it.
24852 (vectorizable_reduction): Properly guard the single_defuse_cycle
24853 path for non-SLP reduction chains where we cannot use it.
24854 Rework reduc_def/index and vector type deduction. Rework
24855 vector operand gathering during reduction op code-gen.
24856 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24857 chains dissolve the chain and leave it to non-SLP reduction
24860 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24862 * tree-data-ref.h (dr_alignment): Declare.
24863 * tree-data-ref.c (dr_alignment): New function.
24864 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24865 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24867 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24869 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24871 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24872 and base_misalignment fields.
24873 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24874 * tree-data-ref.c: Include builtins.h.
24875 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24876 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24877 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24878 * tree-vect-data-refs.c: Include tree-cfg.h.
24879 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24880 fields instead of calculating an alignment here.
24881 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24882 innermost_loop_behavior fields.
24884 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24886 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24888 (DR_STEP_ALIGNMENT): New macro.
24889 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24890 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24891 (create_data_ref): Print it.
24892 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24893 to tell whether the step preserves vector (mis)alignment.
24894 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24895 Move the check for an integer step and generalise to all INTEGER_CST.
24896 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24897 Print the outer step alignment.
24899 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24901 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24902 with offset_alignment.
24903 (DR_ALIGNED_TO): Delete.
24904 (DR_OFFSET_ALIGNMENT): New macro.
24905 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24906 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24907 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24908 (create_data_ref): Likewise.
24909 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24910 (vect_analyze_data_refs): Likewise.
24911 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24912 creating dummy innermost behavior.
24914 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24916 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24917 with a "innermost_loop_behavior *" and refeence tree.
24918 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24919 (create_data_ref): Update call accordingly.
24920 * tree-predcom.c (find_looparound_phi): Likewise.
24922 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24924 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24925 fields with dr_wrt_vec_loop.
24926 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24927 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24928 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24929 (vect_dr_behavior): New function.
24930 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24931 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24932 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24933 track whether the step preserves the misalignment.
24934 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24935 Use vect_dr_behavior.
24936 (vect_setup_realignment): Update call accordingly.
24937 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24938 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24939 call to vect_create_addr_base_for_vector_ref.
24940 (vect_create_cond_for_align_checks): Likewise.
24941 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24942 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24943 (vect_recog_mask_conversion_pattern): Likewise.
24944 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24945 (new_stmt_vec_info): Remove redundant zeroing.
24947 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24949 * common/config/arm/arm-common.c (arm_be8_option): New function.
24950 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24951 (ISA_ARMv6): Add isa_bit_be8.
24952 * config/arm/arm.h (arm_be8_option): Add prototype.
24953 (BE8_SPEC_FUNCTION): New define.
24954 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24955 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24956 (mlittle-endian): Similarly.
24957 (mbe8, mbe32): New options.
24958 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24959 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24961 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24963 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24965 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24967 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24968 (cleanup_tree_cfg_bb): Use it.
24969 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24971 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24973 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24976 * loop-doloop.c (add_test): Update profile.
24978 2017-07-03 Martin Liska <mliska@suse.cz>
24981 * sanopt.c (rewrite_usage_of_param): New function.
24982 (sanitize_rewrite_addressable_params): Likewise.
24983 (pass_sanopt::execute): Call rewrite_usage_of_param.
24985 2017-07-03 Richard Biener <rguenther@suse.de>
24987 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24988 back to using VIEW_CONVERT_EXPR.
24990 2017-07-03 Martin Liska <mliska@suse.cz>
24993 * doc/extend.texi: Document when a resolver function is
24994 generated for target_clones.
24996 2017-07-03 Martin Liska <mliska@suse.cz>
24998 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24999 and shadow memory of auto variables which are subject of
25000 use-after-scope sanitization.
25001 (asan_expand_mark_ifn): Add do set only when is_poison.
25003 2016-07-03 Richard Biener <rguenther@suse.de>
25005 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25007 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25008 (vectorizable_reduction): Perform reduction PHI creation when
25009 visiting a reduction PHI and adjust and simplify code generation
25010 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25011 (vect_transform_loop): Visit reduction PHIs.
25012 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25013 defs into the SLP tree.
25014 (vect_build_slp_tree): Reduction defs terminate the recursion.
25015 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25017 (vect_get_vec_defs_for_stmt_copy): Export.
25018 (vect_get_vec_defs): Likewise.
25019 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25021 (vect_get_vec_defs_for_stmt_copy): Declare.
25022 (vect_get_vec_defs): Likewise.
25024 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25026 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25027 parameter with a "loop" parameter and use it instead of the
25028 loop containing DR_STMT. Don't check simple_iv when doing
25029 BB analysis. Describe the two analysis modes in the comment.
25031 2017-07-03 Tom de Vries <tom@codesourcery.com>
25033 PR tree-optimization/69468
25034 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25035 (find_same_succ_bb): Handle ignore_edge_flags.
25037 2017-07-03 Tom de Vries <tom@codesourcery.com>
25039 PR tree-optimization/81192
25040 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25042 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25044 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25046 2017-07-03 Tom de Vries <tom@codesourcery.com>
25048 PR tree-optimization/81192
25049 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25050 BB_SAME_SUCC (bb) == NULL.
25052 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25054 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25057 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25059 * dumpfile.c: Include profile-count.h
25060 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25062 (insert_cond_bb): Update profile.
25063 * tree-cfg.h (insert_cond_bb): Update prototype.
25064 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25065 * tree-dump.c: Do not include tree-cfg.
25067 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25069 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25071 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25073 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25076 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25078 * tree-complex.c (expand_complex_div_wide): update profile.
25080 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25081 Alan Hayward <alan.hayward@arm.com>
25082 David Sherwood <david.sherwood@arm.com>
25084 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25085 (CORETYPES_H): New define.
25086 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25087 (insn-modes-inline.h, s-modes-inline-h): New rules.
25088 (generated_files): Add insn-modes-inline.h.
25089 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25090 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25091 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25092 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25093 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25094 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25095 (build/gencodes.o, build/genconditions.o): Likewise.
25096 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25097 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25098 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25099 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25100 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25101 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25102 * coretypes.h: Include everything up to real.h for generators.
25103 Include insn-modes.h first. Include wide-int-print.h after
25104 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25105 * machmode.h: Don't include insn-modes.h here.
25106 * function-tests.c: Remove includes of signop.h, machmode.h,
25107 double-int.h and wide-int.h.
25109 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25111 * optc-save-gen.awk: Likewise.
25112 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25113 * godump.c: Remove include of wide-int-print.h.
25114 * pretty-print.h: Likewise.
25115 * wide-int-print.cc: Likewise.
25116 * wide-int.cc: Likewise.
25117 * hash-map-tests.c: Remove include of signop.h.
25118 * hash-set-tests.c: Likewise.
25119 * rtl-tests.c: Likewise.
25120 * mkconfig.sh: Remove include of machmode.h.
25121 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25123 (emit_insn_modes_inline_h): ...this new function. Emit the code
25124 into an insn-modes-inline.h header file, adding appropriate
25125 include guards and end comments.
25126 (emit_insn_modes_c_header): Remove include of machmode.h.
25127 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25129 (main): Handle -i flag and call emit_insn_modes_inline_h when
25132 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25134 * tree-ssa-strlen.c (strinfo): Rename the length field to
25135 nonzero_chars. Add a full_string_p field.
25136 (compare_nonzero_chars, zero_length_string_p): New functions.
25137 (get_addr_stridx): Add an offset_out parameter.
25138 Use compare_nonzero_chars.
25139 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25140 (new_strinfo): Update after above changes to strinfo.
25141 (set_endptr_and_length): Set full_string_p.
25142 (get_string_length): Update after above changes to strinfo.
25143 (unshare_strinfo): Update call to new_strinfo.
25144 (maybe_invalidate): Likewise.
25145 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25146 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25147 as a uhwi instead of an shwi. Update after above changes to
25148 strinfo and new_strinfo.
25149 (zero_length_string): Assert that chainsi contains full strings.
25150 Use zero_length_string_p. Update call to new_strinfo.
25151 (adjust_related_strinfos): Update after above changes to strinfo.
25152 Copy full_string_p from origsi.
25153 (adjust_last_stmt): Use zero_length_string_p.
25154 (handle_builtin_strlen): Update after above changes to strinfo and
25155 new_strinfo. Install the lhs as the string length if the previous
25156 entry didn't describe a full string.
25157 (handle_builtin_strchr): Update after above changes to strinfo
25159 (handle_builtin_strcpy): Likewise.
25160 (handle_builtin_strcat): Likewise.
25161 (handle_builtin_malloc): Likewise.
25162 (handle_pointer_plus): Likewise.
25163 (handle_builtin_memcpy): Likewise. Track nonzero characters
25164 that aren't necessarily followed by a nul terminator.
25165 (handle_char_store): Likewise.
25167 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25169 PR tree-optimization/80769
25170 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25171 for malloc and calloc. Document the new invariant that all related
25172 strinfos have delayed lengths or none do.
25173 (verify_related_strinfos): Move earlier in file.
25174 (set_endptr_and_length): New function, split out from...
25175 (get_string_length): ...here. Also set the lengths of related
25177 (zero_length_string): Assert that chainsi has known (rather than
25179 (adjust_related_strinfos): Likewise.
25181 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25183 PR tree-optimization/81136
25184 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25185 assert that two references with the same misalignment have the same
25186 compile-time misalignment if those compile-time misalignments
25189 2017-07-01 Andi Kleen <ak@linux.intel.com>
25191 * print-tree.c (print_node): Print all attributes.
25193 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25195 * cfg.c (scale_bbs_frequencies): New function.
25196 * cfg.h (scale_bbs_frequencies): Declare it.
25197 * cfgloopanal.c (single_likely_exit): Cleanup.
25198 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25200 (scale_loop_profile): Likewise.
25201 (loop_version): Likewise.
25202 (create_empty_loop_on_edge): Update.
25203 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25204 scale_loop_frequencies, scale_loop_profile, loopify,
25205 loop_version): Update prototypes.
25206 * modulo-sched.c (sms_schedule): Update.
25207 * predict.c (unlikely_executed_edge_p): Also check probability.
25208 (probably_never_executed_edge_p): Fix typo.
25209 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25210 * tree-parloops.c (gen_parallel_loop): Update.
25211 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25212 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25213 * tree-ssa-loop-split.c (split_loop): Update.
25214 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25215 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25216 (vect_loop_versioning): Update.
25217 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25219 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25221 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25223 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25225 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25226 to keep profile consistent.
25228 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25230 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25231 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25232 * profile-count.h (max_safe_multiplier): Make unsigned.
25233 (profile_count::guessed_zero): New.
25235 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25237 * bb-reorder.c (fix_up_crossing_landing_pad,
25238 fix_crossing_conditional_branches): Use make_single_succ_edge
25239 to keep profile consistent.
25241 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25243 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25246 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25249 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25250 the right scopes, make sure cond_jump isn't preserved between multiple
25251 iterations. Search for fallthru edge whenever there are 3+ edges and
25252 use find_fallthru_edge for it.
25254 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25256 Patch by Alexander Monakov <amonakov@ispras.ru>
25257 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25258 probabilities consistently.
25260 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25262 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25263 emit_cmp_and_jump_insns.
25265 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25268 * tree-inline.c (expand_call_inline): Combine profile statuses.
25270 2017-06-30 Andrew Pinski <apinski@cavium.com>
25272 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25273 fold_stmt returned true.
25275 2017-06-30 Nathan Sidwell <nathan@acm.org>
25277 * ggc.h (empty_string): Delete.
25278 * cfgexpand.c (expand_asm_stmt): Use plain "".
25279 * optabs.c (expand_asm_memory_barrier): Likewise.
25280 * stringpool.c (empty_string): Delete.
25281 (digit_vector, digit_string): Delete.
25282 (ggc_alloc_string): Use plain "", don't optimize single digit
25283 strings. Use ggc_alloc_atomic.
25285 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25287 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25288 comparison set and one other set, use the cost of the non-comparison
25291 2017-06-30 Nathan Sidwell <nathan@acm.org>
25293 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25296 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25298 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25299 loops. Remove now unneeded calls to gimple_switch_set_label() that
25300 just set removed labels to NULL_TREE.
25302 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25304 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25305 (set_range_info): ...here. Only call set_range_info_raw if domain
25307 (set_nonzero_bits): Call set_range_info_raw.
25308 * tree-ssanames.h (set_range_info_raw): New.
25310 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25313 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25314 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25315 of nonimmediate_operand and <store_mask_constraint> instead of m
25316 for the input operand. For V8FI iterator, always split if input
25317 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25318 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25319 <store_mask_predicate> instead of register_operand and
25320 <store_mask_constraint> instead of v for the input operand. Make
25321 sure both operands aren't MEMs for if not <mask_applied>.
25323 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25325 * lto-wrapper.c (copy_file) Close both file descriptors before
25328 2017-06-30 Martin Liska <mliska@suse.cz>
25331 * multiple_target.c (create_dispatcher_calls): Make ifunc
25332 also for function that don't have calls or are not referenced.
25334 2017-06-30 Richard Biener <rguenther@suse.de>
25336 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25337 analyze the first scalar stmt. Move vector type computation
25338 for the BB case here from ...
25339 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25340 live operation processing in the SLP case properly.
25342 2017-06-30 Richard Biener <rguenther@suse.de>
25344 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25346 2017-06-30 Martin Liska <mliska@suse.cz>
25349 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25350 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25352 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25354 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25357 2017-06-30 Richard Biener <rguenther@suse.de>
25359 PR tree-optimization/81249
25360 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25361 condition reduction result to original scalar type.
25363 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25365 * profile-count.h (enum profile_quality): Fix typos and whitespace
25368 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25370 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25371 type for branch probabilities.
25373 2017-06-29 Julian Brown <julian@codesourcery.com>
25374 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25376 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25377 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25378 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25379 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25381 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25383 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25384 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25385 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25386 CC usage from generic code to here.
25387 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25388 CC usage into the target macros.
25390 2017-06-29 Maya Rashish <coypu@sdf.org>
25392 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25395 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25397 * arm/arm-builtins.c: Include profile-count.h
25398 * except.c (sjlj_emit_function_enter): Use
25399 profile_probability::unlikely.
25401 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25403 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25404 and tocrel_offset be pointer args rather than implicitly using
25406 (legitimate_constant_pool_address_p, rs6000_emit_move,
25407 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25408 tocrel_offset and use in toc_relative_expr_p call.
25409 (print_operand, print_operand_address): Use static tocrel_base_oac
25410 and tocrel_offset_oac.
25411 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25414 2017-06-29 Maya Rashish <coypu@sdf.org>
25416 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25418 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25420 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25421 objects, take into account only the alignment of 'op0' and 'mode1' if
25424 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25426 * ccmp.c (ccmp_tree_comparison_p): New function.
25427 (ccmp_candidate_p): Update to use above function.
25428 (get_compare_parts): New function.
25429 (expand_ccmp_next): Update to use new functions.
25430 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25432 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25433 take mode as argument.
25434 * ccmp.h (expand_ccmp_expr): Add mode as argument.
25435 * expr.c (expand_expr_real_1): Pass mode as argument.
25437 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
25439 * combine.c (combine_instructions): Print insns to dump_file, together
25442 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25444 * asan.c (asan_emit_stack_protection): Update.
25445 (create_cond_insert_point): Update.
25446 * auto-profile.c (afdo_propagate_circuit): Update.
25447 * basic-block.h (struct edge_def): Turn probability to
25448 profile_probability.
25449 (EDGE_FREQUENCY): Update.
25450 * bb-reorder.c (find_traces_1_round): Update.
25451 (better_edge_p): Update.
25452 (sanitize_hot_paths): Update.
25453 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
25454 (make_single_succ_edge): Update.
25455 (check_bb_profile): Update.
25456 (dump_edge_info): Update.
25457 (update_bb_profile_for_threading): Update.
25458 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
25459 probabilitycount to 0.
25460 * cfgbuild.c (compute_outgoing_frequencies): Update.
25461 * cfgcleanup.c (try_forward_edges): Update.
25462 (outgoing_edges_match): Update.
25463 (try_crossjump_to_edge): Update.
25464 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
25465 (expand_gimple_tailcall): Update.
25466 (construct_init_block): Use make_single_succ_edge.
25467 (construct_exit_block): Use make_single_succ_edge.
25468 * cfghooks.c (verify_flow_info): Update.
25469 (redirect_edge_succ_nodup): Update.
25470 (split_edge): Update.
25471 (account_profile_record): Update.
25472 * cfgloopanal.c (single_likely_exit): Update.
25473 * cfgloopmanip.c (scale_loop_profile): Update.
25474 (set_zero_probability): Remove.
25475 (duplicate_loop_to_header_edge): Update.
25476 * cfgloopmanip.h (loop_version): Update prototype.
25477 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
25478 (force_nonfallthru_and_redirect): Update.
25479 (update_br_prob_note): Update.
25480 (rtl_verify_edges): Update.
25481 (purge_dead_edges): Update.
25482 (rtl_lv_add_condition_to_bb): Update.
25483 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
25484 * cgraphunit.c (init_lowered_empty_function): Update.
25485 (cgraph_node::expand_thunk): Update.
25486 * cilk-common.c: Include profile-count.h
25487 * dojump.c (inv): Remove.
25488 (jumpifnot): Update.
25489 (jumpifnot_1): Update.
25490 (do_jump_1): Update.
25492 (do_jump_by_parts_greater_rtx): Update.
25493 (do_compare_rtx_and_jump): Update.
25494 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
25495 do_jump_1. do_compare_rtx_and_jump): Update prototype.
25496 * dwarf2cfi.c: Include profile-count.h
25497 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
25498 (sjlj_emit_dispatch_table): Likewise.
25499 * explow.c: Include profile-count.h
25500 * expmed.c (emit_store_flag_force): Update.
25501 (do_cmp_and_jump): Update.
25502 * expr.c (compare_by_pieces_d::generate): Update.
25503 (compare_by_pieces_d::finish_mode): Update.
25504 (emit_block_move_via_loop): Update.
25505 (store_expr_with_bounds): Update.
25506 (store_constructor): Update.
25507 (expand_expr_real_2): Update.
25508 (expand_expr_real_1): Update.
25509 * expr.h (try_casesi, try_tablejump): Update prototypes.
25510 * gimple-pretty-print.c (dump_probability): Update.
25511 (dump_profile): New.
25512 (dump_gimple_label): Update.
25513 (dump_gimple_bb_header): Update.
25514 * graph.c (draw_cfg_node_succ_edges): Update.
25515 * hsa-gen.c (convert_switch_statements): Update.
25516 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
25517 (find_if_case_1): Update.
25518 (find_if_case_2): Update.
25519 * internal-fn.c (expand_arith_overflow_result_store): Update.
25520 (expand_addsub_overflow): Update.
25521 (expand_neg_overflow): Update.
25522 (expand_mul_overflow): Update.
25523 (expand_vector_ubsan_overflow): Update.
25524 * ipa-cp.c (good_cloning_opportunity_p): Update.
25525 * ipa-split.c (split_function): Use make_single_succ_edge.
25526 * ipa-utils.c (ipa_merge_profiles): Update.
25527 * loop-doloop.c (add_test): Update.
25528 (doloop_modify): Update.
25529 * loop-unroll.c (compare_and_jump_seq): Update.
25530 (unroll_loop_runtime_iterations): Update.
25531 * lra-constraints.c (lra_inheritance): Update.
25532 * lto-streamer-in.c (input_cfg): Update.
25533 * lto-streamer-out.c (output_cfg): Update.
25534 * mcf.c (adjust_cfg_counts): Update.
25535 * modulo-sched.c (sms_schedule): Update.
25536 * omp-expand.c (expand_omp_for_init_counts): Update.
25537 (extract_omp_for_update_vars): Update.
25538 (expand_omp_ordered_sink): Update.
25539 (expand_omp_for_ordered_loops): Update.
25540 (expand_omp_for_generic): Update.
25541 (expand_omp_for_static_nochunk): Update.
25542 (expand_omp_for_static_chunk): Update.
25543 (expand_cilk_for): Update.
25544 (expand_omp_simd): Update.
25545 (expand_omp_taskloop_for_outer): Update.
25546 (expand_omp_taskloop_for_inner): Update.
25547 * omp-simd-clone.c (simd_clone_adjust): Update.
25548 * optabs.c (expand_doubleword_shift): Update.
25549 (expand_abs): Update.
25550 (emit_cmp_and_jump_insn_1): Update.
25551 (expand_compare_and_swap_loop): Update.
25552 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
25553 * predict.c (predictable_edge_p): Update.
25554 (edge_probability_reliable_p): Update.
25555 (set_even_probabilities): Update.
25556 (combine_predictions_for_insn): Update.
25557 (combine_predictions_for_bb): Update.
25558 (propagate_freq): Update.
25559 (estimate_bb_frequencies): Update.
25560 (force_edge_cold): Update.
25561 * profile-count.c (profile_count::dump): Add missing space into dump.
25562 (profile_count::debug): Add newline.
25563 (profile_count::differs_from_p): Explicitly convert to unsigned.
25564 (profile_count::stream_in): Update.
25565 (profile_probability::dump): New member function.
25566 (profile_probability::debug): New member function.
25567 (profile_probability::differs_from_p): New member function.
25568 (profile_probability::differs_lot_from_p): New member function.
25569 (profile_probability::stream_in): New member function.
25570 (profile_probability::stream_out): New member function.
25571 * profile-count.h (profile_count_quality): Rename to ...
25572 (profile_quality): ... this one.
25573 (profile_probability): New.
25574 (profile_count): Update.
25575 * profile.c (compute_branch_probabilities): Update.
25576 * recog.c (peep2_attempt): Update.
25577 * sched-ebb.c (schedule_ebbs): Update.
25578 * sched-rgn.c (find_single_block_region): Update.
25579 (compute_dom_prob_ps): Update.
25580 (schedule_region): Update.
25581 * sel-sched-ir.c (compute_succs_info): Update.
25582 * stmt.c (struct case_node): Update.
25583 (do_jump_if_equal): Update.
25584 (get_outgoing_edge_probs): Update.
25585 (conditional_probability): Update.
25586 (emit_case_dispatch_table): Update.
25587 (expand_case): Update.
25588 (expand_sjlj_dispatch_table): Update.
25589 (emit_case_nodes): Update.
25590 * targhooks.c: Update.
25591 * tracer.c (better_p): Update.
25592 (find_best_successor): Update.
25593 * trans-mem.c (expand_transaction): Update.
25594 * tree-call-cdce.c: Update.
25595 * tree-cfg.c (gimple_split_edge): Upate.
25596 (move_sese_region_to_fn): Upate.
25597 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
25598 * tree-eh.c (lower_resx): Upate.
25599 (cleanup_empty_eh_move_lp): Upate.
25600 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25601 * tree-inline.c (copy_edges_for_bb): Update.
25602 (copy_cfg_body): Update.
25603 * tree-parloops.c (gen_parallel_loop): Update.
25604 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
25605 (gimple_gen_time_profiler): Update.
25606 * tree-ssa-dce.c (remove_dead_stmt): Update.
25607 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25608 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
25609 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
25610 (unloop_loops): Update.
25611 (try_peel_loop): Update.
25612 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25613 * tree-ssa-loop-split.c (connect_loops): Update.
25614 (split_loop): Update.
25615 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25616 (hoist_guard): Update.
25617 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
25618 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
25619 (value_replacement): Update.
25620 * tree-ssa-reassoc.c (branch_fixup): Update.
25621 * tree-ssa-tail-merge.c (replace_block_by): Update.
25622 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
25623 (create_edge_and_update_destination_phis): Update.
25624 (compute_path_counts): Update.
25625 (recompute_probabilities): Update.
25626 (update_joiner_offpath_counts): Update.
25627 (freqs_to_counts_path): Update.
25628 (duplicate_thread_path): Update.
25629 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
25630 (struct switch_conv_info): Update.
25631 (gen_inbound_check): Update.
25632 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25633 (vect_do_peeling): Update.
25634 (vect_loop_versioning): Update.
25635 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25636 (optimize_mask_stores): Update.
25637 * ubsan.c (ubsan_expand_null_ifn): Update.
25638 * value-prof.c (gimple_divmod_fixed_value): Update.
25639 (gimple_divmod_fixed_value_transform): Update.
25640 (gimple_mod_pow2): Update.
25641 (gimple_mod_pow2_value_transform): Update.
25642 (gimple_mod_subtract): Update.
25643 (gimple_mod_subtract_transform): Update.
25644 (gimple_ic): Update.
25645 (gimple_stringop_fixed_value): Update.
25646 (gimple_stringops_transform): Update.
25647 * value-prof.h: Update.
25649 2017-06-29 Carl Love <cel@us.ibm.com>
25651 * config/rs6000/rs6000-c.c: Add support for built-in functions
25652 vector signed int vec_signed (vector float);
25653 vector signed long long vec_signed (vector double);
25654 vector signed int vec_signed2 (vector double, vector double);
25655 vector signed int vec_signede (vector double);
25656 vector signed int vec_signedo (vector double);
25657 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
25658 instruction generator.
25659 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25660 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
25661 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
25663 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
25664 vunsignede_v2df): Add define_expands.
25665 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
25666 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
25667 VEC_UNSIGNEDO): Add definitions.
25668 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
25669 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
25670 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
25671 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
25672 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
25673 * config/rs6000/altivec.h (vec_signed, vec_signed2,
25674 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
25675 vec_unsignede, vec_unsignedo): Add builtin defines.
25676 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
25678 * doc/extend.texi: Update the built-in documentation file for the
25679 new built-in functions.
25681 2017-06-29 Richard Biener <rguenther@suse.de>
25683 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
25684 reduction chains to LOOP_VINFO_REDUCTIONS.
25685 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
25686 SLP reductions after processing reduction chains.
25688 2017-06-29 Nathan Sidwell <nathan@acm.org>
25690 * builtins.c (fold_builtin_FUNCTION): Use
25691 lang_hooks.decl_printable_name.
25693 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
25695 PR middle-end/81194
25696 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
25697 with only one label.
25698 * stmt.c (expand_case): Assert NCASES is greater than one.
25700 2017-06-29 Richard Biener <rguenther@suse.de>
25702 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
25704 (group_case_labels): Likewise.
25705 (find_taken_edge): Push sanity checking on val to workers...
25706 (find_taken_edge_cond_expr): ... here
25707 (find_taken_edge_switch_expr): ... and here, handle cases
25708 with just a default label.
25709 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
25710 (group_case_labels): Likewise.
25711 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
25712 group_case_labels does anything cleanup the CFG again.
25714 2017-06-29 Bin Cheng <bin.cheng@arm.com>
25716 PR tree-optimization/81196
25717 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
25718 exit condition comparing two IVs.
25720 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
25722 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
25723 profile to the dummy entry at the end of the list of architectures.
25724 * config/arm/arm-cpu-cdata.h: Regenerated.
25726 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25727 Michael Collison <michael.collison@arm.com>
25730 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
25732 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
25733 (*aarch64_reg_<mode>3_minus_mask): New pattern.
25734 (*aarch64_<optab>_reg_di3_mask2): New pattern.
25735 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
25736 of shift when the shift amount is masked with constant equal to
25737 the size of the mode.
25738 * config/aarch64/predicates.md (subreg_lowpart_operator): New
25741 2017-06-29 Martin Liska <mliska@suse.cz>
25743 * config/i386/i386.opt: Change range from [1,5] to [0,5].
25745 2017-06-29 Yury Gribov <tetra2005@gmail.com>
25748 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
25751 (edge_growth_cache_entry::edge_growth_cache_entry): New
25753 (reset_edge_growth_cache): Update to use constructor.
25755 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25757 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
25758 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25759 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
25761 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
25763 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
25764 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
25766 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
25768 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25769 (*-linux-uclibc*): Add t-uclibc tmake_file.
25770 * config/t-musl: New.
25771 * config/t-uclibc: New.
25773 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25775 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25777 (gen_comm_data): Emit architectural setting of arch_prof.
25778 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25780 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25781 (armv8-m.base, armv8-m.main): Likewise.
25782 * arm-protos.h (arm_build_target): Add profile field.
25783 (arch_option): Likewise.
25784 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25786 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25787 arm_active_target.profile.
25789 2017-06-28 Richard Biener <rguenther@suse.de>
25791 PR middle-end/81227
25792 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25793 TYPE_OVERFLOW_WRAPS.
25794 * match.pd (negate_expr_p): Likewise.
25795 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25796 fold_build2, not fold_binary.
25798 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25800 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25801 Convert memory address to Pmode.
25802 (aarch64_print_operand): Assert MEM operands are always Pmode.
25804 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25807 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25808 Remove redundant if.
25809 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25810 * config/arm/aarch-common-protos.h
25811 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25812 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25814 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25817 * multiple_target.c (create_dispatcher_calls): Set the default
25818 clone to be static, not public.
25820 2017-06-28 Richard Biener <rguenther@suse.de>
25822 * tree-vect-loop.c (vectorizable_reduction): Move special
25823 cond reduction IV var creation ...
25824 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25825 parameter. Use STMT_VINFO_VECTYPE.
25826 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25829 2017-06-28 Martin Liska <mliska@suse.cz>
25832 * ipa-visibility.c (non_local_p): Handle visibility.
25834 2017-06-28 Martin Liska <mliska@suse.cz>
25837 * common.opt: Add IntegerRange to various options.
25838 * opt-functions.awk (integer_range_info): New function.
25839 * optc-gen.awk: Add integer_range_info to cl_options struct.
25840 * opts-common.c (decode_cmdline_option): Handle
25841 CL_ERR_INT_RANGE_ARG.
25842 (cmdline_handle_error): Likewise.
25843 * opts.c (print_filtered_help): Show valid interval in
25844 when --help is provided.
25845 * opts.h (struct cl_option): Add range_min and range_max fields.
25846 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25848 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25850 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25851 (x * C EQ/NE y * C): New transformation.
25853 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25855 * genmultilib (combination_space): Accept '+' in option names.
25857 2017-06-28 Martin Liska <mliska@suse.cz>
25860 * asan.c (instrument_derefs): Bail out inner references
25861 that are hard register variables.
25863 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25866 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25867 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25869 2017-06-28 Richard Biener <rguenther@suse.de>
25871 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25872 (vect_get_slp_defs): Adjust.
25873 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25875 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25877 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25878 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25879 (vectorizable_reduction): Adjust.
25880 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25882 (vect_get_slp_defs): Likewise.
25883 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25884 (vectorizable_bswap): Adjust.
25885 (vectorizable_call): Likewise.
25886 (vectorizable_conversion): Likewise.
25887 (vectorizable_assignment): Likewise.
25888 (vectorizable_shift): Likewise.
25889 (vectorizable_operation): Likewise.
25890 (vectorizable_store): Likewise.
25891 (vectorizable_condition): Likewise.
25892 (vectorizable_comparison): Likewise.
25894 2017-06-28 Michael Collison <michael.collison@arm.com>
25897 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25899 (arm_gen_movmemqi): Removed unused variable 'i'.
25900 Convert 'for' loop into 'while' loop.
25901 (arm_expand_prologue): Remove last unnecessary set of insn.
25902 (thumb_pop): Remove unused variable 'pushed_words'.
25903 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25905 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25907 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25908 * config/s390/s390-protos.h: Add prototype of
25909 s390_rel_address_ok_p.
25910 * config/s390/s390.c (s390_got_symbol): New function.
25911 (s390_rel_address_ok_p): New function.
25912 (legitimize_pic_address): Use s390_rel_address_ok_p.
25913 (s390_load_got): Use s390_got_symbol.
25914 (s390_option_override): Issue error if
25915 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25916 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25918 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25920 2017-06-27 Andrew Pinski <apinski@cavium.com>
25922 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25923 (X * copysign (1.0, X)): New pattern.
25924 (X * copysign (1.0, -X)): New pattern.
25925 (copysign (-1.0, CST)): New pattern.
25927 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25929 * genmultilib (combination_space): Remove variable.
25930 Validate reuse rules against regular expression for any sequence
25931 of multilib options in any order.
25933 2017-06-27 Michael Collison <michael.collison@arm.com>
25935 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25936 call aarch64_split_simd_combine.
25937 * (aarch64_combine_internal<mode>): Delete pattern.
25938 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25939 Allow register and subreg operands.
25941 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25943 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25944 specific need, just fallback on defaults.
25945 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25947 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25948 Olivier Hainque <hainque@adacore.com>
25950 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25952 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25953 targets. Pick a default if no particular attempt applied.
25954 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25957 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25959 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25960 (x86_64-wrs-vxworks7): Likewise.
25962 2017-06-27 Marek Polacek <polacek@redhat.com>
25965 * ubsan.c (instrument_null): Check get_base_address's result for null.
25967 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25969 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25971 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25973 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25974 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25975 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25976 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25977 New function types.
25978 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25979 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25980 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25981 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25982 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25983 BUILT_IN_FEUPDATEENV): New builtins.
25984 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25985 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25986 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25987 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25989 (builtin_structptr_types): Adjust size.
25990 * tree.c (builtin_structptr_types): Add four entries.
25992 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25993 Olivier Hainque <hainque@adacore.com>
25995 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25996 (TLS_SYM): New local macro, forcing reference to __tls__ on
25997 link command lines for VxWorks 7 RTPs, triggering initialization
25999 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26000 OS features TLS support, true for RTPs on VxWorks 7.
26001 * config/vxworks.c (vxworks_override_options): Setup emutls
26004 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26006 * predict.c (test_prediction_value_range): Use -1U instead of -1
26007 to avoid narrowing conversion warning.
26008 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26009 to avoid narrowing conversion warning.
26010 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26012 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26014 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26016 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26017 64bit configurations.
26018 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26019 (SIZE_TYPE): Likewise.
26020 * config/vxworks.c (vxworks_emutls_var_fields): Use
26021 long_unsigned_type_node instead of unsigned_type_node as the offset
26022 field type, which is "pointer" mode in emutls.c.
26024 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26027 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26029 PR middle-end/81207
26030 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26031 gimple_vuse copying separately from gimple_vdef copying.
26033 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26035 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26038 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26039 Olivier Hainque <hainque@adacore.com>
26041 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26042 for all vxworks7 targets.
26043 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26044 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26045 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26046 variations for VX6/VX7 and 32/64bits later on in ...
26047 (VXWORKS_LIB_SPEC): Leverage new macros.
26048 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26049 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26051 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26053 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26054 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26056 2017-06-26 Carl Love <cel@us.ibm.com>
26058 * config/rs6000/rs6000-c.c: Add support for built-in functions
26059 vector bool char vec_reve (vector bool char);
26060 vector signed char vec_reve (vector signed char);
26061 vector unsigned char vec_reve (vector unsigned char);
26062 vector bool int vec_reve (vector bool int);
26063 vector signed int vec_reve (vector signed int);
26064 vector unsigned int vec_reve (vector unsigned int);
26065 vector bool long long vec_reve (vector bool long long);
26066 vector signed long long vec_reve (vector signed long long);
26067 vector unsigned long long vec_reve (vector unsigned long long);
26068 vector bool short vec_reve (vector bool short);
26069 vector signed short vec_reve (vector signed short);
26070 vector double vec_reve (vector double);
26071 vector float vec_reve (vector float);
26072 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26073 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26074 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26075 (altivec_vreve): New pattern.
26076 * config/rs6000/altivec.h (vec_reve): New define.
26077 * doc/extend.texi (vec_rev): Update the built-in documentation file
26078 for the new built-in functions.
26080 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26082 PR tree-optimization/71815
26083 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26085 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26087 (slsr_process_phi): Likewise.
26088 (replace_uncond_cands_and_profitable_phis): Don't replace a
26089 multiply candidate with a stride of 1 (copy or cast).
26090 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26092 (lowest_cost_path): Likewise.
26093 (total_savings): Likewise.
26095 2017-06-26 Richard Biener <rguenther@suse.de>
26098 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26099 Use def_builtin_pure for all gather builtins.
26101 2017-06-26 Richard Biener <rguenther@suse.de>
26103 PR tree-optimization/81203
26104 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26105 non-dominating BBs.
26107 2017-06-26 Marek Polacek <polacek@redhat.com>
26110 * doc/invoke.texi: Document -Wmultistatement-macros.
26112 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26114 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26115 arm_neon_ok_no_float_abi effective target.
26117 2017-06-26 Richard Biener <rguenther@suse.de>
26119 PR tree-optimization/80928
26120 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26121 (copy_bbs): Set BB_DUPLICATED flag early.
26122 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26124 (execute_on_shrinking_pred): Likewise.
26125 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26126 BB_DUPLICATED blocks.
26127 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26128 iterate over all PHIs considering removal of *gsi.
26130 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26132 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26135 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26137 * config/rs6000/rs6000-string.c: (expand_block_clear,
26138 do_load_for_compare, select_block_compare_mode,
26139 compute_current_alignment, expand_block_compare,
26140 expand_strncmp_align_check, expand_strn_compare,
26141 expand_block_move, rs6000_output_load_multiple)
26142 Move functions related to string/block move/compare
26143 to a separate file.
26144 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26145 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26146 for this function which is now used in two files.
26147 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26148 * config.gcc: Add rs6000-string.o to extra_objs for
26149 targets powerpc*-*-* and rs6000*-*-*.
26151 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26154 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26155 32-bit, since indexed is not valid for DImode.
26156 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26157 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26158 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26159 (define_peephole2 for Altivec d-form store): Likewise.
26162 * multiple_target.c (create_dispatcher_calls): Only create the
26163 dispatcher call if the function is the default clone of a
26164 versioned function.
26166 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26168 PR middle-end/80902
26169 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26170 a call, force the call to not be a tail call.
26172 2017-06-23 Jeff Law <law@redhat.com>
26174 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26177 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26179 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26180 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26181 handling for early expansion of vector shifts (sl,sr,sra,rl).
26182 (builtin_function_type): Add vector shift right instructions
26183 to the unsigned argument list.
26185 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26187 rtl-optimizatoin/79286
26188 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26189 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26190 trap. PIC register plus a const unspec without offset can never trap.
26192 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26194 * tree.h (builtin_structptr_type): New type.
26195 (builtin_structptr_types): Declare new array.
26196 * tree.c (builtin_structptr_types): New array.
26197 (free_lang_data, build_common_tree_nodes): Use it.
26199 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26202 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26205 2017-06-22 Matt Turner <mattst88@gmail.com>
26207 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26208 Lake models to skylake case. Assume skylake for unknown
26209 models with clflushopt.
26211 2017-06-22 Jeff Law <law@redhat.com>
26213 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26214 frame sizes that do not satisfy aarch64_uimm12_shift.
26216 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26218 * profile-count.h (apply_probability,
26219 apply_scale, probability_in): Fix checks for zero.
26221 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26223 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26224 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26226 2016-06-22 Richard Biener <rguenther@suse.de>
26228 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26229 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26230 REDUC_MAX_EXPR support.
26231 (vectorizable_reduction): Likewise.
26232 (vect_create_epilog_for_reduction): Likewise.
26234 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26236 * match.pd (A / (1 << B) -> A >> B): New.
26237 * generic-match-head.c: Include optabs-tree.h.
26238 * gimple-match-head.c: Likewise.
26239 * optabs-tree.h (target_supports_op_p): New.
26240 * optabs-tree.c (target_supports_op_p): New.
26242 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26244 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26245 $gcc_cv_ld --help output.
26246 (gcc_cv_ld_demangle): Likewise.
26247 (gcc_cv_ld_eh_frame_hdr): Likewise.
26248 (gcc_cv_ld_pie): Likewise.
26249 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26250 (gcc_cv_ld_buildid): Likewise.
26251 (gcc_cv_ld_sysroot): Likewise.
26252 (ld_bndplt_support): Likewise.
26253 (ld_pushpopstate_support): Likewise.
26254 * configure: Regenerate.
26255 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26257 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26260 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26261 operands indexes to avoid gap between operands and match_dups.
26263 2017-06-21 Andrew Pinski <apinski@cavium.com>
26265 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26266 Increment Arith_shift and Arith_shift_reg by 1.
26267 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26269 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26270 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26271 (aarch64_strip_extend): Add new argument and test for it.
26272 (aarch64_cheap_mult_shift_p): New function.
26273 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26274 add a cost if it is true.
26275 Update calls to aarch64_strip_extend.
26276 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26278 2017-06-21 Andrew Pinski <apinski@cavium.com>
26280 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26282 (thunderxt88): Likewise.
26283 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26284 (thunderx_prefetch_tune): New variable.
26285 (thunderx2t99_prefetch_tune): Update for the correct values.
26286 (thunderxt88_tunings): New variable.
26287 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26288 generic_prefetch_tune.
26289 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26291 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26293 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26294 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26295 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26296 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26297 (aarch64_atomic_cas<mode>, GPI): Likewise.
26299 2017-06-21 Martin Liska <mliska@suse.cz>
26301 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26302 statements on cold and hot labels.
26303 * predict.c (tree_estimate_probability_bb): Remove the
26304 prediction from this place.
26306 2017-06-21 Martin Liska <mliska@suse.cz>
26308 PR tree-optimization/79489
26309 * gimplify.c (maybe_add_early_return_predict_stmt): New
26311 (gimplify_return_expr): Call the function.
26312 * predict.c (tree_estimate_probability_bb): Remove handling
26314 * predict.def: Update comment about early return predictor.
26315 * gimple-predict.h (is_gimple_predict): New function.
26316 * predict.def: Change default value of early return to 66.
26317 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26319 * passes.def: Put pass_strip_predict_hints to the beginning of
26322 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26324 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26325 FUNCTION_DECL declarations.
26326 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26328 (dwaf2out_decl): Likewise.
26329 * godump.c (go_early_global_decl): Skip call to the real debug hook
26330 for FUNCTION_DECL declarations.
26331 * passes.c (rest_of_decl_compilation): Skip call to the
26332 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26333 -fdump-go-spec is passed.
26335 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26337 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26338 Reorder for compactness.
26339 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26340 (def_builtin_pure, def_builtin_pure2): New functions.
26341 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26343 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26345 * match.pd (nop_convert): New predicate.
26346 ((A +- CST1) +- CST2): Allow some NOP conversions.
26348 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26351 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26352 with ctors/dtors if GOVD_SHARED is set.
26354 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26356 * config/aarch64/aarch64.md (movti_aarch64):
26357 Emit mov rather than orr.
26358 (movtf_aarch64): Likewise.
26359 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26360 Emit mov rather than orr.
26362 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26364 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26365 Swap alternatives, make integer dup more expensive.
26367 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26369 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26370 Return true for non-tls symbols.
26372 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26374 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26375 (cortex-a75): Likewise.
26376 (cortex-a75.cortex-a55): Likewise.
26377 * config/aarch64/aarch64-tune.md: Regenerate.
26378 * doc/invoke.texi (-mtune): Document new values for -mtune.
26380 2017-06-21 Tom de Vries <tom@codesourcery.com>
26382 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26383 stack_size feature.
26384 (Effective-Target Keywords, Other attributes): Suggest using
26385 dg-add-options stack_size feature to get stack limit in stack_size
26386 effective target documentation.
26388 2017-06-21 Julian Brown <julian@codesourcery.com>
26389 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26391 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26392 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26393 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26395 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26396 attribute type list for neon_multiply.
26397 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26398 attribute type list for neon_multiply.
26399 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26400 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26401 attribute type list for neon_multiply.
26402 * config/arm/types.md (crypto_pmull): Add.
26403 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26404 attribute type list.
26406 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26408 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26411 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26413 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26414 to make sure not to dereference a NULL cost_classes_ptr pointer.
26416 2017-06-20 Carl Love <cel@us.ibm.com>
26418 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26419 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26420 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26421 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26422 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26423 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26424 VMULOSW): New enum "unspec" values.
26425 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26426 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26427 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26428 altivec_vmulosw): New patterns.
26429 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26430 VMULOSW): Add definitions.
26432 2017-06-20 Julia Koval <julia.koval@intel.com>
26434 * config/i386/i386.c: Fix rounding expand for new pattern.
26435 * config/i386/subst.md: Fix pattern (parallel -> unspec).
26437 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26439 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
26440 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
26442 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26444 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
26447 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
26449 * config/aarch64/aarch64-cores.def: Rearrange to sort by
26450 architecture, then by implementer ID.
26451 * config/aarch64/aarch64-tune.md: Regenerate.
26453 2017-06-20 Richard Biener <rguenther@suse.de>
26455 PR middle-end/81097
26456 * fold-const.c (split_tree): Fold to type before negating.
26458 2017-06-20 David Malcolm <dmalcolm@redhat.com>
26460 * diagnostic-show-locus.c
26461 (selftest::test_fixit_deletion_affecting_newline): New function.
26462 (selftest::diagnostic_show_locus_c_tests): Call it.
26464 2017-06-20 Andreas Schwab <schwab@suse.de>
26467 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
26470 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26472 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
26473 __ARM_FEATURE_COPROC according to support.
26475 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26477 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26478 Rewritten to avoid overflow for > 32-bit pointers.
26481 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
26482 by removing enum keyword.
26483 (ubsan_type_descriptor): Likewise. Formatting fix.
26486 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
26487 splitter): Require TARGET_SSE2 in the condition.
26489 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
26492 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
26493 for doing vector set of SFmode on ISA 3.0.
26494 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
26495 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
26497 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
26498 SFmode value into a V4SF variable that was extracted from another
26499 V4SF variable without converting the element to double precision
26500 and back to single precision vector format.
26501 (vsx_insert_extract_v4sf_p9_2): Likewise.
26503 2017-06-19 Jakub Jelinek <jakub@redhat.com>
26505 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
26506 in UWHI to avoid undefined overflow.
26509 * ubsan.h (enum ubsan_encode_value_phase): New.
26510 (ubsan_encode_value): Change second argument to
26511 enum ubsan_encode_value_phase with default value of
26512 UBSAN_ENCODE_VALUE_GENERIC.
26513 * ubsan.c (ubsan_encode_value): Change second argument to
26514 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
26515 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
26516 create_tmp_var_raw instead of create_tmp_var and use a
26518 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
26519 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
26520 ubsan_encode_value callers.
26523 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
26524 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
26525 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
26527 2017-06-19 Richard Biener <rguenther@suse.de>
26529 PR middle-end/81118
26530 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
26531 estimates if we changed anything.
26533 2017-06-19 Richard Biener <rguenther@suse.de>
26535 PR tree-optimization/80887
26536 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
26537 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
26538 simplified lookups, then reset mprts_hook.
26539 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
26541 (try_to_simplify): Likewise.
26543 2017-06-19 Martin Liska <mliska@suse.cz>
26546 * gimplify.c (gimplify_switch_expr):
26547 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
26549 2017-06-19 Martin Liska <mliska@suse.cz>
26551 * doc/install.texi: Document that PGO runs in 4 stages.
26553 2017-06-19 Martin Liska <mliska@suse.cz>
26556 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
26557 to dispatcher function name.
26558 * multiple_target.c (replace_function_decl): New function.
26559 (create_dispatcher_calls): Redirect both edges and references.
26561 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
26563 * profile-count.c (profile_count::dump): Dump quality.
26564 (profile_count::differs_from_p): Update for unsigned val.
26565 * profile-count.h (profile_count_quality): New enum.
26566 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
26568 2017-06-19 Richard Biener <rguenther@suse.de>
26570 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
26571 struct function as arg.
26572 (estimate_numbers_of_iterations): Export overload with loop arg.
26573 (free_numbers_of_iterations_estimates_loop): Use an overload of
26574 free_numbers_of_iterations_estimates instead.
26575 * tree-cfg.c (remove_bb): Adjust.
26576 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
26577 * tree-parloops.c (gen_parallel_loop): Likewise.
26578 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
26580 (tree_unroll_loops_completely): Likewise.
26581 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
26582 Use an overload instead and export.
26583 (estimated_loop_iterations): Adjust.
26584 (max_loop_iterations): Likewise.
26585 (likely_max_loop_iterations): Likewise.
26586 (estimate_numbers_of_iterations): Take struct function as arg
26588 (loop_exits_before_overflow): Adjust.
26589 (free_numbers_of_iterations_estimates_loop): Use an overload.
26590 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
26591 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
26593 2017-06-19 Richard Biener <rguenther@suse.de>
26596 * ipa-prop.c (find_constructor_constant_at_offset): Handle
26597 RANGE_EXPR conservatively.
26599 2017-06-16 Carl Love <cel@us.ibm.com>
26601 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26602 definitions for vec_float, vec_float2, vec_floato,
26603 vec_floate built-ins.
26604 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
26605 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
26607 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
26608 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
26609 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
26610 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
26611 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
26612 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
26613 vec_floato): Add builtin defines.
26614 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
26615 Update the built-in documentation file for the new built-in
26618 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26620 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
26621 (mthumb): Mark as the negative of -marm.
26623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26625 * doc/invoke.texi (ARM Options, -mcpu): Document supported
26627 (ARM Options, -mtune): Document that this accepts the same
26628 extension options as -mcpu.
26629 (ARM Options, -mfpu): Document addition of -mfpu=auto.
26631 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26633 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
26634 permitted extensions.
26636 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26638 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
26639 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
26640 (armv8-m.main): Add option +nodsp.
26641 * config/arm/arm-cpu-cdata.h: Regenerated.
26643 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26645 * config/arm/t-fuchsia: New file.
26646 * config.gcc (arm*-*-fuchsia*): Use it.
26648 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26650 * config/arm/t-symbian: Rewrite for new option infrastructure.
26652 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26654 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
26655 (MULTILIB_REQUIRED): Likewise.
26657 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26659 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
26660 (MULTILIB_RESUE): Likewise.
26661 (MULTILIB_MATCHES): Likewise.
26662 (MULTLIB_REQUIRED): Likewise.
26664 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26666 * config/arm/t-rtems: Rewrite for new option framework.
26668 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26670 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
26671 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
26672 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
26673 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
26674 * config/arm/t-multilib: ... here.
26675 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
26676 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
26677 armv7-a and armv8*-a when A-profile libraries have not been built.
26678 * config/arm/t-rmprofile: Rewrite.
26680 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26682 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
26683 with a backslash. Remove the backslash after substituting unescaped
26685 * doc/fragments.texi (MULTILIB_REUSE): Document it.
26687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26689 * config.gcc: (arm*-*-*): When building a-profile libraries, force
26690 the driver to pass through the default setting of -mfloat-abi.
26691 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
26693 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
26694 (all_feat_combs): New rule.
26695 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
26697 * config/arm/t-aprofile: Rewrite.
26699 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26701 * config/arm/arm.h (FPUTYPE_AUTO): Define.
26702 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
26703 fpu is not specified by the user/command-line.
26704 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
26705 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
26706 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
26707 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
26708 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
26709 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
26711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26713 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
26714 * config/arm/t-arm-elf: Rewritten.
26716 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26718 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
26719 have some floating-point instructions.
26720 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
26721 (TARGET_MAYBE_HARD_FLOAT): New macro.
26722 * config/arm/arm-builtins.c (arm_init_builtins): Use
26723 TARGET_MAYBE_HARD_FLOAT.
26724 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
26726 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26728 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
26729 (configargs.h): Include it.
26730 (arm_print_hint_for_fpu_option): New function.
26731 (arm_parse_fpu_option): New function.
26732 (candidate_extension): New class.
26733 (arm_canon_for_multilib): New function.
26734 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
26735 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
26736 (ARCH_CANONICAL_SPECS): New macro.
26737 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
26739 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26741 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
26742 are set after handling multilib fragments. Set target_cpu_default2
26745 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26747 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
26749 (arm*-*-*): Set target_cpu_default2 to a quoted string.
26750 * config/arm/parsecpu.awk (check_cpu): Validate any extension
26752 (check_arch): Likewise.
26753 * config/arm/arm.c (arm_configure_build_target): Handle
26754 TARGET_CPU_DEFAULT being a string constant. Scan any feature
26755 options in the default.
26757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26759 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
26760 when an option is an alias of another.
26761 * config/arm/parsecpu.awk (optalias): New parser token.
26762 (gen_comm_data): Mark non-alias options as such. Emit entries
26763 for extension aliases.
26764 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
26765 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
26766 (armv6kz, armv6zk, armv6t2): Likewise.
26767 (armv7): Make vfpv3-d16 an alias.
26768 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26770 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26771 Sort in canonical order.
26772 (armv8-a): Sort in canonical order.
26773 (armv8.1-a, armv8.2-a): Likewise.
26774 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26776 (cortex-a9): Sort in canonical order.
26777 * config/arm/arm.c (selftests.h): Include it.
26778 (arm_test_cpu_arch_data): New function.
26779 (arm_run_self_tests): New function.
26780 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26781 (targetm): Move declaration to the end of the file.
26782 * arm-cpu-cdata.h: Regenerated.
26784 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26786 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26787 call to target_mode_check describing the type of option passed.
26788 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26789 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26790 arm_parse_cpu_option_name to match parameters against list of
26792 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26793 arm_arch_core_flags data structure.
26794 * config/arm/arm-cpu_cdata.h: Regenerated.
26796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26798 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26800 (arm_print_hint_for_cpu_option): Likewise.
26801 (arm_print_hint_for_arch_option): Likewise.
26802 (arm_parse_cpu_option_name): Likewise.
26803 (arm_parse_arch_option_name): Likewise.
26804 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26805 of entries in the all_fpus list.
26806 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26807 (arm_parse_cpu_option_name): Declare.
26808 (arm_parse_arch_option_name): Declare.
26809 (arm_parse_option_features): Declare.
26810 (arm_intialize_isa): Declare.
26811 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26813 (gen_comm_data): ... here. Make definitions non-static.
26814 * config/arm/arm-cpu-data.h: Regenerated.
26815 * config/arm/arm-cpu-cdata.h: Regenerated.
26817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26819 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26820 (cpu_arch_extension): New structure.
26821 (cpu_arch_option, arch_option, cpu_option): New structures.
26822 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26823 architecture types.
26824 (gen_data): Generate new format data tables.
26825 * config/arm/arm.c (cpu_tune): New structure.
26826 (cpu_option, processors): Delete.
26827 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26828 (arm_print_hint_for_cpu_option): ... this and ...
26829 (arm_print_hint_for_arch_option): ... this.
26830 (arm_parse_arch_cpu_name): Delete. Replace with ...
26831 (arm_parse_cpu_option_name): ... this and ...
26832 (arm_parse_arch_option_name): ... this.
26833 (arm_unrecognized_feature): Change type of target parameter to
26835 (arm_parse_arch_cpu_features): Delete. Replace with ...
26836 (arm_parse_option_features): ... this.
26837 (arm_configure_build_target): Rework to use new configuration data
26839 (arm_print_tune_info): Rework for new configuration data tables.
26840 * config/arm/arm-cpu-data.h: Regenerated.
26841 * config/arm/arm-cpu.h: Regenerated.
26843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26845 * Makefile.in (OBJS): Move sbitmap.o from here ...
26846 (OBJS-libcommon): ... to here.
26848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26850 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26851 (ISA_ALL_CRYPTO): New macro.
26852 (ISA_ALL_SIMD): New macro
26853 (ISA_ALL_FP): New macro.
26854 * config/arm/arm.c (fpu_bitlist): Update initializer.
26855 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26857 (arm9e): Add fpu. Add option for nofp
26858 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26859 (arm926ej-s, arm1026ej-s): Likewise.
26860 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26861 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26862 neon-fp16, neon-vfpv4, nofp and nosimd.
26863 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26864 (cortex-a8): Add fpu. Add option for nofp.
26865 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26866 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26867 (cortex-r4f): Add fpu.
26868 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26869 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26871 (cortex-r8): Likewise.
26872 (cortex-m4): Add fpu. Add option for nofp.
26873 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26874 (cortex-a17.cortex-a7): Likewise.
26875 (cortex-a32): Add fpu. Add options for crypto and nofp.
26876 (cortex-a35, cortex-a53): Likewise.
26877 (cortex-a57): Add fpu. Add option for crypto.
26878 (cortex-a72, cortex-a73): Likewise.
26879 (exynos-m1): Likewise.
26880 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26881 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26882 (cortex-m33): Add fpu. Add option for nofp.
26883 * config/arm/arm-cpu-cdata.h: Regenerated
26884 * config/arm/arm-cpu-data.h: Regenerated.
26886 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26888 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26889 (armv5te, armv5tej): Likewise.
26890 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26891 (armv7): Add options fp and vfpv3-d16.
26892 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26893 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26895 (armv7ve): Likewise.
26896 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26897 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26898 (armv8-a): Add nocrypto option.
26899 (armv8.1-a, armv8.2-a): Likewise.
26900 (armv8-m.main): add options fp, fp.dp and nofp.
26902 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26904 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26906 (armv8-a+crc): Delete.
26907 (armv8.1-a): Add options simd, crypto and nofp.
26908 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26909 (armv8.2-a+fp16): Delete.
26910 (armv8-m.main): Add option dsp.
26911 (armv8-m.main+dsp): Delete.
26912 (cortex-a8): Add fpu. Add nofp option.
26913 (cortex-a9): Add fpu. Add nofp and nosimd options.
26914 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26915 link to main cpu and architecture data structures.
26916 (gen_comm_data): Only put isa attributes from the main architecture
26918 (option): New statement for architecture and CPU entries.
26919 * arm.c (struct cpu_option): New structure.
26920 (struct processors): Add entry for options.
26921 (arm_unrecognized_feature): New function.
26922 (arm_parse_arch_cpu_name): Ignore any characters after the first
26924 (arm_parse_arch_cpu_feature): New function.
26925 (arm_configure_build_target): Separate out any CPU and architecture
26926 features and parse separately. Don't error out if -mfpu=auto is
26927 used with only an architecture string.
26928 (arm_print_asm_arch_directives): New function.
26929 (arm_file_start): Call it.
26930 * config/arm/arm-cpu-cdata.h: Regenerated.
26931 * config/arm/arm-cpu-data.h: Likewise.
26932 * config/arm/arm-tables.opt: Likewise.
26934 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26936 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26937 assembler when it is not -mfpu=auto.
26939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26941 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26942 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26943 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26944 (ASM_CPU_SPEC): Rewrite.
26945 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26946 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26947 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26948 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26949 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26950 copied string is NUL-terminated. Also strip any characters prefixed
26952 (arm_rewrite_selected_arch): New function.
26953 (arm_rewrite_march): New function.
26955 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26957 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26958 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26959 (march, mcpu, mtune): Convert to string-based options.
26960 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26961 (arm_parse_arch_cpu_name): New function.
26962 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26963 identify selected architecture or CPU.
26964 (arm_option_save): New function.
26965 (TARGET_OPTION_SAVE): Redefine.
26966 (arm_option_restore): Restore string options.
26967 (arm_option_print): Print string options.
26969 2017-06-16 Martin Sebor <msebor@redhat.com>
26971 PR tree-optimization/80933
26972 PR tree-optimization/80934
26973 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26974 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26975 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26976 (gimple_fold_builtin): Call them.
26978 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26980 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26981 as unlikely; update profile.
26983 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26985 * predict.c (force_edge_cold): Handle declaring edges impossible
26988 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26990 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26992 (try_unroll_loop_completely): Fix reporting.
26994 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26996 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26998 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27001 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27002 if given a non-constant argument for an intrinsic which requires a
27005 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27007 * profile.c (compare_freqs): New function.
27008 (branch_prob): Sort edge list.
27009 (find_spanning_tree): Assume that the list is priority sorted.
27011 2017-06-16 Richard Biener <rguenther@suse.de>
27013 PR tree-optimization/81090
27014 * passes.def (pass_record_bounds): Remove.
27015 * tree-pass.h (make_pass_record_bounds): Likewise.
27016 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27017 make_pass_record_bounds): Likewise.
27018 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27019 not free niter estimates at the beginning but at the end.
27020 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27022 2017-06-16 Richard Biener <rguenther@suse.de>
27024 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27025 initializer to workaround ICE in host GCC 4.8.
27027 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27029 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27031 (clone_inlined_nodes): Update.
27033 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27035 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27036 prefetch settings, and enable prefetching by default at -O3.
27038 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27040 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27041 Set flag_prefetch_loop_arrays according to tuning data.
27043 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27045 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27046 New tune structure.
27047 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27048 [Unrelated to main purpose of the patch] Place the pointer field last
27049 to enable type checking errors when tune structure are wrongly merged.
27050 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27051 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27052 (thunderx2t99_prefetch_tune): New tune constants.
27053 (tune_params *_tunings): Update all tunings (no functional change).
27054 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27055 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27056 from tunings structures.
27058 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27061 * ubsan.c (instrument_null): Add T argument, use it instead
27062 of computing it based on IS_LHS.
27063 (instrument_object_size): Likewise.
27064 (pass_ubsan::execute): Adjust instrument_null and
27065 instrument_object_size callers to pass gimple_get_lhs or
27066 gimple_assign_rhs1 result to it. Use instrument_null instead of
27067 calling get_base_address and instrument_mem_ref. Handle
27068 aggregate call arguments for object-size sanitization.
27070 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27072 PR tree-optimization/81089
27073 * tree-vrp.c (is_masked_range_test): Validate operands of
27076 2017-06-15 Martin Sebor <msebor@redhat.com>
27079 * dumpfile.c (dump_register): Avoid calling memset to initialize
27080 a class with a default ctor.
27081 * gcc.c (struct compiler): Remove const qualification.
27082 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27083 * hash-table.h: Ditto.
27084 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27086 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27087 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27089 * params.h (struct param_info): Make struct members non-const.
27090 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27091 with default initialization.
27092 * vec.h (vec_copy_construct, vec_default_construct): New helper
27094 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27095 with vec_copy_construct.
27096 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27097 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27098 * doc/invoke.texi (-Wclass-memaccess): Document.
27100 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27102 * emit-rtl.h (is_leaf): Update comment about local
27103 register allocator.
27105 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27108 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27109 for a variable to have a section before checking if the section has a
27111 Set section to.persistent if persistent attribute is set.
27112 Warn if .persistent attribute is used on an automatic variable.
27114 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27116 PR rtl-optimization/80474
27117 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27119 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27121 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27124 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27126 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27127 (casesi): Emit bounds checking as RTL.
27128 (casesi_internal_mips16_<mode>): Remove bounds checking.
27130 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27132 * config/xtensa/xtensa.c (xtensa_option_override): Append
27133 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27134 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27135 xtensa_doloop_hooks): Define unconditionally.
27136 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27138 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27139 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27140 for it in xtensa_option_override.
27141 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27142 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27144 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27146 * doc/cppopts.texi: Document '-' special value to -MF.
27148 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27150 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27151 (cortex_a53_fconst): Likewise.
27152 (cortex_a53_fpmul): Likewise.
27153 (cortex_a53_f_load_64): Likewise.
27154 (cortex_a53_f_load_many): Likewise.
27155 (cortex_a53_advsimd_alu): Likewise.
27156 (cortex_a53_advsimd_alu_q): Likewise.
27157 (cortex_a53_advsimd_mul): Likewise.
27158 (cortex_a53_advsimd_mul_q): Likewise.
27159 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27160 Add missing fmul, r2f_cvt and fconst cases.
27162 2017-06-14 Richard Biener <rguenther@suse.de>
27164 PR middle-end/81088
27165 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27167 (fold_binary_loc): When associating do not treat pre-existing
27168 TREE_OVERFLOW on literal constants as a reason to allow
27169 TREE_OVERFLOW on associated literal constants.
27171 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27173 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27174 (MASK_FEATURES): New macro.
27175 * config/sparc/sparc.c (sparc_option_override): Remove the special
27176 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27178 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27180 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27181 a division of 0 if non-call exceptions are enabled.
27183 2017-06-14 Andrew Pinski <apinski@cavium.com>
27184 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27187 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27188 Improve vector initialization code gen for only variable case.
27190 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27192 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27194 2017-06-14 Richard Biener <rguenther@suse.de>
27196 PR tree-optimization/81083
27197 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27200 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27202 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27203 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27204 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27205 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27206 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27207 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27209 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27211 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27212 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27214 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27216 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27218 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27220 * config/rs6000/t-rtems: Don't handle SPE.
27222 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27224 * config/rs6000/t-linux: Don't handle SPE.
27226 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27228 * config/rs6000/eabispe.h: Delete file.
27230 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27232 * config/rs6000/t-spe: Delete file.
27234 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27236 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27237 (rs6000_legitimate_offset_address_p): Return false for anything in
27238 V2SImode or V2SFmode.
27240 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27242 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27243 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27244 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27246 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27247 (rs6000_legitimate_offset_address_p): Ditto.
27248 (rs6000_emit_move): Ditto.
27249 (rs6000_init_builtins): Remove V4HI_type_node.
27251 2017-06-13 Martin Liska <mliska@suse.cz>
27254 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27255 (gate_asan): Likewise.
27256 * asan.h (asan_no_sanitize_address_p): Remove the function.
27257 (sanitize_flags_p): New function.
27258 * builtins.def: Fix coding style.
27259 * common.opt: Use renamed enum value.
27260 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27261 * doc/extend.texi: Document no_sanitize attribute.
27262 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27263 to SANITIZE_UNDEFINED_NONDEFAULT.
27264 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27265 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27266 Use sanitize_flags_p.
27267 * gimplify.c (gimplify_function_tree): Likewise.
27268 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27269 * opts.c (parse_no_sanitize_attribute): New function.
27270 (common_handle_option): Use renamed enum value.
27271 * opts.h (parse_no_sanitize_attribute): Declare.
27272 * tree.c (sanitize_flags_p): New function.
27273 * tree.h: Declared here.
27274 * tsan.c: Use sanitize_flags_p.
27275 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27276 (instrument_mem_ref): Likewise.
27277 (instrument_bool_enum_load): Likewise.
27278 (do_ubsan_in_current_function): Remove the function.
27279 (pass_ubsan::execute): Use sanitize_flags_p.
27280 * ubsan.h: Remove do_ubsan_in_current_function
27281 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27282 (dump_function_to_file): Use it here.
27284 2017-06-13 Martin Jambor <mjambor@suse.cz>
27286 PR tree-optimization/80803
27287 PR tree-optimization/81063
27288 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27289 (propagate_subaccesses_across_link): Enqueue subtree whenever
27290 necessary instead of relying on the caller.
27292 2017-06-13 Martin Jambor <mjambor@suse.cz>
27294 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27295 that have a first_link.
27296 (sort_and_splice_var_accesses): Do not check first_link before
27298 (subtree_mark_written_and_enqueue): Likewise.
27299 (propagate_all_subaccesses): Likewise and do not stop at first
27300 parent with a first_link.
27302 2017-06-13 Martin Jambor <mjambor@suse.cz>
27304 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27307 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27309 * match.pd: New pattern.
27311 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27313 * tree-vrp.c (is_masked_range_test): New function.
27314 (register_edge_assert_for): Determine ranges for
27317 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27319 PR tree-optimization/67328
27320 * fold-const.c (maskable_range_p): New function.
27321 (build_range_check): Generate bittests if possible.
27323 2017-06-13 Martin Liska <mliska@suse.cz>
27325 * gimple-pretty-print.c (dump_probability): Add new argument.
27326 (dump_edge_probability): Dump both probability and count.
27327 (dump_gimple_label): Likewise.
27328 (dump_gimple_bb_header): Likewise.
27330 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27333 * config/avr/avr-devices.c: Fix indentation.
27334 * config/avr/gen-avr-mmcu-specs.c: Dito.
27336 2017-06-13 Richard Biener <rguenther@suse.de>
27338 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27339 instead get vector type from stmt_info.
27340 (vectorizable_reduction): Adjust. Remove dead code.
27342 2017-06-13 Richard Biener <rguenther@suse.de>
27344 PR middle-end/81065
27345 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27346 case of C * (x * C2 + C3).
27347 (fold_addr_of_array_ref_difference): Properly fold index difference.
27349 2017-06-12 David S. Miller <davem@davemloft.net>
27352 * config/sparc/sparc.md (return expander): Emit frame blockage if
27353 function uses alloca.
27355 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27357 * combine.c (make_field_assignment): Check len rather than the mode
27358 precision when calling force_to_mode.
27360 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27362 Support multilibs and devices that see flash in RAM address range.
27365 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27366 (avr_mcu_t) <flash_pm_offset>: New field.
27367 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27368 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27369 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27370 (AVR_TINY_PM_OFFSET): Remove macro.
27371 * config/avr/avr.opt (-mshort-calls): New option.
27372 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27373 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27374 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27375 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27376 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27377 instead of avr_arch->have_jmp_call.
27378 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27379 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27380 avr_arch->flash_pm_offset to define.
27381 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27382 new field flash_pm_offset. Add entry for avrxmega3.
27383 (avr_texinfo): Add entry for avrxmega3.
27384 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27385 attiny212, attiny214,
27386 attiny412, attiny414, attiny416, attiny417,
27387 attiny814, attiny816, attiny817,
27388 attiny1614, attiny1616, attiny1617,
27389 attiny3214, attiny3216, attiny3217.
27390 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27391 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27392 (avr_print_operand_address) [AVR_TINY]: Same.
27393 (avr_asm_init_sections) <readonly_data_section>: Only patch
27394 callback if avr_arch->flash_pm_offset = 0.
27395 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27396 for rodata if avr_arch->flash_pm_offset != 0.
27397 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27398 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27399 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27400 (m_options): Append opt_rcall.
27401 (m_dirnames): Append dir_rcall.
27402 * config/avr/t-multilib: Regenerate.
27404 * configure.ac [target=avr]: Check whether avrxmega3 default
27405 linker description file works as needed.
27406 * configure: Regenerate.
27407 * doc/avr-mmcu.texi: Regenerate.
27408 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27409 <__AVR_ARCH__>: Document avrxmega3 and 103.
27410 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27411 <__AVR_SHORT_CALLS__>: Document it.
27412 <__AVR_PM_BASE_ADDRESS__>: Document it.
27413 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27414 (AVR Variable Attributes) <progmem>: Document this is
27415 not needed for avrxmega3.
27416 (AVR Named Address Spaces) <__flash>: Dito.
27418 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27420 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27422 2017-06-12 Doug Rupp <rupp@adacore.com>
27424 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27425 Append vxworks-stdint.h to the tm_file list.
27426 * config/vxworks-stdint.h: New file.
27428 2017-06-12 Martin Liska <mliska@suse.cz>
27430 PR tree-optimization/81041
27431 * tree-profile.c (gimple_gen_ic_func_profiler):
27432 Create an extra BB in profile-generate
27433 (gimple_gen_time_profiler): Likewise.
27435 2017-06-12 Jakub Jelinek <jakub@redhat.com>
27437 PR tree-optimization/81003
27438 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
27439 (update_range_test): Use it instead of force_gimple_operand_gsi.
27441 2017-06-12 Richard Biener <rguenther@suse.de>
27443 PR tree-optimization/81053
27444 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
27445 with backedge value not defined in loop. Simplify def stmt
27448 2017-06-11 Tom de Vries <tom@codesourcery.com>
27451 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
27453 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
27454 nvptx_cannot_force_const_mem.
27456 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27458 * opts.c (finish_options): Move test for flag_split_stack after
27459 it has been initialized.
27461 2017-06-11 Jason Merrill <jason@redhat.com>
27463 * tree.h (id_equal): New.
27464 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
27465 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
27466 instead of strcmp of IDENTIFIER_POINTER.
27468 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27470 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
27471 (mark_all_inlined_calls_cdtor): Fix formating.
27472 (inline_transform): Rescale profile before inlining.
27474 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27476 * cgraph.h (cgraph_edge::clone): Update prototype.
27477 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
27478 (cgraph_node::create_clone): Update.
27479 (cgraph_node::create_version_clone): Update.
27480 * tree-inline.c (copy_bb): Update.
27481 (expand_call_inline): Update.
27483 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27485 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
27486 factored out from ...
27487 (rs6000_emit_prologue): ... here.
27489 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
27491 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
27492 factored out from ...
27493 (rs6000_emit_prologue): ... here.
27495 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27497 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
27499 (handle_missing_profiles): Fix computation of tp_first_run.
27500 (counts_to_freqs): Do not touch freqs when count is 0.
27502 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
27504 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
27507 2017-06-10 Tom de Vries <tom@codesourcery.com>
27509 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27510 attributes): Document signal effective target.
27512 2017-06-10 Tom de Vries <tom@codesourcery.com>
27514 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27515 Document effective target stack_size.
27517 2017-06-09 David Malcolm <dmalcolm@redhat.com>
27519 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
27520 to the edit_context if they can be auto-applied.
27522 2017-06-9 Ian Lance Taylor <iant@golang.org>
27524 * opts.c (finish_options): If -fsplit-stack, disable implicit
27525 -forder-blocks-and-partition.
27526 * doc/invoke.texi (Optimize Options): Document that when using
27527 -fsplit-stack -forder-blocks-and-partition is not implicitly
27530 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27532 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
27533 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
27534 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
27535 * builtins.def (abort, trap, unreachable): Declare cold.
27536 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
27537 * tree-core.h (ECF_COLD): New.
27538 * tree.c (set_call_expr_flags): Handle ECF_COLD.
27539 (build_common_builtin_nodes): Mark unreachable and abort as cold.
27541 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27543 * predict.c (unlikely_executed_stmt_p): Cleanup.
27545 2017-06-09 Richard Biener <rguenther@suse.de>
27547 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
27548 model if the ref is always written to.
27550 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27552 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
27554 2017-06-09 Tamar Christina <tamar.christina@arm.com>
27556 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
27559 2017-06-09 Tom de Vries <tom@codesourcery.com>
27562 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
27563 "target cannot support label values" when encountering LABEL_REF.
27565 2017-06-09 Martin Liska <mliska@suse.cz>
27567 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
27568 (gimple_gen_ic_func_profiler): Emit direct comparison
27569 of __gcov_indirect_call_callee with NULL.
27570 (gimple_gen_time_profiler): Change probability from
27571 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
27573 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27575 * profile.c (edge_gcov_counts): Turn to pointer.
27576 (compute_branch_probabilities, compute_branch_probabilities): Update.
27577 (branch_prob): Do not clear edge_gcov_count.
27578 * profile.h (edge_gcov_counts): Turn to pointer.
27579 (edge_gcov_count): Update.
27581 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
27583 * gimple.h (gimple_check_failed): Mark cold.
27585 2017-06-09 Richard Biener <rguenther@suse.de>
27587 PR tree-optimization/66623
27588 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
27589 refactor check_reduction into two parts, properly computing
27590 whether we have to check reduction validity for outer loop
27593 2017-06-09 Richard Biener <rguenther@suse.de>
27595 PR tree-optimization/79483
27596 * graphite-scop-detection.c (order): New global.
27597 (get_order): Compute bb to order mapping that satisfies code
27598 generation constraints.
27599 (cmp_pbbs): New helper.
27600 (build_scops): Start domwalk at entry block, sort generated
27603 2017-06-09 Richard Biener <rguenther@suse.de>
27605 PR middle-end/81007
27606 * ipa-polymorphic-call.c
27607 (ipa_polymorphic_call_context::restrict_to_inner_class):
27608 Skip FIELD_DECLs with error_mark_node type.
27609 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
27612 2017-06-09 Martin Liska <mliska@suse.cz>
27614 * predict.c (struct branch_predictor): New struct.
27615 (test_prediction_value_range): New test.
27616 (predict_c_tests): New function.
27617 * selftest-run-tests.c (selftest::run_tests): Run the function.
27618 * selftest.h: Declare new tests.
27620 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
27623 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
27624 gen_add3_insn did not fail.
27625 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
27626 r0, construct that number in a temporary reg and add that reg to r0.
27627 If asked to put the result in r0 as well, fail.
27629 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
27631 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27632 for early expansion of vec_eqv.
27634 2017-06-08 Jakub Jelinek <jakub@redhat.com>
27636 PR middle-end/81005
27637 * ubsan.c (instrument_null): Avoid pointless code temporary.
27638 (pass_ubsan::execute): Instrument aggregate arguments of calls.
27640 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
27644 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
27647 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
27650 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27652 * predict.c (unlikely_executed_edge_p): Move ahead.
27653 (probably_never_executed_edge_p): Use it.
27655 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27657 PR middle-end/79988
27658 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
27659 gimple_call_builtin_p call.
27661 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27663 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
27664 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
27665 rtl_check_failed_type2, rtl_check_failed_code1,
27666 rtl_check_failed_code2, rtl_check_failed_code_mode,
27667 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
27668 rtvec_check_failed_bounds, rtl_check_failed_flag,
27669 _fatal_insn_not_found, _fatal_insn): Likewise.
27670 * tree.h (tree_contains_struct_check_failed,
27671 tree_check_failed, tree_not_check_failed,
27672 tree_class_check_failed, tree_range_check_failed,
27673 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
27674 tree_vec_elt_check_failed, phi_node_elt_check_failed,
27675 tree_operand_check_failed, omp_clause_check_failed,
27676 omp_clause_operand_check_failed, omp_clause_range_check_failed):
27679 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27681 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
27682 flag_branch_probabilities.
27683 * ipa-inline.c (edge_badness): Likewise.
27684 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
27685 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
27686 * predict.c (maybe_hot_frequency_p): Likewise.
27687 (probably_never_executed): Likewise.
27688 * sched-ebb.c (schedule_ebbs): Likewise.
27689 * sched-rgn.c (find_single_block_region): Likewise.
27690 * tracer.c (tail_duplicate): Likewise.
27692 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27694 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
27695 longer requires x_flag_profile_use.
27697 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27699 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
27700 instead of flag_reorder_blocks_and_partition.
27701 * dbxout.c (dbxout_function_end): Likewise.
27702 * dwarf2out.c (gen_subprogram_die): Likewise.
27703 * haifa-sched.c (sched_create_recovery_edges): Likewise.
27704 * hw-doloop.c (reorg_loops): Likewise.
27705 * varasm.c (assemble_start_function,
27706 assemble_end_function): Likewise.
27707 (decide_function_section): Do not check for
27708 flag_reorder_blocks_and_partition.
27710 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
27712 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27714 (chkp_get_hard_register_fake_addr_expr): Ditto.
27715 (chkp_build_addr_expr): Add check for hard reg case.
27716 (chkp_parse_array_and_component_ref): Ditto.
27717 (chkp_find_bounds_1): Ditto.
27718 (chkp_process_stmt): Don't generate bounds store for
27721 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27723 * predict.c (maybe_hot_bb_p): Do not check profile status.
27724 (maybe_hot_edge_p): Likewise.
27725 (probably_never_executed): Check for zero counts even if profile
27727 (unlikely_executed_edge_p): New function.
27728 (unlikely_executed_stmt_p): New function.
27729 (unlikely_executed_bb_p): New function.
27730 (set_even_probabilities): Use unlikely predicates.
27731 (combine_predictions_for_bb): Likewise.
27732 (predict_paths_for_bb): Likewise.
27733 (predict_paths_leading_to_edge): Likewise.
27734 (determine_unlikely_bbs): New function.
27735 (estimate_bb_frequencies): Use it.
27736 (compute_function_frequency): Use zero counts even if profile is
27738 * profile-count.h: Fix typo.
27740 2017-08-08 Julia Koval <julia.koval@intel.com>
27742 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
27743 _mm512_mask_cvtsepi16_storeu_epi8,
27744 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
27745 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
27746 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
27747 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
27748 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
27749 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
27750 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
27751 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
27752 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
27753 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
27754 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
27755 __builtin_ia32_pmovuswb256mem_mask,
27756 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
27757 __builtin_ia32_pmovwb512mem_mask): New builtins.
27759 2017-08-08 Julia Koval <julia.koval@intel.com>
27761 PR target/73350,80862
27762 * config/i386/subst.md (round): Fix round pattern.
27763 * config/i386/i386.c (ix86_erase_embedded_rounding):
27764 Fix erasing rounding for the fixed pattern.
27766 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
27768 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27770 2017-06-08 Martin Liska <mliska@suse.cz>
27772 PR gcov-profile/80911
27773 * gcov.c (block_info::block_info): New constructor.
27775 2017-06-07 Carl Love <cel@us.ibm.com>
27777 * config/rs6000/rs6000-c: The return type of the following
27778 built-in functions was implemented as int not long long. Fix sign
27779 of return value for the unsigned version of vec_mulo and vec_mule.
27780 vector unsigned long long vec_bperm (vector unsigned long long,
27781 vector unsigned char)
27782 vector signed long long vec_mule (vector signed int,
27784 vector unsigned long long vec_mule (vector unsigned int,
27785 vector unsigned int)
27786 vector signed long long vec_mulo (vector signed int,
27788 vector unsigned long long vec_mulo (vector unsigned int,
27789 vector unsigned int)
27790 * doc/extend.texi: Fix the documentation for the built-in
27793 2017-06-07 Carl Love <cel@us.ibm.com>
27796 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27799 2017-06-07 Carl Love <cel@us.ibm.com>
27801 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27802 support, Generate doublehv for signed int/float for BE case only.
27804 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27806 * doc/invoke.texi (mcx16): Rewrite.
27808 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27810 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27811 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27812 *mov<mode>_softfloat, and an anonymous splitter): Use
27813 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27815 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27817 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27819 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27820 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27821 (rs6000_debug_reg_global): Adjust.
27822 (rs6000_init_hard_regno_mode_ok): Adjust.
27823 (rs6000_dbx_register_number): Adjust.
27824 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27825 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27826 Remove SPE_ACC and SPEFSCR.
27827 (REG_ALLOC_ORDER): Ditto.
27828 (FRAME_POINTER_REGNUM): Change to 111.
27829 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27830 (REG_CLASS_NAMES): Ditto.
27831 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27832 (REGISTER_NAMES): Ditto.
27833 (ADDITIONAL_REG_NAMES): Ditto.
27834 (rs6000_reg_names): Ditto.
27835 * config/rs6000/rs6000.md: Renumber some register number
27838 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27840 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27842 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27843 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27845 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27846 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27847 Delete the SPE high registers.
27848 (REG_ALLOC_ORDER): Ditto.
27849 (enum reg_class): Remove SPE_HIGH_REGS.
27850 (REG_CLASS_NAMES): Ditto.
27851 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27852 (REGISTER_NAMES): Ditto.
27853 (rs6000_reg_names): Ditto.
27854 * doc/tm.texi.in: Remove SPE as example.
27855 * doc/tm.texi: Regenerate.
27857 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27859 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27860 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27861 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27862 * config/rs6000/rs6000.md (type): Remove "brinc".
27864 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27866 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27867 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27868 * config/rs6000/linuxspe.h: Delete file.
27869 * config/rs6000/rs6000.md: Don't include spe.md.
27870 * config/rs6000/spe.h: Delete file.
27871 * config/rs6000/spe.md: Delete file.
27872 * config/rs6000/t-rs6000: Remove spe.md.
27874 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27876 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27877 (reg_or_none500mem_operand): Delete.
27878 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27879 instead of reg_or_none500mem_operand.
27881 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27883 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27884 handling of SPE flags.
27885 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27887 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27889 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27891 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27892 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27893 paired_divv2sf3): Similar.
27894 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27895 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27896 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27898 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27899 Rename the paired_* instruction patterns.
27900 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27902 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27903 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27904 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27905 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27906 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27907 PAIRED_VECTOR_MODE.
27908 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27909 (spe_func_has_64bit_regs_p): Delete.
27910 (spe_expand_predicate_builtin): Delete.
27911 (spe_expand_evsel_builtin): Delete.
27912 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27913 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27914 (invalid_e500_subreg): Delete.
27915 (rs6000_legitimize_address): Always force_reg op2 as well, for
27916 paired single memory accesses.
27917 (rs6000_member_type_forces_blk): Delete.
27918 (rs6000_spe_function_arg): Delete.
27919 (rs6000_expand_unop_builtin): Delete SPE handling.
27920 (rs6000_expand_binop_builtin): Ditto.
27921 (spe_expand_stv_builtin): Delete.
27922 (bdesc_2arg_spe): Delete.
27923 (spe_expand_builtin): Delete.
27924 (spe_expand_predicate_builtin): Delete.
27925 (spe_expand_evsel_builtin): Delete.
27926 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27927 (spe_init_builtins): Delete.
27928 (spe_func_has_64bit_regs_p): Delete.
27929 (savres_routine_name): Delete "info" parameter. Adjust callers.
27930 (rs6000_emit_stack_reset): Ditto.
27931 (rs6000_dwarf_register_span): Delete.
27932 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27933 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27934 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27936 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27938 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27939 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27940 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27941 mulv2sf3, divv2sf3): Delete expanders.
27943 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27945 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27947 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27949 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27950 * config/rs6000/rs6000.c: Ditto.
27952 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27954 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27955 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27956 comparison_operator.
27958 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27960 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27961 * config/rs6000/rs6000.opt: Ditto.
27962 * config/rs6000/t-rtems: Ditto.
27964 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27966 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27967 TARGET_E500_SINGLE by 0, simplify.
27968 * config/rs6000/rs6000.c: Ditto.
27969 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27970 (spe_build_register_parallel): Delete.
27971 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27972 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27973 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27974 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27975 (E500_CONVERT): Delete.
27976 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27978 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27980 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27981 * config/rs6000/dfp.md: Ditto.
27982 (negdd2, *negdd2_fpr): Merge.
27983 (absdd2, *absdd2_fpr): Merge.
27984 (negtd2, *negtd2_fpr): Merge.
27985 (abstd2, *abstd2_fpr): Merge.
27986 * config/rs6000/e500.h: Delete file.
27987 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27988 TARGET_FPRS by 1 and simplify.
27989 * config/rs6000/rs6000-c.c: Ditto.
27990 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27991 TARGET_DF_SPE by 0.
27992 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27994 * config/rs6000/rs6000.md: Ditto.
27995 (floatdidf2, *floatdidf2_fpr): Merge.
27996 (move_from_CR_gt_bit): Delete.
27997 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27998 (E500_CR_IOR_COMPARE): Delete.
27999 (All patterns that require !TARGET_FPRS): Delete.
28000 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28002 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28004 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28006 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28008 * graphds.c (add_edge): Intitialize edge's attached data.
28009 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28010 pointer parameter. Call pointed function on each edge during
28011 graph traversing. Skip traversing the edge when the function
28013 (graphds_dfs, graphds_scc): Ditto.
28014 (for_each_edge): New parameter. Pass the new parameter to callback
28016 * graphds.h (skip_edge_callback): New function pointer type.
28017 (graphds_dfs, graphds_scc): New function pointer parameter.
28018 (graphds_edge_callback, for_each_edge): New parameter.
28020 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28022 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28023 out code checking if runtime alias check is possible to below ...
28024 Call the new function.
28025 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28026 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28028 2017-06-07 Marek Polacek <polacek@redhat.com>
28031 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28032 TYPE_OVERFLOW_WRAPS check.
28034 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28036 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28037 if versioning is required.
28038 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28039 peeling with the check for versioning.
28041 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28043 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28044 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28045 Set true to new parameter if new ssa variable is defined.
28046 (vect_gen_vector_loop_niters): Refactor. Set range information
28047 for the new vector loop bound variable.
28048 (vect_do_peeling): Ditto.
28050 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28052 * tree-affine.c (ssa.h): Include header file.
28053 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28054 has wrapping overflow behavior.
28056 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28058 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28060 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28062 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28063 (tree_to_aff_combination): ... here.
28065 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28067 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28068 reg_pressure model function.
28069 (ivopts_global_cost_for_size): Delete.
28070 (determine_set_costs, iv_ca_recount_cost): Call new model function
28071 ivopts_estimate_reg_pressure.
28073 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28075 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28076 expensive than udiv. Remove floating point cases from mod.
28078 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28080 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28081 Increase idiv cost.
28083 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28085 * config/aarch64/aarch64.md
28086 (copysignsf3): Fix mask generation.
28088 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28090 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28092 (class dump_manager): Add register_dumps method.
28093 * dumpfile.c: Include langhooks.h.
28094 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28095 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28096 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28097 (dump_manager::dump_register): Start with 512 entries instead of 32.
28098 (dump_manager::register_dumps): New method.
28099 * toplev.c (general_init): Instead of invoking register_dumps
28100 langhook, invoke register_dumps method on the dump manager.
28101 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28104 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28106 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28107 Say that # requires an associated define_split to exist, and that
28108 the define_split must be suitable for use after register allocation.
28110 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28112 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28113 (compute_outgoing_frequencies): Also initialize zero counts.
28114 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28115 around loops; preserve more of profile when nothing changes.
28117 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28119 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28121 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28122 * config/arm/arm-cpu-cdata.h: Regenerate.
28123 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28124 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28125 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28126 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28127 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28129 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28131 * config/arm/t-rmprofile: Likewise.
28132 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28134 2017-06-06 David S. Miller <davem@davemloft.net>
28137 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28138 blockage if function uses alloca.
28140 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28142 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28143 New "uid" fields to hold pretty-print IDs of group and ref.
28144 Memory references are now identified as <group_id>:<ref_id>
28145 instead of using [random] addresses.
28146 (dump_mem_details): Simplify, no functional change.
28147 (dump_mem_ref): Simplify and make output more concise.
28148 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28149 (find_or_create_group): Initialize group uid.
28150 (record_ref): Initialize ref uid. Improve debug output.
28151 (prune_group_by_reuse, should_issue_prefetch_p,)
28152 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28153 (mark_nontemporal_store, determine_loop_nest_reuse):
28154 Improve debug output.
28156 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28158 * dbgcnt.def (prefetch): New debug counter.
28159 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28160 (schedule_prefetches): Stop issueing prefetches if debug counter
28163 2017-06-06 Tom de Vries <tom@codesourcery.com>
28165 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28166 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28168 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28170 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28171 Use aarch64_reg_or_zero predicate for operand 4.
28172 (aarch64_compare_and_swap<mode> define_insn_and_split):
28173 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28174 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28176 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28178 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28179 (arm_compute_save_core_reg_mask): This.
28180 (thumb1_compute_save_reg_mask): Rename into ...
28181 (thumb1_compute_save_core_reg_mask): This.
28182 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28183 (arm_compute_frame_layout): Likewise.
28185 2017-06-06 Richard Biener <rguenther@suse.de>
28187 PR tree-optimization/80974
28188 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28189 keep or clear leaders SSA info.
28191 2017-06-06 Tom de Vries <tom@codesourcery.com>
28193 * config/nvptx/nvptx.c (split_mode_p): New function.
28194 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28196 2017-06-06 Tom de Vries <tom@codesourcery.com>
28198 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28200 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28203 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28206 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28208 * shrink-wrap.c (handle_simple_exit): Update profile.
28209 (try_shrink_wrapping): Upate profile.
28211 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28213 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28214 (tree_guess_outgoing_edge_probabilities): New.
28215 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28216 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28218 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28220 * ipa-split.c (split_function): Initialize return bb profile.
28222 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28224 * profile.c (compute_branch_probabilities): Also initialize
28225 EXIT_BLOCK profile.
28227 2017-06-06 Richard Biener <rguenther@suse.de>
28229 PR tree-optimization/80928
28230 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28231 (vect_analyze_loop_operations): Properly guard analysis for
28233 (vect_transform_loop): Likewise.
28234 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28235 (vect_model_induction_cost): Do not cost for pure SLP.
28236 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28237 of induction in inner loop vectorization.
28238 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28239 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28240 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28242 (vect_analyze_slp_cost_1): Cost induction.
28243 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28244 (vect_get_slp_vect_defs): Likewise.
28245 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28246 (vect_transform_stmt): Handle SLP reductions.
28247 * tree-vectorizer.h (vectorizable_induction): Adjust.
28249 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28251 * config/rs6000/rs6000.c (make_resolver_func): Update
28252 init_lowered_empty_function call.
28254 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28256 * doc/invoke.texi: Document the -fprofile-abs-path option.
28257 * common.opt (fprofile-abs-path): New option.
28258 * gcov-io.h (gcov_write_filename): Declare.
28259 * gcov-io.c (gcov_write_filename): New function.
28260 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28261 * profile.c (output_location): Likewise.
28263 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28265 * shring-wrap.c: Revert accidental commit.
28267 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28269 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28271 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28273 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28275 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28276 profile in callgraph edge.
28277 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28278 (apply_scale): Likewise.
28279 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28280 Also scale profile when inlining function with zero profile.
28281 (initialize_cfun): Update exit block profile even when it is zero.
28282 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28283 when profile is read.
28285 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28287 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28288 (CLONE_*): New constants to define the processors we can generate
28289 code for with the target_clone attribute.
28290 (rs6000_clone_map): New array to identify which clone processors
28291 the current program is running on.
28292 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28293 target_clone attribute.
28294 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28295 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28296 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28297 (cpu_expand_builtin): Add support for target_clone attribute.
28298 (rs6000_valid_attribute_p): Allow "default" attribute.
28299 (get_decl_name): New debug function to simplify printing the
28300 current function name in debugging statements.
28301 (rs6000_clone_priority): New functions to support the target_clone
28302 attribute, and be able to generate code to switch between ISA 2.05
28303 through ISA 3.0 (power6 through power9).
28304 (rs6000_compare_version_priority): Likewise.
28305 (rs6000_get_function_versions_dispatcher): Likewise.
28306 (make_resolver_func): Likewise.
28307 (add_condition_to_bb): Likewise.
28308 (dispatch_function_versions): Likewise.
28309 (rs6000_generate_version_dispatcher_body): Likewise.
28310 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28311 (fusion_gpr_load_p): Fix a spacing issue.
28312 * doc/extend.texi (Common Function Attributes): Document that the
28313 PowerPC supports the target_clone attribute.
28315 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28317 * config/arm/arm.h: explain F symbol found in description of ARM
28318 register allocation in its legend.
28320 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28322 * config/mips/frame-header-opt.c: Include profile-count.h.
28323 * config/riscv/riscv.c: Include profile-count.h
28325 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28327 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28329 (sm_set_flag_if_changed): Add bbs field.
28330 (execute_sm_if_changed_flag_set): Pass BBS.
28331 (execute_sm): Update.
28333 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28335 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28338 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28340 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28341 (peephole2): New peephole2 to emit the above.
28342 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28344 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28346 * config/aarch64/aarch64.c (define_peephole2 above
28347 *sub_<shift>_<mode>): New peephole.
28349 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28351 * config/i386/i386.c (make_resolver_func): Update.
28352 * Makefile.in: Add profile-count.h and profile-count.o
28353 * auto-profile.c (afdo_indirect_call): Update to new API.
28354 (afdo_set_bb_count): Update.
28355 (afdo_propagate_edge): Update.
28356 (afdo_propagate_circuit): Update.
28357 (afdo_calculate_branch_prob): Update.
28358 (afdo_annotate_cfg): Update.
28359 * basic-block.h: Include profile-count.h
28360 (struct edge_def): Turn count to profile_count.
28361 (struct basic_block_def): Likewie.
28362 (REG_BR_PROB_BASE): Move to profile-count.h
28363 (RDIV): Move to profile-count.h
28364 * bb-reorder.c (max_entry_count): Turn to profile_count.
28365 (find_traces): Update.
28366 (rotate_loop):Update.
28367 (connect_traces):Update.
28368 (sanitize_hot_paths):Update.
28369 * bt-load.c (migrate_btr_defs): Update.
28370 * cfg.c (RDIV): Remove.
28371 (init_flow): Use alloc_block.
28372 (alloc_block): Uninitialize count.
28373 (unchecked_make_edge): Uninitialize count.
28374 (check_bb_profile): Update.
28375 (dump_edge_info): Update.
28376 (dump_bb_info): Update.
28377 (update_bb_profile_for_threading): Update.
28378 (scale_bbs_frequencies_int): Update.
28379 (scale_bbs_frequencies_gcov_type): Update.
28380 (scale_bbs_frequencies_profile_count): New.
28381 * cfg.h (update_bb_profile_for_threading): Update.
28382 (scale_bbs_frequencies_profile_count): Declare.
28383 * cfgbuild.c (compute_outgoing_frequencies): Update.
28384 (find_many_sub_basic_blocks): Update.
28385 * cfgcleanup.c (try_forward_edges): Update.
28386 (try_crossjump_to_edge): Update.
28387 * cfgexpand.c (expand_gimple_tailcall): Update.
28388 (construct_exit_block): Update.
28389 * cfghooks.c (verify_flow_info): Update.
28390 (dump_bb_for_graph): Update.
28391 (split_edge): Update.
28392 (make_forwarder_block): Update.
28393 (duplicate_block): Update.
28394 (account_profile_record): Update.
28395 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28396 (get_estimated_loop_iterations): Update.
28397 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28398 (single_likely_exit): Update.
28399 * cfgloopmanip.c (scale_loop_profile): Update.
28401 (set_zero_probability): Update.
28402 (lv_adjust_loop_entry_edge): Update.
28403 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28404 (purge_dead_edges): Update.
28405 (rtl_account_profile_record): Update.
28406 * cgraph.c (cgraph_node::create): Uninitialize count.
28407 (symbol_table::create_edge): Uninitialize count.
28408 (cgraph_update_edges_for_call_stmt_node): Update.
28409 (cgraph_edge::dump_edge_flags): Update.
28410 (cgraph_node::dump): Update.
28411 (cgraph_edge::maybe_hot_p): Update.
28412 * cgraph.h: Include profile-count.h
28413 (create_clone), create_edge, create_indirect_edge): Update.
28414 (cgraph_node): Turn count to profile_count.
28415 (cgraph_edge0: Likewise.
28416 (make_speculative, clone): Update.
28417 (create_edge): Update.
28418 (init_lowered_empty_function): Update.
28419 * cgraphclones.c (cgraph_edge::clone): Update.
28420 (duplicate_thunk_for_node): Update.
28421 (cgraph_node::create_clone): Update.
28422 * cgraphunit.c (cgraph_node::analyze): Update.
28423 (cgraph_node::expand_thunk): Update.
28424 * final.c (dump_basic_block_info): Update.
28425 * gimple-streamer-in.c (input_bb): Update.
28426 * gimple-streamer-out.c (output_bb): Update.
28427 * graphite.c (print_global_statistics): Update.
28428 (print_graphite_scop_statistics): Update.
28429 * hsa-brig.c: Include basic-block.h.
28430 * hsa-dump.c: Include basic-block.h.
28431 * hsa-gen.c (T sum_slice): Update.
28432 (convert_switch_statements):Update.
28433 * hsa-regalloc.c: Include basic-block.h.
28434 * ipa-chkp.c (chkp_produce_thunks): Update.
28435 * ipa-cp.c (struct caller_statistics): Update.
28436 (init_caller_stats): Update.
28437 (gather_caller_stats): Update.
28438 (ipcp_cloning_candidate_p): Update.
28439 (good_cloning_opportunity_p): Update.
28440 (get_info_about_necessary_edges): Update.
28441 (dump_profile_updates): Update.
28442 (update_profiling_info): Update.
28443 (update_specialized_profile): Update.
28444 (perhaps_add_new_callers): Update.
28445 (decide_about_value): Update.
28446 (ipa_cp_c_finalize): Update.
28447 * ipa-devirt.c (struct odr_type_warn_count): Update.
28448 (struct decl_warn_count): Update.
28449 (struct final_warning_record): Update.
28450 (possible_polymorphic_call_targets): Update.
28451 (ipa_devirt): Update.
28452 * ipa-fnsummary.c (redirect_to_unreachable): Update.
28453 * ipa-icf.c (sem_function::merge): Update.
28454 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
28455 * ipa-inline.c (compute_uninlined_call_time): Update.
28456 (compute_inlined_call_time): Update.
28457 (want_inline_small_function_p): Update.
28458 (want_inline_self_recursive_call_p): Update.
28459 (edge_badness): Update.
28460 (lookup_recursive_calls): Update.
28461 (recursive_inlining): Update.
28462 (inline_small_functions): Update.
28463 (dump_overall_stats): Update.
28464 (dump_inline_stats): Update.
28465 * ipa-profile.c (ipa_profile_generate_summary): Update.
28466 (ipa_propagate_frequency): Update.
28467 (ipa_profile): Update.
28468 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28469 * ipa-utils.c (ipa_merge_profiles): Update.
28470 * loop-doloop.c (doloop_modify): Update.
28471 * loop-unroll.c (report_unroll): Update.
28472 (unroll_loop_runtime_iterations): Update.
28473 * lto-cgraph.c (lto_output_edge): Update.
28474 (lto_output_node): Update.
28475 (input_node): Update.
28476 (input_edge): Update.
28477 (merge_profile_summaries): Update.
28478 * lto-streamer-in.c (input_cfg): Update.
28479 * lto-streamer-out.c (output_cfg): Update.
28480 * mcf.c (create_fixup_graph): Update.
28481 (adjust_cfg_counts): Update.
28482 (sum_edge_counts): Update.
28483 * modulo-sched.c (sms_schedule): Update.
28484 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
28485 * predict.c (maybe_hot_count_p): Update.
28486 (probably_never_executed): Update.
28487 (dump_prediction): Update.
28488 (combine_predictions_for_bb): Update.
28489 (propagate_freq): Update.
28490 (handle_missing_profiles): Update.
28491 (counts_to_freqs): Update.
28492 (rebuild_frequencies): Update.
28493 (force_edge_cold): Update.
28494 * predict.h: Include profile-count.h
28495 (maybe_hot_count_p, counts_to_freqs): UPdate.
28496 * print-rtl-function.c: Do not include cfg.h
28497 * print-rtl.c: Include basic-block.h
28498 * profile-count.c: New file.
28499 * profile-count.h: New file.
28500 * profile.c (is_edge_inconsistent): Update.
28501 (correct_negative_edge_counts): Update.
28502 (is_inconsistent): Update.
28503 (set_bb_counts): Update.
28504 (read_profile_edge_counts): Update.
28505 (compute_frequency_overlap): Update.
28506 (compute_branch_probabilities): Update; Initialize and deinitialize
28508 (branch_prob): Update.
28509 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
28510 (edge_gcov_count): New.
28511 (bb_gcov_count): New.
28512 * shrink-wrap.c (try_shrink_wrapping): Update.
28513 * tracer.c (better_p): Update.
28514 * trans-mem.c (expand_transaction): Update.
28515 (ipa_tm_insert_irr_call): Update.
28516 (ipa_tm_insert_gettmclone_call): Update.
28517 * tree-call-cdce.c: Update.
28518 * tree-cfg.c (gimple_duplicate_sese_region): Update.
28519 (gimple_duplicate_sese_tail): Update.
28520 (gimple_account_profile_record): Update.
28521 (execute_fixup_cfg): Update.
28522 * tree-inline.c (copy_bb): Update.
28523 (copy_edges_for_bb): Update.
28524 (initialize_cfun): Update.
28525 (freqs_to_counts): Update.
28526 (copy_cfg_body): Update.
28527 (expand_call_inline): Update.
28528 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
28529 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
28530 (try_unroll_loop_completely): Update.
28531 (try_peel_loop): Update.
28532 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
28533 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
28534 * tree-ssa-loop-split.c (connect_loops): Update.
28535 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
28536 * tree-ssa-reassoc.c (branch_fixup): Update.
28537 * tree-ssa-tail-merge.c (replace_block_by): Update.
28538 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
28539 (compute_path_counts): Update.
28540 (update_profile): Update.
28541 (recompute_probabilities): Update.
28542 (update_joiner_offpath_counts): Update.
28543 (estimated_freqs_path): Update.
28544 (freqs_to_counts_path): Update.
28545 (clear_counts_path): Update.
28546 (ssa_fix_duplicate_block_edges): Update.
28547 (duplicate_thread_path): Update.
28548 * tree-switch-conversion.c (case_bit_test_cmp): Update.
28549 (struct switch_conv_info): Update.
28550 * tree-tailcall.c (decrease_profile): Update.
28551 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
28552 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
28553 * value-prof.c (check_counter): Update.
28554 (gimple_divmod_fixed_value): Update.
28555 (gimple_mod_pow2): Update.
28556 (gimple_mod_subtract): Update.
28557 (gimple_ic_transform): Update.
28558 (gimple_stringop_fixed_value): Update.
28559 * value-prof.h (gimple_ic): Update.
28561 2017-06-02 Carl Love <cel@us.ibm.com>
28563 * config/rs6000/rs6000-c: Add support for built-in functions
28564 vector double vec_doublee (vector signed int);
28565 vector double vec_doublee (vector unsigned int);
28566 vector double vec_doublee (vector float);
28567 vector double vec_doubleh (vector signed int);
28568 vector double vec_doubleh (vector unsigned int);
28569 vector double vec_doubleh (vector float);
28570 vector double vec_doublel (vector signed int);
28571 vector double vec_doublel (vector unsigned int);
28572 vector double vec_doublel (vector float);
28573 vector double vec_doubleo (vector signed int);
28574 vector double vec_doubleo (vector unsigned int);
28575 vector double vec_doubleo (vector float);.
28576 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
28577 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
28579 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
28580 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
28581 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
28583 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
28584 vec_doublel, vec_doubleh.
28585 * doc/extend.texi: Update the built-in documentation file for the
28586 new built-in functions.
28588 2017-06-02 David Malcolm <dmalcolm@redhat.com>
28591 * ipa-inline-analysis.c (free_growth_caches): Set
28592 edge_removal_hook_holder to NULL after removing it.
28594 2017-06-02 Sudakshina Das <sudi.das@arm.com>
28596 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
28597 comparision with zero.
28599 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28600 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28601 for early expansion of vec_min and vec_max builtins.
28602 (builtin_function_type): Add min/max unsigned variants to those
28603 identified as having unsigned arguments.
28605 2017-06-02 Olivier Hainque <hainque@adacore.com>
28607 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
28609 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28611 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
28612 Use VALL_F16 iterator rather than VALL.
28614 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28616 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
28617 Emit CBNZ inside loop when doing a strong exchange and comparing
28618 against zero. Generate the CC flags after the loop.
28620 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
28622 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
28623 (dl_section_ref): New.
28624 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
28625 On AIX, append an expression to subtract the size of the
28626 section length to dl_section_ref.
28628 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
28630 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28631 for early expansion of vector absolute builtins.
28633 2017-06-02 Richard Biener <rguenther@suse.de>
28635 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
28636 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
28638 2017-06-02 Richard Biener <rguenther@suse.de>
28640 PR tree-optimization/80948
28641 * tree-tailcall.c (find_tail_calls): Track stmts to move in
28642 stmt order as well.
28644 2017-06-02 Richard Biener <rguenther@suse.de>
28646 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
28648 * tree-vect-stmts.c (process_use): Do not mark backedge defs
28649 for inductions as relevant.
28651 2017-06-02 Richard Biener <rguenther@suse.de>
28653 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
28654 (vectorizable_induction): ... this. Remove dead code.
28656 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
28658 * builtins. (expand_builtin_alloca): Remove second parameter and
28659 infer its value from the first parameter instead.
28660 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
28662 2017-06-02 Jakub Jelinek <jakub@redhat.com>
28664 PR rtl-optimization/80903
28665 * loop-doloop.c (add_test): Unshare sequence.
28667 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
28669 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
28671 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
28673 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
28675 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
28676 xlogue_layout::get_instance, logue_layout::xlogue_layout,
28677 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
28678 (xlogue_layout::get_stub_rtx): Make static.
28679 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
28680 (xlogue_layout::compute_stub_managed_regs): Rename to...
28681 (xlogue_layout::count_stub_managed_regs): ...this.
28682 (xlogue_layout::is_stub_managed_reg): New function.
28683 (xlogue_layout::m_stub_names): Rename to...
28684 (xlogue_layout::s_stub_names): ...this, make static.
28685 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
28686 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
28687 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
28688 xlogue_layout::s_stub_names): Instantiate statics.
28689 (stub_managed_regs): Remove.
28690 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
28691 (disable_call_ms2sysv_xlogues): Rename to...
28692 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
28693 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
28695 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
28696 change after reload_completed.
28697 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
28699 (ix86_expand_prologue): Likewise.
28700 (ix86_expand_epilogue): Likewise.
28701 (ix86_expand_split_stack_prologue): Likewise.
28702 (ix86_compute_frame_layout): Remove frame parameter ...
28703 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
28704 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
28706 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
28707 (ix86_frame): Move from here ...
28708 * config/i386/i386.h (ix86_frame): ... to here.
28709 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
28710 complete ix86_frame data structure instead. Remove some_ld_name.
28712 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28714 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
28715 symbols that hold a DECL_VALUE_EXPR.
28717 2017-06-01 Martin Jambor <mjambor@suse.cz>
28719 PR tree-optimization/80898
28720 * tree-sra.c (process_subtree_disqualification): Removed.
28721 (disqualify_candidate): Do not acll
28722 process_subtree_disqualification.
28723 (subtree_mark_written_and_enqueue): New function.
28724 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
28725 RHS has been disqualified and re-queue LHS if necessary. Apart
28726 from that, ignore disqualified RHS.
28728 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28730 * config/s390/s390.c (s390_emit_epilogue): Disable early return
28731 address fetch for z10 or later.
28733 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28735 * config/arc/arc.md (tst_movb): Add guard when splitting.
28737 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28739 * config/arc/arc.c (arc_can_eliminate): Test against
28740 arc_frame_pointer_needed.
28742 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28744 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
28745 to prevent store reordering.
28746 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
28747 (type): Add block type.
28748 (stack_tie): Define special instruction to be used in
28751 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28753 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
28754 constraint. It is not valid for the pattern.
28755 (noncommutative_binary_comparison): Likewise.
28757 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28759 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
28762 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28764 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
28765 be used by the reg-alloc.
28767 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28769 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28770 reg-alloc when having mul64 or mul32x16 instructions.
28771 (mulsidi3): Likewise.
28772 (umulsidi3): Likewise.
28773 (mulsi32x16): New pattern.
28774 (mulsi64): Likewise.
28775 (mulsidi64): Likewise.
28776 (umulsidi64): Likewise.
28777 (MUL32x16_REG): Define.
28778 (mul64_600): Use MUL32x16_REG.
28779 (mac64_600): Likewise.
28780 (umul64_600): Likewise.
28781 (umac64_600): Likewise.
28783 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28785 * config/arc/arc.md (mulsi3_700): Make it commutative.
28787 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28789 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28791 (*sign_extendsidi2_insn): Likewise for movstosw.
28793 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28795 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28796 the type of the input discriminant value. Convert the
28797 discriminant value of signedness vary.
28799 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28801 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28802 Add to -Wall section.
28804 2017-06-01 Richard Biener <rguenther@suse.de>
28806 PR middle-end/66313
28807 * fold-const.c (fold_plusminus_mult_expr): If the factored
28808 factor may be zero use a wrapping type for the inner operation.
28809 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28810 and handle moved defs.
28811 (process_assignment): Properly guard the unary op case. Return a
28812 tri-state indicating that moving the stmt before the call may allow
28813 to continue. Pass through to_move.
28814 (find_tail_calls): Handle moving unrelated defs before
28817 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28820 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28821 splitter result in the canonical way.
28823 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28825 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28826 also for 32bit target. Update insn attributes.
28827 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28829 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28831 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28832 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28833 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28834 (_mm_maskz_min_ss): New intrinsics.
28836 2017-05-31 Martin Liska <mliska@suse.cz>
28838 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28839 Change comment style to one we normally use.
28840 (vectorizable_reduction): Likewise.
28841 (vectorizable_induction): Likewise.
28842 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28843 (vectorizable_call): Likewise.
28844 (vectorizable_simd_clone_call): Likewise.
28845 (vectorizable_conversion): Likewise.
28846 (vectorizable_assignment): Likewise.
28847 (vectorizable_shift): Likewise.
28848 (vectorizable_operation): Likewise.
28849 (vectorizable_store): Likewise.
28850 (vectorizable_load): Likewise.
28851 * tree-vectorizer.h: Likewise.
28853 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28855 * passes.c (emergency_dump_function): New.
28856 * tree-pass.h (emergency_dump_function): Declare.
28857 * plugin.c (plugins_internal_error_function): Remove.
28858 * plugin.h (plugins_internal_error_function): Remove declaration.
28859 * toplev.c (internal_error_function): New static function. Use it...
28860 (general_init): ...here.
28862 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28864 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28865 (arc_rtx_costs): Add costs for new patterns.
28866 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28867 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28869 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28871 * tree-ssa-strlen.c (get_next_strinfo): New function.
28872 (get_stridx_plus_constant): Use it.
28873 (zero_length_string): Likewise.
28874 (adjust_related_strinfos): Likewise.
28875 (adjust_last_stmt): Likewise.
28877 2017-05-31 Richard Biener <rguenther@suse.de>
28880 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28881 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28883 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28885 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28886 loop_vinfo argument and use of dependence distance vectors.
28887 Check instead whether the two references differ only in their
28888 initial value and assume that they have the same alignment if the
28889 difference is a multiple of the vector alignment.
28890 (vect_analyze_data_refs_alignment): Update call accordingly.
28892 2017-05-31 Martin Liska <mliska@suse.cz>
28895 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28897 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28899 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28900 (create_intersect_range_checks): Move from ...
28901 * tree-data-ref.c (create_intersect_range_checks_index)
28902 (create_intersect_range_checks): ... to here.
28903 (create_runtime_alias_checks): New function factored from ...
28904 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28905 here. Call above function.
28906 * tree-data-ref.h (create_runtime_alias_checks): New function.
28908 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28910 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28911 segment length for dr_b and compute it in wide_int.
28913 2017-05-31 Richard Biener <rguenther@suse.de>
28915 PR tree-optimization/80906
28916 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28917 and pass through iv_map.
28918 (copy_bb_and_scalar_dependences): Adjust.
28919 (translate_pending_phi_nodes): Likewise.
28920 (copy_loop_close_phi_args): Handle code-generating IVs instead
28923 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28925 * diagnostic-color.c (color_dict): Add "type-diff".
28926 (parse_gcc_colors): Update comment.
28927 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28928 -fdiagnostics-show-template-tree and -fno-elide-type.
28929 (GCC_COLORS): Add type-diff to example.
28931 (-fdiagnostics-show-template-tree): New.
28932 (-fno-elide-type): New.
28933 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28934 the pp_format_decoder callback. Call any m_format_postprocessor's
28936 (pretty_printer::pretty_printer): Initialize
28937 m_format_postprocessor.
28938 (pretty_printer::~pretty_printer): Delete any
28939 m_format_postprocessor.
28940 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28941 (class format_postprocessor): New class.
28942 (struct pretty_printer::format_decoder): Document the new parameters.
28943 (struct pretty_printer::m_format_postprocessor): New field.
28944 * tree-diagnostic.c (default_tree_printer): Update for new
28945 bool and const char ** params.
28946 * tree-diagnostic.h (default_tree_printer): Likewise.
28948 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28950 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28951 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28952 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28953 rs6000_gen_cell_microcode code.
28954 (rs6000_final_prescan_insn): Delete.
28955 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28956 "warn-cell-microcode" entries.
28957 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28958 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28959 throughout. Change cc_reg_not_micro_cr0_operand to
28960 cc_reg_not_cr0_operand throughout.
28961 (*extendhi<mode>2_noload): Delete.
28962 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28963 (mwarn-cell-microcode): Delete.
28964 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28965 -mgen-cell-microcode and -mwarn-cell-microcode.
28967 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28970 * config/i386/constraints.md (Yd): New constraint.
28972 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28973 and (?Yd, r) alternatives. Update insn attributes.
28974 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28975 and (?*Yd, r) alternatives. Update insn attributes.
28976 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28978 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28980 * gimplify.c (gimplify_modify_expr): Don't create a
28981 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28984 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28986 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28988 2017-05-30 Richard Biener <rguenther@suse.de>
28990 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28991 and reduc_def fields.
28992 (STMT_VINFO_REDUC_TYPE): New define.
28993 (STMT_VINFO_REDUC_DEF): Likewise.
28994 (vect_force_simple_reduction): Adjust prototype.
28995 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28996 (vect_is_simple_reduction): Remove check_reduction argument.
28997 (vect_force_simple_reduction): Adjust and set
28998 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28999 (vectorizable_reduction): Do not re-do reduction analysis
29000 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29001 * tree-parloops.c (gather_scalar_reductions): Adjust.
29003 2017-05-30 Richard Biener <rguenther@suse.de>
29005 PR middle-end/80901
29006 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29009 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29011 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29012 Introduce unknown_misalignment parameter and remove vf.
29013 (vect_peeling_hash_get_lowest_cost):
29014 Pass unknown_misalignment parameter.
29015 (vect_enhance_data_refs_alignment):
29016 Fix unsupportable data ref treatment.
29018 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29020 * tree-vect-data-refs.c (vect_get_data_access_cost):
29021 Workaround for SLP handling.
29022 (vect_enhance_data_refs_alignment):
29023 Compute costs for doing no peeling at all, compare to the best
29024 peeling costs so far and avoid peeling if cheaper.
29026 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29028 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29029 Return peeling info and set costs to zero for unlimited cost
29031 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29032 with unknown misalignment. Compute and costs for unknown
29033 misalignment, compare them to the costs for known misalignment
29034 and choose the cheapest for peeling.
29036 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29038 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29039 (vect_get_peeling_costs_all_drs): Create function.
29040 (vect_peeling_hash_get_lowest_cost):
29041 Use vect_get_peeling_costs_all_drs.
29042 (vect_peeling_supportable): Create function.
29043 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29045 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29047 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29048 DR_HAS_NEGATIVE_STEP.
29049 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29050 (vect_enhance_data_refs_alignment): Use.
29051 (vect_duplicate_ssa_name_ptr_info): Use.
29052 * tree-vectorizer.h (dr_misalignment): Use.
29053 (known_alignment_for_access_p): Use.
29055 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29058 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29060 (has_section_name): New function.
29062 2017-05-30 Martin Liska <mliska@suse.cz>
29065 * auto-profile.c (get_function_decl_from_block): Fix
29068 2017-05-30 Richard Biener <rguenther@suse.de>
29070 PR middle-end/80876
29071 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29073 2017-05-30 Martin Liska <mliska@suse.cz>
29075 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29076 * dumpfile.h (struct dump_file_info): Remove ctors.
29078 2017-05-30 Martin Liska <mliska@suse.cz>
29080 * predict.def: Fix GNU coding style.
29082 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29084 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29085 Mark 'to' argument with ATTRIBUTE_UNUSED.
29087 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29089 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29090 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29091 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29094 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29096 * doc/install.texi (Options specification): Restore entry of
29097 --enable-sjlj-exceptions.
29099 2017-05-27 Michael Eager <eager@eagercon.com>
29102 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29104 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29106 * config/microblaze/microblaze.h
29107 (FIXED_REGISTERS): Update in macro.
29108 (CALL_USED_REGISTERS): Update in macro.
29110 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29112 * doc/install.texi: Add links to macOS binary distributions.
29114 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29118 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29120 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29122 2017-05-26 Martin Liska <mliska@suse.cz>
29124 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29126 2017-05-26 Martin Liska <mliska@suse.cz>
29128 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29129 always leading ';; '.
29130 (dump_bb_info): Likewise.
29131 (brief_dump_cfg): Likewise.
29132 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29133 * dumpfile.c: Remove usage of TDF_VERBOSE.
29134 * dumpfile.h (enum dump_kind): Likewise.
29135 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29136 * print-tree.c (debug_verbose): Remove.
29137 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29138 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29139 * tree-diagnostic.c (default_tree_printer): Replace
29140 TDF_DIAGNOSTIC with TDF_SLIM.
29142 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29144 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29145 in parameter loop, rather than loop_vinfo.
29146 (create_intersect_range_checks): Ditto.
29147 (vect_create_cond_for_alias_checks): Update call to above functions.
29149 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29151 PR tree-optimization/80815
29152 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29153 for merging runtime alias checks. Handle negative DR_STEPs.
29155 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29157 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29159 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29160 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29161 out code pruning runtime alias checks.
29162 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29163 factored out from above.
29164 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29166 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29168 (prune_runtime_alias_test_list): New decalaration.
29170 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29172 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29173 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29174 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29175 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29176 (operator==, comp_dr_with_seg_len_pair): Ditto.
29177 (vect_prune_runtime_alias_test_list): Ditto.
29179 2017-05-26 Martin Liska <mliska@suse.cz>
29182 * params.def: Bound partial-inlining-entry-probability param.
29184 2017-05-26 Marek Polacek <polacek@redhat.com>
29187 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29190 2017-05-26 Richard Biener <rguenther@suse.de>
29192 PR tree-optimization/80842
29193 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29196 2017-05-26 Richard Biener <rguenther@suse.de>
29198 PR tree-optimization/80844
29199 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29201 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29203 * doc/md.texi (Machine Constraints): Update x86 family
29204 machine constraints section to match 'config/i386/constraints.md'.
29206 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29208 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29210 2017-05-25 Nathan Sidwell <nathan@acm.org>
29212 * doc/invoke.texi (--enable-languages): Update documentation.
29214 2017-05-25 Martin Liska <mliska@suse.cz>
29216 * dumpfile.c: Add TDF_FOLDING.
29217 * dumpfile.h (enum dump_kind): Likewise.
29218 * genmatch.c (dt_simplify::gen_1): Use it.
29220 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29222 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29224 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29226 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29227 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29229 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29231 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29232 * match.pd (X == C): Rewrite it here.
29233 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29234 with_certain_nonzero_bits2): New predicates.
29235 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29237 2017-05-24 Nathan Sidwell <nathan@acm.org>
29239 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29242 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29245 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29247 * config/powerpcspe: New port. Files are copied from the rs6000
29248 port, with "rs6000" in filenames replaced by "powerpcspe".
29250 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29252 PR rtl-optimization/80754
29253 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29255 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29257 * config/sparc/sparc.md (length): Return the correct value for -mflat
29258 sibcalls to match output_sibcall.
29260 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29264 * config/rs6000/rs6000.c (struct machine_function): Add new field
29266 (rs6000_get_separate_components): Init that field, use it.
29267 (rs6000_components_for_bb): Use the field.
29269 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29271 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29273 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29275 PR middle-end/80823
29276 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29278 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29281 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29282 address against address_operand predicate.
29283 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29285 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29287 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29288 parameters passed indirectly.
29290 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29292 * config/i386/i386.md (*movdi_internal): Remove SSE4
29293 alternative 18 (?r, *v). Update insn attributes.
29294 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29295 Update insn attributes.
29296 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29297 Update insn attributes.
29298 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29299 alternative 1 (r, v). Remove isa attribute.
29300 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29301 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29302 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29304 2017-05-23 Tom de Vries <tom@codesourcery.com>
29306 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29309 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29311 * cgraphunit.c (symbol_table::process_new_functions): Update.
29312 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29313 (inline_generate_summary): Rename to ...
29314 (ipa_fn_summary_generate): ... this one.
29315 (inline_read_summary): Rename to ...
29316 (ipa_fn_summary_read): ... this one.
29317 (inline_write_summary): Rename to ...
29318 (ipa_fn_summary_write): ... this one.
29319 (inline_free_summary): Rename to ...
29320 (ipa_free_fn_summary): ... this one.
29321 (pass_data_local_fn_summary, pass_local_fn_summary,
29322 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29323 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29324 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29325 make_pass_ipa_fn_summary): New.
29326 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29327 inline_write_summary, inline_free_summary): Remove.
29328 (ipa_free_fn_summary) : New.
29329 * ipa-inline.c (ipa_inline): Update.
29330 (pass_ipa_inline): Do not generate summaries.
29331 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29333 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29334 and add pass_ipa_fn_summary.
29335 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29337 (make_pass_inline_parameters): Remove.
29339 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29341 * omp-low.c (struct omp_context): Remove "default_kind" member.
29344 * omp-offload.c (execute_oacc_device_lower): Remove the
29345 parallelism dimensions function attributes for unparallelized
29346 OpenACC kernels constructs.
29348 2017-05-23 Martin Liska <mliska@suse.cz>
29350 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29352 (cgraph_edge::make_speculative): Likewise.
29353 (cgraph_edge::resolve_speculation): Likewise.
29354 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29355 (cgraph_node::dump): Likewise.
29356 * cgraph.h: Likewise.
29357 * cgraphunit.c (analyze_functions): Likewise.
29358 (symbol_table::compile): Likewise.
29359 * ipa-cp.c (print_all_lattices): Likewise.
29360 (determine_versionability): Likewise.
29361 (initialize_node_lattices): Likewise.
29362 (ipcp_verify_propagated_values): Likewise.
29363 (estimate_local_effects): Likewise.
29364 (update_profiling_info): Likewise.
29365 (create_specialized_node): Likewise.
29366 (perhaps_add_new_callers): Likewise.
29367 (decide_about_value): Likewise.
29368 (decide_whether_version_node): Likewise.
29369 (identify_dead_nodes): Likewise.
29370 (ipcp_store_bits_results): Likewise.
29371 * ipa-devirt.c (dump_targets): Likewise.
29372 (ipa_devirt): Likewise.
29373 * ipa-icf.c (sem_item::dump): Likewise.
29374 (sem_function::equals): Likewise.
29375 (sem_variable::equals): Likewise.
29376 (sem_item_optimizer::read_section): Likewise.
29377 (sem_item_optimizer::execute): Likewise.
29378 (congruence_class::dump): Likewise.
29379 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29380 (dump_inline_summary): Likewise.
29381 (estimate_node_size_and_time): Likewise.
29382 (inline_analyze_function): Likewise.
29383 * ipa-inline-transform.c (inline_call): Likewise.
29384 * ipa-inline.c (report_inline_failed_reason): Likewise.
29385 (want_early_inline_function_p): Likewise.
29386 (edge_badness): Likewise.
29387 (update_edge_key): Likewise.
29388 (inline_small_functions): Likewise.
29389 * ipa-profile.c (ipa_profile): Likewise.
29390 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29391 (ipa_make_edge_direct_to_target): Likewise.
29392 (remove_described_reference): Likewise.
29393 (ipa_impossible_devirt_target): Likewise.
29394 (propagate_controlled_uses): Likewise.
29395 (ipa_print_node_params): Likewise.
29396 (ipcp_transform_function): Likewise.
29397 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29398 (propagate_pure_const): Likewise.
29399 * ipa-reference.c (generate_summary): Likewise.
29400 (read_write_all_from_decl): Likewise.
29401 (propagate): Likewise.
29402 (ipa_reference_read_optimization_summary): Likewise.
29403 * ipa-utils.c (ipa_merge_profiles): Likewise.
29404 * ipa.c (walk_polymorphic_call_targets): Likewise.
29405 (symbol_table::remove_unreachable_nodes): Likewise.
29406 (ipa_single_use): Likewise.
29407 * passes.c (execute_todo): Likewise.
29408 * predict.c (drop_profile): Likewise.
29409 * symtab.c (symtab_node::get_dump_name): New function.
29410 (symtab_node::dump_name): Likewise.
29411 (symtab_node::dump_asm_name): Likewise.
29412 (symtab_node::dump_references): Likewise.
29413 (symtab_node::dump_referring): Likewise.
29414 (symtab_node::dump_base): Likewise.
29415 (symtab_node::debug_symtab): Likewise.
29416 * tree-sra.c (convert_callers_for_node): Likewise.
29417 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29418 * value-prof.c (init_node_map): Likewise.
29420 2017-05-23 Martin Liska <mliska@suse.cz>
29422 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29423 and symtab_node::debug_symtab to symbol_table::debug.
29424 * cgraphunit.c (analyze_functions): Use the renamed function.
29425 (symbol_table::compile): Likewise.
29426 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29427 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29428 * passes.c (execute_todo): Likewise.
29429 * symtab.c (symbol_table::dump): New function.
29430 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29432 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29434 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
29435 that nonconst implies exec.
29437 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29439 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29440 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29441 (inline_edge_summary_vec): Turn into ...
29442 (ipa_call_summaries): ... this one.
29443 (redirect_to_unreachable, edge_set_predicate,
29444 evaluate_properties_for_edge, inline_summary_alloc,
29445 reset_ipa_call_summary, reset_inline_summary,
29446 inline_summary_t::duplicate): Update.
29447 (inline_edge_duplication_hook): Turn to ...
29448 (ipa_call_summary_t::duplicate): ... this one.
29449 (inline_edge_removal_hook): Turn to ...
29450 (ipa_call_summary_t::remove): ... this one.
29451 (dump_inline_edge_summary): Turn to ...
29452 (dump_ipa_call_summary): ... this one.
29453 (estimate_function_body_sizes): Update.
29454 (inline_update_callee_summaries): Update.
29455 (remap_edge_change_prob): Update.
29456 (remap_edge_summaries): Update.
29457 (inline_merge_summary): Update.
29458 (do_estimate_edge_time): Update.
29459 (inline_generate_summary): Update.
29460 (inline_read_section): Update.
29461 (inline_read_summary): Update.
29462 (inline_free_summary): Update.
29463 * ipa-inline.c (can_inline_edge_p): Update.
29464 (compute_inlined_call_time): Update.
29465 (want_inline_small_function_p): Update.
29466 (edge_badness): Update.
29467 (early_inliner): Update.
29468 * ipa-inline.h (inline_edge_summary): Turn to ...
29469 (ipa_call_summary): ... this one.
29470 (ipa_call_summary_t): New class.
29471 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29472 (ipa_call_summaries): New.
29473 (inline_edge_summary): Remove.
29474 (estimate_edge_growth): Update.
29475 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29476 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29477 * ipa-split.c (execute_split_functions): Update.
29478 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29480 2017-05-23 Tom de Vries <tom@codesourcery.com>
29482 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29483 attributes): Document rdrand effective target.
29485 2017-05-23 Tom de Vries <tom@codesourcery.com>
29487 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
29488 attributes): Sort alphabetically.
29490 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
29492 * config/avr/genmultilib.awk: Use gsub instead of gensub.
29494 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29497 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
29498 V2DF/V2DI splat into two separate patterns, one that handles
29499 registers, and the other that only handles memory. Drop support
29500 for splatting from a GPR on ISA 2.07 and then splitting the
29501 splat into direct move and splat.
29502 (vsx_splat_<mode>_reg): Likewise.
29503 (vsx_splat_<mode>_mem): Likewise.
29505 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
29507 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
29509 2017-05-22 Jakub Jelinek <jakub@redhat.com>
29511 PR middle-end/80809
29512 * omp-low.c (finish_taskreg_remap): New function.
29513 (finish_taskreg_scan): If unit size of ctx->record_type
29514 is non-constant, unshare the size expression and replace
29515 decls in it with possible outer var refs.
29517 PR middle-end/80809
29518 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
29519 GOVD_SHARED rather than GOVD_PRIVATE with it.
29520 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
29521 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
29523 PR middle-end/80853
29524 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
29525 as last argument to build_outer_var_ref for pointer bases of array
29526 section reductions.
29528 2017-05-19 Martin Sebor <msebor@redhat.com>
29530 * print-tree.c (print_node): Print DECL_READ_P flag.
29532 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29534 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
29535 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
29536 * cgraph.c: Likewise.
29537 * cgraphunit.c: Likewise.
29538 * gengtype.c: Likewise.
29539 * ipa-cp.c: Likewise.
29540 * ipa-devirt.c: Likewise.
29541 * ipa-icf.c: Likewise.
29542 * ipa-predicate.c: Likewise.
29543 * ipa-profile.c: Likewise.
29544 * ipa-prop.c: Likewise.
29545 * ipa-split.c: Likewise.
29547 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
29548 edge_predicate_pool, dump_inline_hints,
29549 inline_summary::account_size_time, redirect_to_unreachable,
29550 edge_set_predicate, set_hint_predicate,
29551 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
29552 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
29553 inline_summary_t::remove, remap_hint_predicate_after_duplication,
29554 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
29555 ipa_call_summary_t::remove, initialize_growth_caches,
29556 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
29557 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
29558 mark_modified, unmodified_parm_1, unmodified_parm,
29559 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
29560 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
29561 compute_bb_predicates, will_be_nonconstant_expr_predicate,
29562 will_be_nonconstant_predicate, record_modified_bb_info,
29563 get_minimal_bb, record_modified, param_change_prob,
29564 phi_result_unknown_predicate, predicate_for_phi_result,
29565 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
29566 estimate_function_body_sizes, compute_inline_parameters,
29567 compute_inline_parameters_for_curren, pass_data_inline_parameters,
29568 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
29569 inline_update_callee_summaries, remap_edge_change_prob,
29570 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
29571 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
29572 inline_analyze_function, inline_summary_t::insert,
29573 inline_generate_summary, read_ipa_call_summary, inline_read_section,
29574 inline_read_summary, write_ipa_call_summary, inline_write_summary,
29575 inline_free_summary): Move to ipa-fnsummary.h
29576 (predicate_t): Remove.
29577 * ipa-fnsummary.c: New file.
29578 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
29579 (enum inline_hints_vals, inline_hints, agg_position_info,
29580 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
29581 inline_summaries, ipa_call_summary, ipa_call_summary_t,
29582 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
29583 dump_inline_summary, dump_inline_hints, inline_generate_summary,
29584 inline_read_summary, inline_write_summary, inline_free_summary,
29585 inline_analyze_function, initialize_inline_failed,
29586 inline_merge_summary, inline_update_overall_summary,
29587 compute_inline_parameters): Move to ipa-fnsummary.h
29588 * ipa-fnsummary.h: New file.
29589 * ipa-inline-transform.h: Include ipa-inline.h.
29590 * ipa-inline.c: LIkewise.
29592 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29594 * ipa-inline.c (edge_badness): Use inlined_time instead of
29595 inline_summaries->get.
29597 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29599 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
29601 2017-05-22 Nathan Sidwell <nathan@acm.org>
29603 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
29604 (fdump-lang): Document 'raw' option.
29605 * dumpfile.h (TDI_tu): Delete.
29606 * dumpfile.c (dump_files): Remove translation-unit.
29607 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
29609 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
29611 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
29612 command option from $(AWK) call.
29613 * config/avr/genmultilib.awk: Simplify and rewrite so that it
29614 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
29615 [FORMAT]: Remove handling of variable.
29616 * config/avr/t-multilib: Regenerate.
29618 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29620 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
29622 (dump_inline_summary): Do not print self_time.
29623 (estimate_function_body_sizes): Do not set self_time.
29624 (compute_inline_parameters): Likewise.
29625 (inline_read_section, inline_write_summary): Do not stream self_time.
29626 * ipa-inline.h (inline_summary): Drop self_time.
29628 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
29630 * ipa-inline-analysis.c (account_size_time): Rename to ...
29631 (inline_summary::account_size_time): ... this one.
29632 (reset_ipa_call_summary): Turn to ...
29633 (ipa_call_summary::reset): ... this one.
29634 (reset_inline_summary): Turn to ...
29635 (inline_summary::reset): ... this one.
29636 (inline_summary_t::remove): Update.
29637 (inline_summary_t::duplicate): Update.
29638 (ipa_call_summary_t::remove): Update.
29639 (dump_inline_summary): Update.
29640 (estimate_function_body_sizes): Update.
29641 (compute_inline_parameters): Update.
29642 (estimate_node_size_and_time): Update.
29643 (inline_merge_summary): Update.
29644 (inline_update_overall_summary): Update.
29645 (inline_read_section): Update.
29646 (inline_write_summary): Update.
29647 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
29648 add account_size_time and reset member functions.
29649 (ipa_call_summary): Add reset function.
29650 * ipa-predicate.h (predicate::operator &): Constify.
29652 2017-05-22 Richard Biener <rguenther@suse.de>
29654 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
29656 2017-05-19 Jason Merrill <jason@redhat.com>
29658 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
29660 2017-05-19 Marek Polacek <polacek@redhat.com>
29663 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
29664 TYPE_OVERFLOW_WRAPS checks.
29666 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
29668 * tree-core.h (enum omp_clause_default_kind): Add
29669 "OMP_CLAUSE_DEFAULT_PRESENT".
29670 * tree-pretty-print.c (dump_omp_clause): Handle it.
29671 * gimplify.c (enum gimplify_omp_var_data): Add
29672 "GOVD_MAP_FORCE_PRESENT".
29673 (gimplify_adjust_omp_clauses_1): Map it to
29674 "GOMP_MAP_FORCE_PRESENT".
29675 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
29677 * gimplify.c (oacc_default_clause): Clarify.
29679 2017-05-19 Nathan Sidwell <nathan@acm.org>
29681 LANG_HOOK_REGISTER_DUMPS
29682 * toplev.c (general_init): Call register dump lang hook.
29683 * doc/invoke.texi: Document -fdump-lang option family.
29684 * dumpfile.c (dump_files): Remove class dump here.
29685 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
29686 * dumpfile.h (tree_dump_index): Remove TDI_class.
29687 * langhooks-def.h (lhd_register_dumps): Declare.
29688 (LANG_HOOKS_REGISTER_DUMPS): Define.
29689 (LANG_HOOKS_INITIALIZER): Add it.
29690 * langhooks.c (lhd_register_dumps): Define.
29691 * langhooks.h (struct lang_hooks): Add register_dumps.
29693 2017-05-19 Nathan Sidwell <nathan@acm.org>
29695 * context.h (context::set_passes): New.
29696 * context.c (context::context): Do not create pass manager.
29697 * toplev.c (general_init): Create pass manager here.
29699 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
29701 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
29702 use this splitter if two add or or instructions would also work for
29703 the constant we want to generate.
29705 2017-05-19 Richard Biener <rguenther@suse.de>
29708 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
29709 predicate evaluation.
29711 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29713 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
29715 * ipa-inline.c (want_inline_small_function_p): Do not cast to
29718 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
29720 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
29721 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
29722 (inline_edge_summary_vec): Turn into ...
29723 (ipa_call_summaries): ... this one.
29724 (redirect_to_unreachable, edge_set_predicate,
29725 evaluate_properties_for_edge, inline_summary_alloc,
29726 reset_ipa_call_summary, reset_inline_summary,
29727 inline_summary_t::duplicate): Update.
29728 (inline_edge_duplication_hook): Turn to ...
29729 (ipa_call_summary_t::duplicate): ... this one.
29730 (inline_edge_removal_hook): Turn to ...
29731 (ipa_call_summary_t::remove): ... this one.
29732 (dump_inline_edge_summary): Turn to ...
29733 (dump_ipa_call_summary): ... this one.
29734 (estimate_function_body_sizes): Update.
29735 (inline_update_callee_summaries): Update.
29736 (remap_edge_change_prob): Update.
29737 (remap_edge_summaries): Update.
29738 (inline_merge_summary): Update.
29739 (do_estimate_edge_time): Update.
29740 (inline_generate_summary): Update.
29741 (inline_read_section): Update.
29742 (inline_read_summary): Update.
29743 (inline_free_summary): Update.
29744 * ipa-inline.c (can_inline_edge_p): Update.
29745 (compute_inlined_call_time): Update.
29746 (want_inline_small_function_p): Update.
29747 (edge_badness): Update.
29748 (early_inliner): Update.
29749 * ipa-inline.h (inline_edge_summary): Turn to ...
29750 (ipa_call_summary): ... this one.
29751 (ipa_call_summary_t): New class.
29752 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
29753 (ipa_call_summaries): New.
29754 (inline_edge_summary): Remove.
29755 (estimate_edge_growth): Update.
29756 * ipa-profile.c (ipa_propagate_frequency_1): Update.
29757 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29758 * ipa-split.c (execute_split_functions): Update.
29759 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
29761 2017-05-19 Richard Biener <rguenther@suse.de>
29763 PR middle-end/80764
29764 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
29766 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
29768 * config/rs6000/rs6000.c (struct machine_function): Add field
29769 fpr_is_wrapped_separately.
29770 (rs6000_get_separate_components): Use 64 components. Handle the
29771 new FPR components.
29772 (rs6000_components_for_bb): Handle the FPR components.
29773 (rs6000_emit_prologue_components): Handle the FPR components.
29774 (rs6000_emit_epilogue_components): Handle the FPR components.
29775 (rs6000_set_handled_components): Handle the FPR components.
29776 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29777 that are already separately shrink-wrapped.
29778 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29779 that are already separately shrink-wrapped.
29781 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29784 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29787 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29788 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29789 cases where the register allocator uses a move and an offsettable
29790 memory operation to/from a FPR register on ISA 2.06/2.07.
29791 (define_peephole2 for Altivec d-form store): Likewise.
29793 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29796 * config/i386/mmx.md (*mov<mode>_internal): Enable
29797 alternatives 11, 12, 13 and 14 also for 32bit targets.
29798 Remove alternatives 15, 16, 17 and 18.
29799 * config/i386/sse.md (vec_concatv2di): Change
29800 alternative (!x, *y) to (x, ?!*Yn).
29802 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29804 * dumpfile.h (enum dump_kind): Remove stray comma.
29806 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29808 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29809 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29810 predicate::num_conditions
29811 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29812 (CHANGED): turn into predicate::changed.
29813 (agg_position_info): Move to ipa-predicate.h
29814 (add_condition, predicate::add_clause, predicate::operator &=,
29815 predicate::or_with, predicate::evaluate, predicate::probability,
29816 dump_condition, dump_clause, predicate::dump,
29817 predicate::remap_after_duplication, predicate::remap_after_inlining,
29818 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29819 (evaluate_conditions_for_known_args): Update.
29820 (set_cond_stmt_execution_predicate): Update.
29821 * ipa-inline.h: Include ipa-predicate.h
29822 (condition, inline_param_summary, conditions, agg_position_info,
29823 predicate): Move to ipa-predicate.h
29824 * ipa-predicate.c: New file.
29825 * ipa-predicate.h: New file.
29827 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29829 * final.c (leaf_function_p): Check we are not in a sequence.
29831 2017-05-18 Martin Liska <mliska@suse.cz>
29833 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29834 * dumpfile.c (dump_register): Use new enum dump_kind.
29835 (get_dump_file_name): Likewise.
29836 (dump_enable_all): Likewise.
29837 (dump_switch_p_1): Likewise.
29838 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29839 * dumpfile.h (enum dump_kind): New enum type.
29840 (struct dump_file_info): Create constructor and
29841 format fields and comments.
29842 * passes.c (pass_manager::register_one_dump_file):
29844 * statistics.c (statistics_early_init): Likewise.
29845 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29846 TDF_TREE with TDF_SLIM.
29847 (gather_memory_references_ref): Likewise.
29849 2017-05-18 Martin Liska <mliska@suse.cz>
29851 * vec.h (struct vnull): Use it.
29853 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29855 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29856 (true_predicate, false_predicate, true_predicate_p,
29857 false_predicate_p): Remove.
29858 (single_cond_predicate, not_inlined_predicate): Turn to member function
29860 (add_condition): Update.
29861 (add_clause): Turn to...
29862 (predicate::add_clause): ... this one; update; allow passing NULL
29864 (and_predicates): Turn to ...
29865 (predicate::operator &=): ... this one.
29866 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29867 (or_predicates): Turn to ...
29868 (predicate::or_with): ... this one.
29869 (evaluate_predicate): Turn to ...
29870 (predicate::evaluate): ... this one.
29871 (predicate_probability): Turn to ...
29872 (predicate::probability): ... this one.
29873 (dump_condition): Update.
29874 (dump_predicate): Turn to ...
29875 (predicate::dump): ... this one.
29876 (account_size_time): Update.
29877 (edge_set_predicate): Update.
29878 (set_hint_predicate): UPdate.
29879 (evaluate_conditions_for_known_args): Update.
29880 (evaluate_properties_for_edge): Update.
29881 (remap_predicate_after_duplication): Turn to...
29882 (predicate::remap_after_duplication): ... this one.
29883 (remap_hint_predicate_after_duplication): Update.
29884 (inline_summary_t::duplicate): UPdate.
29885 (dump_inline_edge_summary): Update.
29886 (dump_inline_summary): Update.
29887 (set_cond_stmt_execution_predicate): Update.
29888 (set_switch_stmt_execution_predicate): Update.
29889 (compute_bb_predicates): Update.
29890 (will_be_nonconstant_expr_predicate): Update.
29891 (will_be_nonconstant_predicate): Update.
29892 (phi_result_unknown_predicate): Update.
29893 (predicate_for_phi_result): Update.
29894 (array_index_predicate): Update.
29895 (estimate_function_body_sizes): Update.
29896 (estimate_node_size_and_time): Update.
29897 (estimate_ipcp_clone_size_and_time): Update.
29898 (remap_predicate): Rename to ...
29899 (predicate::remap_after_inlining): ... this one.
29900 (remap_hint_predicate): Update.
29901 (inline_merge_summary): Update.
29902 (inline_update_overall_summary): Update.
29903 (estimate_size_after_inlining): Update.
29904 (read_predicate): Rename to ...
29905 (predicate::stream_in): ... this one.
29906 (read_inline_edge_summary): Update.
29907 (write_predicate): Rename to ...
29908 (predicate::stream_out): ... this one.
29909 (write_inline_edge_summary): Update.
29910 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29911 (clause_t): Turn to uint32_t
29912 (predicate): Turn to class; implement constructor and operators
29914 (size_time_entry): Update.
29915 (inline_summary): Update.
29916 (inline_edge_summary): Update.
29918 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29920 * fold-const.c (fold_binary_loc): Move transformation...
29921 * match.pd (C - X CMP X): ... here.
29923 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29925 * config/sparc/sparc.c (sparc_option_override): Set function
29926 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29927 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29929 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29931 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29933 2017-05-18 Marek Polacek <polacek@redhat.com>
29936 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29937 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29938 gimple_assign_load_p.
29940 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29942 PR middle-end/80692
29943 * real.c (do_compare): Give decimal_do_compare preference over
29944 comparing just the signs.
29946 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29948 * doc/md.texi (Canonicalization of Instructions): Describe the
29949 canonical form of instructions that inherently set a condition
29952 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29954 PR middle-end/80775
29955 * tree-cfg.c: Move deletion of unreachable case statements to after
29956 the merging of consecutive case labels.
29958 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29960 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29961 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29962 restoring of callee-saved registers.
29964 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29966 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29967 * config/visium/visium.c (single_set_and_flags): Likewise.
29968 * config/visium/visium.md (Substitutions): Likewise.
29970 2017-05-17 Martin Liska <mliska@suse.cz>
29972 * cfg.c: Introduce dump_flags_t type and
29973 use it instead of int type.
29975 * cfghooks.c: Likewise.
29976 * cfghooks.h (struct cfg_hooks): Likewise.
29977 * cfgrtl.c: Likewise.
29978 * cfgrtl.h: Likewise.
29979 * cgraph.c (cgraph_node::get_body): Likewise.
29980 * coretypes.h: Likewise.
29981 * domwalk.c: Likewise.
29982 * domwalk.h: Likewise.
29983 * dumpfile.c (struct dump_option_value_info): Likewise.
29984 (dump_enable_all): Likewise.
29985 (dump_switch_p_1): Likewise.
29986 (opt_info_switch_p): Likewise.
29987 * dumpfile.h (enum tree_dump_index): Likewise.
29988 (struct dump_file_info): Likewise.
29989 * genemit.c: Likewise.
29990 * generic-match-head.c: Likewise.
29991 * gengtype.c (open_base_files): Likewise.
29992 * gimple-pretty-print.c: Likewise.
29993 * gimple-pretty-print.h: Likewise.
29994 * graph.c (print_graph_cfg): Likewise.
29995 * graphite-scop-detection.c (dot_all_sese): Likewise.
29996 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29997 * loop-unroll.c (report_unroll): Likewise.
29998 * passes.c (pass_manager::register_one_dump_file): Likewise.
29999 * print-tree.c: Likewise.
30000 * statistics.c: Likewise.
30001 * tree-cfg.c: Likewise.
30002 * tree-cfg.h: Likewise.
30003 * tree-dfa.c: Likewise.
30004 * tree-dfa.h: Likewise.
30005 * tree-dump.c (dump_function): Likewise.
30006 * tree-dump.h (struct dump_info): Likewise.
30007 * tree-pretty-print.c: Likewise.
30008 * tree-pretty-print.h: Likewise.
30009 * tree-ssa-live.c: Likewise.
30010 * tree-ssa-live.h: Likewise.
30011 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30012 * tree-vect-loop.c: Likewise.
30013 * tree-vect-slp.c: Likewise.
30015 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30016 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30018 PR tree-optimization/80457
30019 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30020 of all arguments to a statement as scalar_to_vec operations.
30021 (vectorizable_call): Adjust call to vect_model_simple_cost for
30023 (vectorizable_conversion): Likewise.
30024 (vectorizable_assignment): Likewise.
30025 (vectorizable_shift): Likewise.
30026 (vectorizable_operation): Likewise.
30027 (vectorizable_comparison): Likewise.
30028 (vect_is_simple_cond): Record the def types for operands.
30029 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30030 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30031 for statement argument count.
30033 2017-05-16 Carl Love <cel@us.ibm.com>
30035 * config/rs6000/rs6000-c: Add support for built-in functions
30036 vector unsigned long long vec_bperm (vector unsigned long long,
30037 vector unsigned char)
30038 vector signed long long vec_mule (vector signed int,
30040 vector unsigned long long vec_mule (vector unsigned int,
30041 vector unsigned int)
30042 vector signed long long vec_mulo (vector signed int,
30044 vector unsigned long long vec_mulo (vector unsigned int,
30045 vector unsigned int)
30046 vector signed char vec_sldw (vector signed char,
30047 vector signed char,
30049 vector unsigned char vec_sldw (vector unsigned char,
30050 vector unsigned char,
30052 vector signed short vec_sldw (vector signed short,
30053 vector signed short,
30055 vector unsigned short vec_sldw (vector unsigned short,
30056 vector unsigned short,
30058 vector signed int vec_sldw (vector signed int,
30061 vector unsigned int vec_sldw (vector unsigned int,
30062 vector unsigned int,
30064 vector signed long long vec_sldw (vector signed long long,
30065 vector signed long long,
30067 vector unsigned long long vec_sldw (vector unsigned long long,
30068 vector unsigned long long,
30070 * config/rs6000/rs6000-c: Add support for built-in functions
30071 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30072 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30073 * doc/extend.texi: Update the built-in documentation for the
30074 new built-in functions.
30076 2017-05-16 Marek Polacek <polacek@redhat.com>
30080 * tree.c (save_expr): Don't fold the expression.
30082 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30084 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30085 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30086 and (?*y,m). Update insn attributes.
30088 2017-05-16 Martin Liska <mliska@suse.cz>
30090 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30091 flags argument of print_gimple_stmt, print_gimple_expr,
30092 print_generic_stmt and print_generic_expr.
30093 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30094 * coretypes.h: Likewise.
30095 * except.c (dump_eh_tree): Likewise.
30096 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30097 * gimple-pretty-print.h: Likewise.
30098 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30099 (backprop::push_to_worklist): Likewise.
30100 (backprop::pop_from_worklist): Likewise.
30101 (backprop::process_use): Likewise.
30102 (backprop::intersect_uses): Likewise.
30103 (note_replacement): Likewise.
30104 * gimple-ssa-store-merging.c
30105 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30106 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30107 (pass_store_merging::execute): Likewise.
30108 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30109 (ssa_base_cand_dump_callback): Likewise.
30110 (dump_incr_vec): Likewise.
30111 (replace_refs): Likewise.
30112 (replace_mult_candidate): Likewise.
30113 (create_add_on_incoming_edge): Likewise.
30114 (create_phi_basis): Likewise.
30115 (insert_initializers): Likewise.
30116 (all_phi_incrs_profitable): Likewise.
30117 (introduce_cast_before_cand): Likewise.
30118 (replace_one_candidate): Likewise.
30119 * gimplify.c (gimplify_expr): Likewise.
30120 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30121 (set_rename): Likewise.
30122 (rename_uses): Likewise.
30123 (copy_loop_phi_nodes): Likewise.
30124 (add_close_phis_to_merge_points): Likewise.
30125 (copy_loop_close_phi_args): Likewise.
30126 (copy_cond_phi_args): Likewise.
30127 (graphite_copy_stmts_from_block): Likewise.
30128 (translate_pending_phi_nodes): Likewise.
30129 * graphite-poly.c (print_pdr): Likewise.
30130 (dump_gbb_cases): Likewise.
30131 (dump_gbb_conditions): Likewise.
30132 (print_scop_params): Likewise.
30133 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30134 (build_cross_bb_scalars_use): Likewise.
30135 (gather_bbs::before_dom_children): Likewise.
30136 * hsa-dump.c (dump_hsa_immed): Likewise.
30137 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30138 (get_replacement_map): Likewise.
30139 * ipa-inline-analysis.c (dump_condition): Likewise.
30140 (estimate_function_body_sizes): Likewise.
30141 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30142 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30143 * ipa-prop.c (ipa_dump_param): Likewise.
30144 (ipa_print_node_jump_functions_for_edge): Likewise.
30145 (ipa_modify_call_arguments): Likewise.
30146 (ipa_modify_expr): Likewise.
30147 (ipa_dump_param_adjustments): Likewise.
30148 (ipa_dump_agg_replacement_values): Likewise.
30149 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30150 * ipa-pure-const.c (check_stmt): Likewise.
30151 (pass_nothrow::execute): Likewise.
30152 * ipa-split.c (execute_split_functions): Likewise.
30153 * omp-offload.c (dump_oacc_loop_part): Likewise.
30154 (dump_oacc_loop): Likewise.
30155 * trans-mem.c (tm_log_emit): Likewise.
30156 (tm_memopt_accumulate_memops): Likewise.
30157 (dump_tm_memopt_set): Likewise.
30158 (dump_tm_memopt_transform): Likewise.
30159 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30160 (print_loop): Likewise.
30161 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30162 (chkp_gather_checks_info): Likewise.
30163 (chkp_get_check_result): Likewise.
30164 (chkp_remove_check_if_pass): Likewise.
30165 (chkp_use_outer_bounds_if_possible): Likewise.
30166 (chkp_reduce_bounds_lifetime): Likewise.
30167 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30168 (chkp_mark_completed_bounds): Likewise.
30169 (chkp_register_incomplete_bounds): Likewise.
30170 (chkp_mark_invalid_bounds): Likewise.
30171 (chkp_maybe_copy_and_register_bounds): Likewise.
30172 (chkp_build_returned_bound): Likewise.
30173 (chkp_get_bound_for_parm): Likewise.
30174 (chkp_build_bndldx): Likewise.
30175 (chkp_get_bounds_by_definition): Likewise.
30176 (chkp_generate_extern_var_bounds): Likewise.
30177 (chkp_get_bounds_for_decl_addr): Likewise.
30178 * tree-chrec.c (chrec_apply): Likewise.
30179 * tree-data-ref.c (dump_data_reference): Likewise.
30180 (dump_subscript): Likewise.
30181 (dump_data_dependence_relation): Likewise.
30182 (analyze_overlapping_iterations): Likewise.
30183 * tree-inline.c (expand_call_inline): Likewise.
30184 (tree_function_versioning): Likewise.
30185 * tree-into-ssa.c (dump_defs_stack): Likewise.
30186 (dump_currdefs): Likewise.
30187 (dump_names_replaced_by): Likewise.
30188 (dump_update_ssa): Likewise.
30189 (update_ssa): Likewise.
30190 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30191 * tree-parloops.c (build_new_reduction): Likewise.
30192 (try_create_reduction_list): Likewise.
30193 (ref_conflicts_with_region): Likewise.
30194 (oacc_entry_exit_ok_1): Likewise.
30195 (oacc_entry_exit_single_gang): Likewise.
30196 * tree-pretty-print.h: Likewise.
30197 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30198 (get_scalar_evolution): Likewise.
30199 (add_to_evolution): Likewise.
30200 (get_loop_exit_condition): Likewise.
30201 (analyze_evolution_in_loop): Likewise.
30202 (analyze_initial_condition): Likewise.
30203 (analyze_scalar_evolution): Likewise.
30204 (instantiate_scev): Likewise.
30205 (number_of_latch_executions): Likewise.
30206 (gather_chrec_stats): Likewise.
30207 (final_value_replacement_loop): Likewise.
30208 (scev_const_prop): Likewise.
30209 * tree-sra.c (dump_access): Likewise.
30210 (disqualify_candidate): Likewise.
30211 (create_access): Likewise.
30212 (reject): Likewise.
30213 (maybe_add_sra_candidate): Likewise.
30214 (create_access_replacement): Likewise.
30215 (analyze_access_subtree): Likewise.
30216 (analyze_all_variable_accesses): Likewise.
30217 (sra_modify_assign): Likewise.
30218 (initialize_constant_pool_replacements): Likewise.
30219 (find_param_candidates): Likewise.
30220 (decide_one_param_reduction): Likewise.
30221 (replace_removed_params_ssa_names): Likewise.
30222 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30223 * tree-ssa-copy.c (dump_copy_of): Likewise.
30224 (copy_prop_visit_cond_stmt): Likewise.
30225 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30226 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30227 (record_equivalences_from_stmt): Likewise.
30228 * tree-ssa-dse.c (compute_trims): Likewise.
30229 (delete_dead_call): Likewise.
30230 (delete_dead_assignment): Likewise.
30231 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30232 (forward_propagate_into_cond): Likewise.
30233 (pass_forwprop::execute): Likewise.
30234 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30235 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30237 (move_computations_worker): Likewise.
30238 (execute_sm): Likewise.
30239 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30240 (remove_exits_and_undefined_stmts): Likewise.
30241 (remove_redundant_iv_tests): Likewise.
30242 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30243 (adjust_iv_update_pos): Likewise.
30244 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30245 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30246 (value_replacement): Likewise.
30247 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30248 * tree-ssa-pre.c (print_pre_expr): Likewise.
30249 (get_representative_for): Likewise.
30250 (create_expression_by_pieces): Likewise.
30251 (insert_into_preds_of_block): Likewise.
30252 (eliminate_insert): Likewise.
30253 (eliminate_dom_walker::before_dom_children): Likewise.
30254 (eliminate): Likewise.
30255 (remove_dead_inserted_code): Likewise.
30256 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30257 * tree-ssa-reassoc.c (get_rank): Likewise.
30258 (eliminate_duplicate_pair): Likewise.
30259 (eliminate_plus_minus_pair): Likewise.
30260 (eliminate_not_pairs): Likewise.
30261 (undistribute_ops_list): Likewise.
30262 (eliminate_redundant_comparison): Likewise.
30263 (update_range_test): Likewise.
30264 (optimize_range_tests_var_bound): Likewise.
30265 (optimize_vec_cond_expr): Likewise.
30266 (rewrite_expr_tree): Likewise.
30267 (rewrite_expr_tree_parallel): Likewise.
30268 (linearize_expr): Likewise.
30269 (break_up_subtract): Likewise.
30270 (linearize_expr_tree): Likewise.
30271 (attempt_builtin_powi): Likewise.
30272 (attempt_builtin_copysign): Likewise.
30273 (transform_stmt_to_copy): Likewise.
30274 (transform_stmt_to_multiply): Likewise.
30275 (dump_ops_vector): Likewise.
30276 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30277 (print_scc): Likewise.
30278 (set_ssa_val_to): Likewise.
30279 (visit_reference_op_store): Likewise.
30280 (visit_use): Likewise.
30281 (sccvn_dom_walker::before_dom_children): Likewise.
30282 (run_scc_vn): Likewise.
30283 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30285 (expr_hash_elt::print): Likewise.
30286 (const_and_copies::pop_to_marker): Likewise.
30287 (const_and_copies::record_const_or_copy_raw): Likewise.
30288 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30289 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30290 (dump_predicates): Likewise.
30291 (find_uninit_use): Likewise.
30292 (warn_uninitialized_phi): Likewise.
30293 (pass_late_warn_uninitialized::execute): Likewise.
30294 * tree-ssa.c (verify_vssa): Likewise.
30295 (verify_ssa): Likewise.
30296 (maybe_optimize_var): Likewise.
30297 * tree-vrp.c (dump_value_range): Likewise.
30298 (dump_all_value_ranges): Likewise.
30299 (dump_asserts_for): Likewise.
30300 (register_edge_assert_for_2): Likewise.
30301 (vrp_visit_cond_stmt): Likewise.
30302 (vrp_visit_switch_stmt): Likewise.
30303 (vrp_visit_stmt): Likewise.
30304 (vrp_visit_phi_node): Likewise.
30305 (simplify_cond_using_ranges_1): Likewise.
30306 (fold_predicate_in): Likewise.
30307 (evrp_dom_walker::before_dom_children): Likewise.
30308 (evrp_dom_walker::push_value_range): Likewise.
30309 (evrp_dom_walker::pop_value_range): Likewise.
30310 (execute_early_vrp): Likewise.
30312 2017-05-16 Richard Biener <rguenther@suse.de>
30314 * dwarf2out.c (loc_list_from_tree_1): Do not create
30315 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30317 2017-05-16 Richard Biener <rguenther@suse.de>
30319 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30321 (note_variable_value_in_expr): If we resolved the decl ref
30322 do not push to the stack.
30324 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30326 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30327 operations in fast-math mode.
30328 (vaddq_f16): Likewise.
30329 (vmul_f16): Likewise.
30330 (vmulq_f16): Likewise.
30331 (vsub_f16): Likewise.
30332 (vsubq_f16): Likewise.
30333 * config/arm/neon.md (add<mode>3): New.
30335 (fma:<VH:mode>3): New. Also remove outdated comment.
30338 2017-05-16 Martin Liska <mliska@suse.cz>
30342 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30343 (odr_types_equivalent_p): Likewise.
30345 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30347 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30349 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30352 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30353 non-interunit SSE move alternatives with '?'.
30354 (zero-extendsidi peephole2): New peephole to skip intermediate
30355 general register in SSE zero-extend sequence.
30357 2017-05-15 Jeff Law <law@redhat.com>
30359 * reorg.c (relax_delay_slots): Create a new variable to hold
30360 the temporary target rather than clobbering TARGET_LABEL.
30362 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30363 missing argument to extract_bit_field call.
30364 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30366 2017-05-15 Martin Liska <mliska@suse.cz>
30369 * gcc.c (process_command): Do not allow empty argument of -o option.
30371 2017-05-15 Renlin Li <renlin.li@arm.com>
30373 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30374 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30375 * config/aarch64/constraints.md (Usf): Add long call check.
30376 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30377 (call_value): Likewise.
30378 (sibcall): Likewise.
30379 (sibcall_value): Likewise.
30381 (call_value_insn): New.
30382 (sibcall_insn): Update rtx pattern.
30383 (sibcall_value_insn): Likewise.
30384 (call_internal): Remove.
30385 (call_value_internal): Likewise.
30386 (sibcall_internal): Likewise.
30387 (sibcall_value_internal): Likewise.
30388 (call_reg): Likewise.
30389 (call_symbol): Likewise.
30390 (call_value_reg): Likewise.
30391 (call_value_symbol): Likewise.
30393 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30396 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30398 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30400 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30401 compatible with CCGOCmode and with CCZmode.
30403 2017-05-14 Martin Sebor <msebor@redhat.com>
30405 PR middle-end/77671
30406 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30407 (gimple_fold_builtin_snprintf): Same.
30408 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30409 (gimple_fold_builtin_snprintf): Same.
30410 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30411 of character types.
30412 (is_call_safe): New function.
30413 (try_substitute_return_value): Call it.
30414 (try_simplify_call): New function.
30415 (pass_sprintf_length::handle_gimple_call): Call it.
30417 2017-05-14 Martin Sebor <msebor@redhat.com>
30419 PR middle-end/80669
30420 * builtins.c (expand_builtin_stpncpy): Simplify.
30422 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30424 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30425 * config/i386/i386.h
30426 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30427 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30428 (struct machine_function): Add new members call_ms2sysv,
30429 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30430 (struct machine_frame_state): New fields sp_realigned and
30431 sp_realigned_offset.
30432 * config/i386/i386.c
30433 (enum xlogue_stub): New enum.
30434 (enum xlogue_stub_sets): New enum.
30435 (class xlogue_layout): New class.
30436 (struct ix86_frame): New fields stack_realign_allocate_offset,
30437 stack_realign_offset and outlined_save_offset. Modify comments to
30438 detail stack layout when using out-of-line stubs.
30439 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
30440 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
30441 -mcall-ms2sysv-xlogues.
30442 (stub_managed_regs): New static variable.
30443 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
30444 registers managed by out-of-line stub.
30445 (disable_call_ms2sysv_xlogues): New function.
30446 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
30447 m->call_ms2sysv when appropriate and compute frame layout for
30449 (sp_valid_at, fp_valid_at): New inline functions.
30450 (choose_basereg): New function.
30451 (choose_baseaddr): Add align parameter, use choose_basereg and modify
30453 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
30454 Use align parameter of choose_baseaddr to generated aligned SSE movs
30456 (pro_epilogue_adjust_stack): Modify to track
30457 machine_frame_state::sp_realigned.
30458 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
30459 (ix86_nsaved_sseregs): Likewise.
30460 (ix86_emit_save_regs): Likewise.
30461 (ix86_emit_save_regs_using_mov): Likewise.
30462 (ix86_emit_save_sse_regs_using_mov): Likewise.
30463 (get_scratch_register_on_entry): Likewise.
30464 (gen_frame_set): New function.
30465 (gen_frame_load): Likewise.
30466 (gen_frame_store): Likewise.
30467 (emit_outlined_ms2sysv_save): Likewise.
30468 (emit_outlined_ms2sysv_restore): Likewise.
30469 (ix86_expand_prologue): Modify stack re-alignment code and call
30470 emit_outlined_ms2sysv_save when appropriate.
30471 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
30472 parameter rtx_insn *insn, which allows the function to be used to only
30473 generate the notes.
30474 (ix86_expand_epilogue): Modify validity checks of frame and stack
30475 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
30476 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
30477 * config/i386/predicates.md
30478 (save_multiple): New predicate.
30479 (restore_multiple): Likewise.
30480 * config/i386/sse.md
30481 (save_multiple<mode>): New pattern.
30482 (save_multiple_realign<mode>): Likewise.
30483 (restore_multiple<mode>): Likewise.
30484 (restore_multiple_and_return<mode>): Likewise.
30485 (restore_multiple_leave_return<mode>): Likewise.
30486 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
30488 2017-05-14 Julia Koval <julia.koval@intel.com>
30490 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
30491 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
30492 (__builtin_ia32_xsetbv): New builtins.
30493 * config/i386/i386.c (ix86_expand_special_args_builtin):
30495 (ix86_expand_builtin): Special expand for new intrinsics.
30496 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
30497 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
30498 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
30500 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30502 * cfganal.c (inverted_post_order_compute): Change argument type
30504 * cfganal.h (inverted_post_order_compute): Adjust prototype.
30505 * df-core.c (rest_of_handle_df_initialize): Adjust.
30506 (rest_of_handle_df_finish): Likewise.
30507 (df_analyze_1): Likewise.
30508 (df_analyze): Likewise.
30509 (loop_inverted_post_order_compute): Change argument to be a vec *.
30510 (df_analyze_loop): Adjust.
30511 (df_get_n_blocks): Likewise.
30512 (df_get_postorder): Likewise.
30513 * df.h (struct df_d): Change field to be a vec.
30514 * lcm.c (compute_laterin): Adjust.
30515 (compute_available): Likewise.
30516 * lra-lives.c (lra_create_live_ranges_1): Likewise.
30517 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
30518 * tree-ssa-pre.c (compute_antic): Likewise.
30520 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30522 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
30523 (depth_first_search::depth_first_search): Change structure init
30524 function to this constructor.
30525 (depth_first_search::add_bb): Rename function to this member.
30526 (depth_first_search::execute): Likewise.
30527 (flow_dfs_compute_reverse_finish): Adjust.
30529 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30531 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
30532 (longest_simple_path): Likewise.
30533 * shrink-wrap.c (spread_components): Likewise.
30534 (disqualify_problematic_components): Likewise.
30535 (emit_common_heads_for_components): Likewise.
30536 (emit_common_tails_for_components): Likewise.
30537 (insert_prologue_epilogue_for_components): Likewise.
30539 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30541 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
30544 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30546 * df-core.c (df_set_blocks): Start using auto_bitmap.
30547 (df_compact_blocks): Likewise.
30548 * df-problems.c (df_rd_confluence_n): Likewise.
30549 * df-scan.c (df_insn_rescan_all): Likewise.
30550 (df_process_deferred_rescans): Likewise.
30551 (df_update_entry_block_defs): Likewise.
30552 (df_update_exit_block_uses): Likewise.
30553 (df_entry_block_bitmap_verify): Likewise.
30554 (df_exit_block_bitmap_verify): Likewise.
30555 (df_scan_verify): Likewise.
30556 * lra-constraints.c (lra_constraints): Likewise.
30557 (undo_optional_reloads): Likewise.
30558 (lra_undo_inheritance): Likewise.
30559 * lra-remat.c (calculate_gen_cands): Likewise.
30560 (do_remat): Likewise.
30561 * lra-spills.c (assign_spill_hard_regs): Likewise.
30562 (spill_pseudos): Likewise.
30563 * tree-ssa-pre.c (bitmap_set_and): Likewise.
30564 (bitmap_set_subtract_values): Likewise.
30566 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30568 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
30569 management with auto_bitmap.
30570 (fix_inter_tick): Likewise.
30571 (fix_recovery_deps): Likewise.
30572 * ira.c (add_store_equivs): Likewise.
30573 (find_moveable_pseudos): Likewise.
30574 (split_live_ranges_for_shrink_wrap): Likewise.
30575 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
30576 (rtx_reuse_manager::seen_def_p): Likewise.
30577 (rtx_reuse_manager::set_seen_def): Likewise.
30578 * print-rtl.h (class rtx_reuse_manager): Likewise.
30580 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30582 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
30584 (migrate_btr_def): Likewise.
30585 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
30586 * df-core.c (loop_post_order_compute): Likewise.
30587 (loop_inverted_post_order_compute): Likewise.
30588 * hsa-common.h: Likewise.
30589 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
30590 * init-regs.c (initialize_uninitialized_regs): Likewise.
30591 * ipa-inline.c (resolve_noninline_speculation): Likewise.
30592 (inline_small_functions): Likewise.
30593 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
30594 * ira.c (combine_and_move_insns): Likewise.
30595 (build_insn_chain): Likewise.
30596 * loop-invariant.c (find_invariants): Likewise.
30597 * lower-subreg.c (propagate_pseudo_copies): Likewise.
30598 * predict.c (tree_predict_by_opcode): Likewise.
30599 (predict_paths_leading_to): Likewise.
30600 (predict_paths_leading_to_edge): Likewise.
30601 (estimate_loops_at_level): Likewise.
30602 (estimate_loops): Likewise.
30603 * shrink-wrap.c (try_shrink_wrapping): Likewise.
30604 (spread_components): Likewise.
30605 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
30606 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
30607 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
30608 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
30609 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
30610 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
30611 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
30612 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
30613 (mark_threaded_blocks): Likewise.
30614 (thread_through_all_blocks): Likewise.
30615 * tree-ssa.c (verify_ssa): Likewise.
30616 (execute_update_addresses_taken): Likewise.
30617 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
30619 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30621 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
30623 (post_order_compute): Likewise.
30624 (inverted_post_order_compute): Likewise.
30625 (pre_and_rev_post_order_compute_fn): Likewise.
30627 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30629 * genrecog.c (int_set::int_set): Explicitly construct our
30630 auto_vec base class.
30631 * vec.h (auto_vec::auto_vec): New constructor.
30633 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30635 * bitmap.h (class auto_bitmap): New constructor taking
30636 bitmap_obstack * argument.
30638 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
30640 * bitmap.h (class auto_bitmap): Change type of m_bits to
30641 bitmap_head, and adjust ctor / dtor and member operators.
30643 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
30645 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
30646 when returned register mode doesn't match original mode.
30648 2017-05-12 Jeff Law <law@redhat.com>
30649 Jakub Jelinek <jakub@redhat.com>
30651 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
30652 we look for cc setter after the compare-elim changes.
30653 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
30654 within the vector to match what compare-elim now expects.
30655 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
30656 (xorsi3_flags, one_cmplsi2_flags): Likewise.
30658 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
30659 after the compare-elim changes.
30660 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
30661 the vector to match what compare-elim now expects.
30662 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
30663 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
30664 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
30665 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
30666 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
30668 * config/visium/visium.c (single_set_and_flags): Fix where
30669 we look for cc setter after the compare-elim changes.
30670 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
30671 with the vector to match what compare-elim now expects.
30672 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
30673 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
30674 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
30675 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
30676 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
30677 (neg<mode>2_insn_set_overflow): Likewise.
30679 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
30681 PR middle-end/79794
30682 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
30683 maybe_expand_insn call, set ops[0].target. If still set after call,
30684 set alt_rtl. Add extra arg to recursive calls.
30685 (extract_bit_field): Add alt_rtl argument. Pass to
30687 * expmed.h (extract_bit_field): Fix prototype.
30688 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
30689 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
30690 to extract_bit_field_calls.
30691 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
30692 Pass alt_rtl to extract_bit_field calls.
30693 * calls.c (store_unaligned_arguments_into_psuedos)
30694 load_register_parameters): Pass extra NULL to extract_bit_field calls.
30695 * optabs.c (maybe_legitimize_operand): Clear op->target when call
30697 * optabs.h (struct expand_operand): Add target bitfield.
30699 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30701 * compare-elim.c (try_eliminate_compare): Canonicalize
30702 operation with embedded compare to
30703 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
30704 (set (reg) (operation)].
30706 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
30708 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
30711 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
30712 cost of adding a carry flag for ADC instruction.
30713 [case MINUS]: Ignore the cost of subtracting a carry flag
30714 for SBB instruction.
30716 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
30718 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
30720 * config/rs6000/bmiintrin.h: New file.
30721 * config/rs6000/bmi2intrin.h: New file.
30722 * config/rs6000/x86intrin.h: New file.
30724 2017-05-12 Jeff Law <law@redhat.com>
30726 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
30729 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
30731 PR middle-end/80707
30732 * tree-cfg.c: Remove cfg edges of unreachable case statements.
30734 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30736 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30737 early expansion of vector divide builtins.
30738 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
30739 builtins identified as having unsigned arguments.
30741 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30743 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
30744 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
30745 expansion of vector logical operations (and, andc, or, xor,
30748 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30750 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
30751 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
30753 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
30755 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
30756 early GIMPLE expansion of vector multiplies.
30758 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30760 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
30761 TARGET_HAVE_MOVT conditional.
30762 (movt splitter): Likewise.
30764 2017-05-12 Richard Biener <rguenther@suse.de>
30766 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
30767 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30768 Fold all stmts not inplace.
30770 2017-05-12 Richard Biener <rguenther@suse.de>
30772 PR tree-optimization/80713
30773 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30774 inserted_exprs bit for not removed stmts.
30776 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30778 PR middle-end/69921
30779 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30780 parallelized" attribute for parallelized OpenACC kernels.
30781 * omp-offload.c (execute_oacc_device_lower): Use it.
30783 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30784 Set "oacc kernels" attribute.
30785 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30786 parameter. Adjust all users.
30787 (oacc_fn_attrib_kernels_p): Remove function.
30788 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30789 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30790 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30791 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30792 assert "oacc kernels" attribute is set.
30794 2017-05-11 Carl Love <cel@us.ibm.com>
30796 * config/rs6000/rs6000-c: Add support for built-in functions
30797 vector unsigned char vec_popcnt (vector signed char)
30798 vector unsigned char vec_popcnt (vector unsigned char)
30799 vector unsigned short vec_popcnt (vector signed short)
30800 vector unsigned short vec_popcnt (vector unsigned short)
30801 vector unsigned int vec_popcnt (vector signed int)
30802 vector unsigned int vec_popcnt (vector unsigned int)
30803 vector unsigned long long vec_popcnt (vector signed long long)
30804 vector unsigned long long vec_popcnt (vector unsigned long long)
30805 vector signed long long vec_slo (vector signed long long,
30806 vector signed char)
30807 vector signed long long vec_slo (vector signed long long,
30808 vector unsigned char)
30809 vector unsigned long long vec_slo (vector unsigned long long,
30810 vector signed char)
30811 vector unsigned long long vec_slo (vector unsigned long long,
30812 vector unsigned char)
30813 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30814 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30815 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30816 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30817 * doc/extend.texi: Update the built-in documentation file for the
30818 new built-in functions.
30820 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30822 * attribs.h (sorted_attr_string): Move machine independent
30823 functions for target clone support from the i386 port to common
30824 code. Rename ix86_function_versions to common_function_versions.
30825 Rename make_name to make_unique_name.
30826 (common_function_versions): Likewise.
30827 (make_unique_name): Likewise.
30828 (make_dispatcher_decl): Likewise.
30829 (is_function_default_version): Likewise.
30830 * attribs.c (attr_strcmp): Likewise.
30831 (sorted_attr_string): Likewise.
30832 (common_function_versions): Likewise.
30833 (make_unique_name): Likewise.
30834 (make_dispatcher_decl): Likewise.
30835 (is_function_default_version): Likewise.
30836 * config/i386/i386.c (attr_strcmp): Likewise.
30837 (sorted_attr_string): Likewise.
30838 (ix86_function_versions): Likewise.
30839 (make_name): Likewise.
30840 (make_dispatcher_decl): Likewise.
30841 (is_function_default_version): Likewise.
30842 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30844 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30847 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30848 Account for direct move costs for vec_construct of integer
30851 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30854 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30855 (UNSPEC_STX_ATOMIC): Ditto.
30856 (loaddi_via_sse): New insn.
30857 (storedi_via_sse): Ditto.
30858 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30859 Update corresponding peephole2 patterns.
30860 (atomic_storedi_fpu): Ditto.
30862 2017-05-11 Julia Koval <julia.koval@intel.com>
30864 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30865 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30867 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30868 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30869 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30871 2017-05-11 Nathan Sidwell <nathan@acm.org>
30873 * graphite-poly.c: Include dumpfile.h.
30875 * dumpfle.h (dump_function): Declare here ...
30876 * tree-dump.h (dump_function): ... not here.
30877 * dumpfile.c: #include tree-cfg.h.
30878 (dump_function): Move here from ...
30879 * tree-dump.c (dump_function): ... here.
30880 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30881 * graphite-poly.c: Don't include tree-dump.h.
30882 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30883 * print-tree.c: Likewise.
30884 * stor-layout.c: Likewise.
30885 * tree-nested.c: Likewise.
30887 * dumpfile.c (dump_start): Use TDF_FLAGS.
30888 (dump_enable_all): Fix TDF_KIND check thinko.
30890 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30892 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30893 array entries to represent two legal parameterizations of the
30894 overloaded __builtin_cmpb function, as represented by the
30895 P6_OV_BUILTIN_CMPB constant.
30896 (altivec_resolve_overloaded_builtin): Add special case handling
30897 for the __builtin_cmpb function, as represented by the
30898 P6_OV_BUILTIN_CMPB constant.
30899 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30900 (BU_P6_64BIT_2): New macro.
30901 (BU_P6_OVERLOAD_2): New macro
30902 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30903 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30904 (CMPB): Add overload support to represent both 32-bit and 64-bit
30905 compare-bytes function.
30906 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30907 support for TARGET_CMPB.
30908 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30909 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30910 documentation of the __builtin_cmpb overloaded built-in function.
30912 2017-05-11 Richard Biener <rguenther@suse.de>
30914 PR tree-optimization/80705
30915 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30916 bases are not vectorizable.
30918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30920 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30921 when counting register pressure.
30923 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30925 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30926 (iv_ca_more_deps): Renamed to ...
30927 (iv_ca_compare_deps): ... this.
30928 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30930 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30932 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30934 (determine_group_iv_costs): ... here.
30935 (find_inv_vars_cb): Record inv var if it's not recorded before.
30937 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30939 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30940 (get_shiftadd_cost): Ditto.
30942 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30944 * tree-ssa-address.c: Include header file.
30945 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30947 (add_to_parts): Refactor.
30948 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30949 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30952 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30954 PR tree-optimization/53090
30955 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30957 (extract_cond_operands): Detect condition with IV on both sides
30958 and return COMP_IV_EXPR_2.
30959 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30960 (rewrite_use_compare): Simplify by removing call to function
30961 extract_cond_operands.
30963 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30965 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30966 (extract_cond_operands): Detect condition comparing against non-
30967 invariant bound and return appropriate enum value.
30968 (find_interesting_uses_cond): Update use of extract_cond_operands.
30969 Handle its return value accordingly.
30970 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30972 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30974 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30975 nonlinear iv_use computation in loop invariant sensitive way.
30977 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30979 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30980 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30982 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30984 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30985 (dump_cand): Support iv_cand.inv_exprs.
30986 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30988 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30991 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30993 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30995 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30996 as local function. Include necessary header files.
30997 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30999 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31001 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31003 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31005 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31006 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31007 RSHIFT_EXPR and BIT_NOT_EXPR.
31009 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31011 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31012 (adjust_setup_cost): New parameter supporting round up adjustment.
31013 (struct address_cost_data): Delete.
31014 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31015 (split_address_cost, ptr_difference_cost): Delete.
31016 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31017 (struct ainc_cost_data): New struct.
31018 (get_address_cost_ainc): New function.
31019 (get_address_cost, get_computation_cost): Reimplement.
31020 (determine_group_iv_cost_address): Record inv_expr for all uses of
31022 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31023 (iv_ca_has_deps): Reimplemented to ...
31024 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31026 (iv_ca_extend): Call iv_ca_more_deps.
31028 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31030 * tree-ssa-address.c (struct mem_address): Move to header file.
31031 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31032 * tree-ssa-address.h (struct mem_address): Move from C file.
31033 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31035 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31037 * tree-affine.h (aff_combination_type): New interface.
31038 (aff_combination_zero_p): Remove static.
31039 (aff_combination_const_p): New interface.
31040 (aff_combination_singleton_var_p): New interfaces.
31042 2017-05-11 Richard Biener <rguenther@suse.de>
31044 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31045 Skip unreachable blocks and destinations.
31046 (eliminate): Move stmt removal and fixup ...
31047 (fini_eliminate): ... here. Skip inserted exprs.
31048 (pass_pre::execute): Move fini_pre after fini_eliminate.
31049 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31050 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31051 PRE to get rid of dead code that has invalid SSA form and
31052 split critical edges again.
31054 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31056 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31058 2017-05-11 Richard Biener <rguenther@suse.de>
31060 * passes.c (execute_function_todo): Verify loops if they are
31061 said to be up-to-date.
31062 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31063 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31065 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31068 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31069 handle calling assemble_external ourself.
31072 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31073 modes with zero size. Enhance comment.
31075 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31077 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31078 built-ins for vec_xl and vec_xst with short and char pointer
31081 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31083 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31084 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31085 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31086 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31087 (_mm_maskz_min_round_ss): New intrinsics.
31088 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31089 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31090 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31091 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31092 (__builtin_ia32_minss_mask_round): New builtins.
31093 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31094 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31095 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31097 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31098 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31100 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31103 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31105 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31106 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31107 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31108 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31109 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31110 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31111 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31112 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31113 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31114 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31115 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31116 (__builtin_ia32_mulss_mask_round): New builtins.
31117 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31118 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31119 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31121 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31122 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31124 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31127 2017-05-10 Julia Koval <julia.koval@intel.com>
31129 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31130 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31131 (_mm256_setr_m128i): New intrinsics.
31133 2017-05-10 Julia Koval <julia.koval@intel.com>
31135 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31136 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31137 (_mm_maskz_rcp14_ss): New intrinsics.
31138 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31139 (__builtin_ia32_rcp14ss_mask): New builtins.
31140 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31142 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31144 PR tree-optimization/51513
31145 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31146 (assert_unreachable_fallthru_edge_p): Use it.
31147 (group_case_labels_stmt): Likewise.
31148 * tree-cfg.h: Prototype it.
31149 * stmt.c: Include cfghooks.h and tree-cfg.h.
31150 (emit_case_dispatch_table) <gap_label>: New local variable.
31151 Use it to fill dispatch table gaps.
31152 Test for default_label before updating probabilities.
31153 (expand_case) <default_label>: Remove unneeded initialization.
31154 Test for unreachable default case statement and remove its edge.
31155 Set default_label accordingly.
31156 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31158 2017-05-10 Carl Love <cel@us.ibm.com>
31160 * config/rs6000/rs6000-c: Add support for built-in functions
31161 vector signed char vec_neg (vector signed char)
31162 vector signed short int vec_neg (vector short int)
31163 vector signed int vec_neg (vector signed int)
31164 vector signed long long vec_neg (vector signed long long)
31165 vector float vec_neg (vector float)
31166 vector double vec_neg (vector double)
31167 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31169 * config/rs6000/altivec.h: Add define for vec_neg
31170 * doc/extend.texi: Update the built-in documentation for the
31171 new built-in functions.
31173 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31175 PR tree-optimization/77644
31176 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31178 2017-05-10 Nathan Sidwell <nathan@acm.org>
31180 * dumpfile.h (TDI_lang_all): New.
31181 (TDF_KIND): New. Renumber others
31182 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31184 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31186 (get_dump_file_name): Adjust suffix generation.
31187 (dump_enable_all): Use TDF_KIND.
31188 * doc/invoke.texi (-fdump-lang-all): Document.
31190 * dumpfile.h: Tabify.
31192 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31195 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31196 Move member access before delete.
31198 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31200 * tree-inline.c (expand_call_inline): Split block at stmt
31203 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31206 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31207 are now unused after splitting mov{sf,sd}_hardfloat.
31208 (f32_lr2): Likewise.
31209 (f32_lm): Likewise.
31210 (f32_lm2): Likewise.
31211 (f32_li): Likewise.
31212 (f32_li2): Likewise.
31213 (f32_lv): Likewise.
31214 (f32_sr): Likewise.
31215 (f32_sr2): Likewise.
31216 (f32_sm): Likewise.
31217 (f32_sm2): Likewise.
31218 (f32_si): Likewise.
31219 (f32_si2): Likewise.
31220 (f32_sv): Likewise.
31221 (f32_dm): Likewise.
31222 (f32_vsx): Likewise.
31223 (f32_av): Likewise.
31224 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31225 For movsf, order stores so the VSX stores occur before the GPR
31226 store which encourages the register allocator to use a traditional
31227 FPR instead of a GPR. For movsd, order the stores so that the GPR
31228 store comes before the VSX stores to allow the power6 to work.
31229 This is due to the power6 not having a 32-bit integer store
31230 instruction from a FPR.
31231 (movsf_hardfloat): Likewise.
31232 (movsd_hardfloat): Likewise.
31234 2017-05-09 Martin Sebor <msebor@redhat.com>
31236 PR translation/80280
31237 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31240 PR translation/80280
31241 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31242 data member added in r247778.
31243 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31245 2017-05-09 Nathan Sidwell <nathan@acm.org>
31247 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31249 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31252 2017-05-09 Marek Polacek <polacek@redhat.com>
31254 * doc/invoke.texi: Fix typo.
31256 2017-05-09 Richard Biener <rguenther@suse.de>
31258 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31259 (vrp_val_is_min): Likewise.
31260 (set_value_range_to_value): Likewise.
31261 (set_value_range_to_nonnegative): Likewise.
31262 (gimple_assign_nonzero_p): Likewise.
31263 (gimple_stmt_nonzero_p): Likewise.
31264 (vrp_int_const_binop): Likewise. Remove unreachable case.
31265 (adjust_range_with_scev): Adjust comments.
31266 (compare_range_with_value): Likewise.
31267 (extract_range_from_phi_node): Likewise.
31268 (test_for_singularity): Likewise.
31270 2017-05-09 Richard Biener <rguenther@suse.de>
31272 * tree-vrp.c (get_single_symbol): Add assert that we don't
31273 get overflowed constants as invariant part.
31274 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31275 checks. Use wi::cmp instead of recursing for integer constants.
31276 (compare_values): Just ignore whether we assumed undefined
31277 overflow instead of failing the compare.
31278 (extract_range_for_var_from_comparison_expr): Add comment before the
31279 TREE_NO_WARNING sets.
31280 (test_for_singularity): Likewise.
31281 (extract_range_from_comparison): Do not disable optimization
31282 when we assumed undefined overflow.
31283 (extract_range_basic): Remove init of unused var.
31285 2017-05-09 Richard Biener <rguenther@suse.de>
31287 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31288 (extract_range_from_multiplicative_op_1): Adjust.
31289 (extract_range_from_binary_expr_1): Use int_const_binop.
31291 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31294 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31295 rs6000_store_data_bypass_p in seven define_bypass directives and
31296 in several comments.
31297 * config/rs6000/rs6000-protos.h: Add prototype for
31298 rs6000_store_data_bypass_p function.
31299 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31300 function implements slightly different (rs6000-specific) semantics
31301 than store_data_bypass_p, returning false rather than aborting
31302 with assertion error when arguments do not satisfy the
31303 requirements of store data bypass.
31304 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31305 rs6000_store_data_bypass_p.
31307 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31309 * config/xtensa/xtensa-protos.h
31310 (xtensa_initial_elimination_offset): New declaration.
31311 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31312 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31313 macro definition, add case for FRAME_POINTER_REGNUM when
31314 FRAME_GROWS_DOWNWARD.
31315 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31316 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31317 xtensa_initial_elimination_offset.
31319 2017-05-08 Nathan Sidwell <nathan@acm.org>
31321 * doc/invoke.texi: Alphabetize -fdump options.
31323 2017-05-08 Martin Sebor <msebor@redhat.com>
31325 PR translation/80280
31326 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31328 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31330 * target.def (compute_frame_layout): New optional target hook.
31331 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31332 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31333 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31335 * reload1.c (verify_initial_elim_offsets): Likewise.
31336 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31337 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31338 (arm_get_frame_offsets): Split up into this ...
31339 (arm_compute_frame_layout): ... and this function.
31341 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31343 * config/aarch64/constraints.md (Usa): New constraint.
31344 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31346 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31348 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31349 with_multilib_list after it has been checked.
31351 2017-05-08 Richard Biener <rguenther@suse.de>
31353 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31354 (bitmap_set_subtract_values): Likewise.
31356 2017-05-08 Richard Biener <rguenther@suse.de>
31358 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31359 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31361 (gimple_stmt_nonzero_warnv_p): Rename to ...
31362 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31364 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31365 (extract_range_basic): Adjust, do not disable propagation on
31366 strict overflow sensitive simplification.
31367 (vrp_visit_cond_stmt): Likewise.
31369 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31371 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31372 body size unconditionally.
31374 2017-05-07 Jeff Law <law@redhat.com>
31377 2017-05-06 Jeff Law <law@redhat.com>
31378 PR tree-optimization/78496
31379 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31382 PR tree-optimization/78496
31383 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31384 (simplify_stmt_using_ranges): Call it.
31385 (vrp_dom_walker::before_dom_children): Extract equivalences
31386 from an ASSERT_EXPR with an equality comparison against a
31389 2017-05-06 Jeff Law <law@redhat.com>
31391 PR tree-optimization/78496
31392 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31395 PR tree-optimization/78496
31396 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31397 (simplify_stmt_using_ranges): Call it.
31398 (vrp_dom_walker::before_dom_children): Extract equivalences
31399 from an ASSERT_EXPR with an equality comparison against a
31402 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31404 * lra-constraints.c (lra_copy_reg_equiv): New function.
31405 (split_reg): Use it to copy equivalence information from the
31406 original register to the spill register.
31408 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31410 PR rtl-optimization/75964
31411 * simplify-rtx.c (simplify_const_relational_operation): Remove
31412 invalid handling of comparisons of integer ABS.
31414 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31416 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31417 initialize to zero.
31418 (init_regs): Remove declaration.
31419 (function_arg_advance_32): Initialize error_p as boolean variable.
31421 2017-05-05 Nathan Sidwell <nathan@acm.org>
31423 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31424 lines. Use for (;;).
31426 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31428 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31429 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31430 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31431 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31432 VF=2 that require versioning.
31434 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31436 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
31439 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31441 * diagnostic.h (diagnostic_override_option_index): Convert from
31442 macro to inline function.
31444 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31446 * diagnostic.c (last_module_changed_p): New function.
31447 (set_last_module): New function.
31448 (diagnostic_report_current_module): Convert macro usage to
31449 the above functions.
31450 * diagnostic.h (diagnostic_context::last_module): Strengthen
31451 from const line_map * to const line_map_ordinary *.
31452 (diagnostic_last_module_changed): Delete macro.
31453 (diagnostic_set_last_module): Delete macro.
31455 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31457 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
31458 with diagnostic_report_diagnostic.
31459 (diagnostic_n_impl_richloc): Likewise.
31460 * diagnostic.h (report_diagnostic): Delete macro.
31461 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
31462 with diagnostic_report_diagnostic.
31463 * substring-locations.c (format_warning_va): Likewise.
31465 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31467 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
31468 save/restor of format_spec. Move option-printing code to...
31469 (print_option_information): ...this new function, and
31470 reimplement by simply printing to the pretty_printer,
31471 rather than appending to the format string.
31473 2017-05-05 David Malcolm <dmalcolm@redhat.com>
31475 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
31476 handling logic into...
31477 (update_effective_level_from_pragmas): ...this new function.
31479 2017-05-04 Andrew Waterman <andrew@sifive.com>
31481 * config/riscv/riscv.opt (mstrict-align): New option.
31482 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
31483 (SLOW_UNALIGNED_ACCESS): Define.
31484 (riscv_slow_unaligned_access): Declare.
31485 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
31487 (riscv_slow_unaligned_access): New variable.
31488 (rocket_tune_info): Set slow_unaligned_access to true.
31489 (optimize_size_tune_info): Set slow_unaligned_access to false.
31490 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
31491 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
31492 (riscv_option_override): Set riscv_slow_unaligned_access.
31493 * doc/invoke.texi: Add -mstrict-align to RISC-V.
31495 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
31497 * config/riscv/riscv.md: Unify indentation.
31499 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
31504 * config/rs6000/rs6000.md (u code attribute): Add FIX and
31506 (extendsi<mode>2): Add support for doing sign extension via
31507 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
31508 don't have ISA 3.0 instructions.
31509 (extendsi<mode>2 splitter): Likewise.
31510 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
31511 generate the normal insns since SImode can now go in vector
31512 registers. Disallow the special UNSPECs needed for previous
31513 machines to hide SImode being used. Add new insns
31514 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
31515 (fix_trunc<mode>si2_stfiwx): Likewise.
31516 (fix_trunc<mode>si2_internal): Likewise.
31517 (fixuns_trunc<mode>si2): Likewise.
31518 (fixuns_trunc<mode>si2_stfiwx): Likewise.
31519 (fctiw<u>z_<mode>_smallint): Likewise.
31520 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
31521 of floating point to 32-bit integer from doing a direct move to
31522 the GPR registers to do a store.
31523 (fctiwz_<mode>): Break long line.
31525 2017-05-05 Bin Cheng <bin.cheng@arm.com>
31527 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
31528 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
31529 (addr_list, addr_offset_valid_p): New.
31530 (split_address_groups): Check offset validity with above function.
31531 (gt-tree-ssa-loop-ivopts.h): Include header file.
31533 2017-05-05 Nathan Sidwell <nathan@acm.org>
31535 * config.gcc (arm*-*-*): Add missing 'fi'.
31537 2017-05-05 Steve Ellcey <sellcey@cavium.com>
31539 * doc/invoke.texi (-fopt-info): Explicitly say order of options
31540 included in -fopt-info does not matter.
31541 * doc/optinfo.texi (-fopt-info): Fix description of default
31542 behavour. Explicitly say order of options included in -fopt-info
31545 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
31547 * config.gcc: Allow combinations of aprofile and rmprofile values for
31548 --with-multilib-list.
31549 * config/arm/t-multilib: New file.
31550 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
31551 variables. Remove setting of ISA and floating-point ABI in
31552 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
31553 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
31554 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
31555 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
31557 * config/arm/t-rmprofile: Likewise except for the matches changes.
31558 * doc/install.texi (--with-multilib-list): Document the combination of
31559 aprofile and rmprofile values and warn about pitfalls in doing that.
31561 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31563 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
31564 (movdi_aarch64): Likewise.
31566 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31568 PR tree-optimization/80632
31569 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
31571 (build_arrays): Initialize it for virtual phis.
31572 (fix_phi_nodes): Use it for virtual phis.
31574 PR tree-optimization/80558
31575 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
31576 [x, y] op z into [x op, y op z] for op & or | if conditions
31579 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
31580 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31583 * config/arm/arm.md (use_literal_pool): Remove.
31584 (64-bit immediate split): No longer takes cost into consideration
31585 if arm_disable_literal_pool is enabled.
31586 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
31587 used when arm_disable_literal_pool is enabled.
31588 (arm_max_const_double_inline_cost): Remove use of
31589 arm_disable_literal_pool.
31590 (push_minipool_fix): Add assert.
31591 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
31592 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
31593 (no_literal_pool_sf_immediate): New.
31595 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31597 PR tree-optimization/80613
31598 * tree-ssa-dce.c (propagate_necessity): Remove cases for
31599 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
31601 2017-05-05 Richard Biener <rguenther@suse.de>
31603 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
31605 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
31607 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
31608 of this flag from insn conditions due to removal from r247495.
31610 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
31612 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
31614 (arm_early_store_addr_dep_ptr): Likewise.
31615 * config/arm/aarch-common-protos.h
31616 (arm_early_load_addr_dep_ptr): Add prototype.
31617 (arm_early_store_addr_dep_ptr): Likewise.
31618 * config/arm/cortex-a53.md: Add new bypasses.
31620 2017-05-05 Jakub Jelinek <jakub@redhat.com>
31622 * tree.c (next_type_uid): Change type to unsigned.
31623 (type_hash_canon): Decrement back next_type_uid if
31624 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
31625 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
31628 2017-05-04 Martin Sebor <msebor@redhat.com>
31630 * builtins.c: Fix a trivial typo in a comment.
31632 PR middle-end/79234
31633 * builtins.c (check_sizes): Adjust to handle reading past the end.
31634 Avoid printing excessive upper bound of ranges. Use %E to print
31635 tree nodes instead of converting them to %wu.
31636 (expand_builtin_memchr): New function.
31637 (compute_dest_size): Rename...
31638 (compute_objsize): ...to this.
31639 (expand_builtin_memcpy): Adjust.
31640 (expand_builtin_mempcpy): Adjust.
31641 (expand_builtin_strcat): Adjust.
31642 (expand_builtin_strcpy): Adjust.
31643 (check_strncat_sizes): Adjust.
31644 (expand_builtin_strncat): Adjust.
31645 (expand_builtin_strncpy): Adjust and simplify.
31646 (expand_builtin_memset): Adjust.
31647 (expand_builtin_bzero): Adjust.
31648 (expand_builtin_memcmp): Adjust.
31649 (expand_builtin): Handle memcmp.
31650 (maybe_emit_chk_warning): Check strncat just once.
31652 2017-05-04 Martin Sebor <msebor@redhat.com>
31654 PR preprocessor/79214
31655 PR middle-end/79222
31656 PR middle-end/79223
31657 * builtins.c (check_sizes): Add inlining context and issue
31658 warnings even when -Wno-system-headers is set.
31659 (check_strncat_sizes): Same.
31660 (expand_builtin_strncat): Same.
31661 (expand_builtin_memmove): New function.
31662 (expand_builtin_stpncpy): Same.
31663 (expand_builtin): Handle memmove and stpncpy.
31665 2017-05-04 Bin Cheng <bin.cheng@arm.com>
31667 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
31668 which is not used any more.
31670 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31672 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
31674 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31676 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
31677 (cortexa53_tunings): Likewise.
31678 (cortexa57_tunings): Likewise.
31679 (cortexa72_tunings): Likewise.
31680 (cortexa73_tunings): Likewise.
31682 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31684 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
31685 Set loop alignment to 8.
31687 2017-05-04 Martin Sebor <msebor@redhat.com>
31689 PR translation/80280
31690 * builtins.c (expand_builtin_object_size): Add missing quoting to
31691 %D and like directives.
31692 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
31693 (hsa_type_for_tree_type): Same.
31694 (verify_function_arguments): Same.
31695 * symtab.c (symbol_table::change_decl_assembler_name): Same.
31696 * varasm.c (get_section): Same.
31699 2017-05-04 Martin Sebor <msebor@redhat.com>
31701 PR translation/80280
31702 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
31704 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
31706 * config/aarch64/aarch64.c (generic_addrcost_table):
31707 Change HI/TI mode setting.
31709 2017-05-04 Martin Jambor <mjambor@suse.cz>
31711 PR tree-optimization/80622
31712 * tree-sra.c (comes_initialized_p): New function.
31713 (build_accesses_from_assign): Only set write lazily when
31714 comes_initialized_p is false.
31715 (analyze_access_subtree): Use comes_initialized_p.
31716 (propagate_subaccesses_across_link): Assert !comes_initialized_p
31717 instead of testing for PARM_DECL.
31719 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31721 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
31722 constraint on operand 0 to allow more general addressing modes.
31723 Adjust output template.
31724 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
31726 * config/aarch64/aarch64-protos.h
31727 (aarch64_address_valid_for_prefetch_p): Declare prototype.
31728 * config/aarch64/constraints.md (Dp): New address constraint.
31729 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
31732 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
31734 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
31735 update use of estimate_ipcp_clone_size_and_time.
31736 (estimate_local_effects): Update use of
31737 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
31738 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
31739 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
31740 Return nonspecialized time.
31742 2017-05-04 Richard Biener <rguenther@suse.de>
31744 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
31745 for the last VUSE which def dominates the PHI. Directly call
31747 (get_continuation_for_phi_1): Remove.
31749 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
31751 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
31752 to explain the use of truncating division. Cap the number of
31753 iterations to the maximum given by nb_iterations_upper_bound,
31756 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
31758 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
31759 * configure: Regenerate.
31760 * config.in: Regenerate.
31761 * config/i386/driver-mingw32.c: new file.
31762 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
31763 * config.host: Link driver-mingw32.o on MinGW host.
31764 * doc/install.texi: Document new --enable-mingw-wildcard configure
31767 2017-05-04 Marek Polacek <polacek@redhat.com>
31769 PR tree-optimization/80612
31770 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31772 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31773 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31775 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31776 (movt splitter): Likewise.
31777 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31778 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31779 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31780 block for Thumb-1 with MOVT.
31781 (thumb2_legitimate_address_p): Move code block ...
31782 (can_avoid_literal_pool_for_label_p): ... into this new function.
31783 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31785 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31786 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31787 "M-profile targets with the MOVT instruction".
31789 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31791 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31792 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31793 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31795 2017-05-04 Martin Liska <mliska@suse.cz>
31797 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31798 variable cond_code.
31800 2017-05-04 Richard Biener <rguenther@suse.de>
31802 * tree.c (array_at_struct_end_p): Handle arrays at struct
31803 end with flexarrays more conservatively. Refactor and treat
31804 arrays of arrays or aggregates more strict. Fix
31805 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31806 * tree.c (array_at_struct_end_p): Adjust prototype.
31807 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31808 * gimple-fold.c (get_range_strlen): Likewise.
31809 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31811 2017-05-04 Richard Biener <rguenther@suse.de>
31813 PR tree-optimization/31130
31814 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31816 (supports_overflow_infinity): Likewise.
31817 (is_negative_overflow_infinity): Likewise.
31818 (is_positive_overflow_infinity): Likewise.
31819 (is_overflow_infinity): Likewise.
31820 (stmt_overflow_infinity): Likewise.
31821 (overflow_infinity_range_p): Likewise.
31822 (usable_range_p): Remove as always returning true.
31823 (make_overflow_infinity): Remove.
31824 (negative_overflow_infinity): Likewise.
31825 (positive_overflow_infinity): Likewise.
31826 (avoid_overflow_infinity): Likewise.
31827 (set_value_range): Adjust accordingly.
31828 (set_value_range_to_nonnegative): Likewise, remove now unused
31829 overflow_infinity arg.
31830 (vrp_operand_equal_p): Adjust.
31831 (update_value_range): Likewise.
31832 (range_int_cst_singleton_p): Likewise.
31833 (operand_less_p): Likewise.
31834 (compare_values_warnv): Likewise.
31835 (extract_range_for_var_from_comparison_expr): Likewise.
31836 (vrp_int_const_binop): Likewise.
31837 (zero_nonzero_bits_from_vr): Likewise.
31838 (extract_range_from_multiplicative_op_1): Likewise.
31839 (extract_range_from_binary_expr_1): Likewise.
31840 (extract_range_from_unary_expr): Likewise.
31841 (extract_range_from_comparison): Likewise.
31842 (extract_range_basic): Likewise.
31843 (adjust_range_with_scev): Likewise.
31844 (compare_ranges): Likewise.
31845 (compare_range_with_value): Likewise.
31846 (dump_value_range): Likewise.
31847 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31849 (simplify_cond_using_ranges): Adjust.
31851 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31853 * brig-builtins.def: Added a builtin for class_f64.
31854 * builtin-types.def: Added a builtin type needed by class_f64.
31856 2017-05-03 Jason Merrill <jason@redhat.com>
31858 * timevar.def: Add TV_CONSTEXPR.
31860 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31862 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31864 2017-05-03 Martin Jambor <mjambor@suse.cz>
31866 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31867 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31868 * ipa-cp.c (ipcp_propagate_stage): Do not call
31869 ipa_update_after_lto_read.
31870 * ipa-inline.c (ipa_inline): Likewise.
31872 2017-05-03 Martin Jambor <mjambor@suse.cz>
31874 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31875 tag. Added a default constructor and a destructor.
31876 (ipa_edge_args_sum_t): New class;
31877 (ipa_edge_args_sum): Declare.
31878 (ipa_edge_args_vector): Remove declaration.
31879 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31880 (ipa_free_edge_args_substructures): Remove declaration.
31881 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31882 (ipa_edge_args_info_available_for_edge_p): Likewise.
31883 * ipa-prop.c (ipa_edge_args_vector): Removed.
31884 (edge_removal_hook_holder): Likewise.
31885 (edge_duplication_hook_holder): Likewise.
31886 (ipa_edge_args_sum): New variable.
31887 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31888 ipa_edge_args_vector.
31889 (ipa_free_edge_args_substructures): Likewise.
31890 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31891 ipa_edge_args_vector.
31892 (ipa_edge_removal_hook): Turned into method
31893 ipa_edge_args_sum_t::remove.
31894 (ipa_edge_duplication_hook): Turned into method
31895 ipa_edge_args_sum_t::duplicate.
31896 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31897 registering edge hooks.
31898 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31899 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31900 ipa_edge_args_sum instead of ipa_edge_args_vector.
31901 * ipa-profile.c (ipa_profile): Likewise.
31903 2017-05-03 Martin Jambor <mjambor@suse.cz>
31905 * symbol-summary.h (function_summary): New method exists.
31906 (function_summary::symtab_removal): Deallocate through release.
31907 (call_summary): New class.
31908 (gt_ggc_mx): New overload.
31909 (gt_pch_nx): Likewise.
31910 (gt_pch_nx): Likewise.
31912 2017-05-03 Jeff Law <law@redhat.com>
31914 PR tree-optimization/78496
31915 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31916 from simplify_cond_using_ranges. Split off code to walk
31917 backwards through casts into ...
31918 (simplify_cond_using_ranges_2): New function.
31919 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31920 (execute_vrp): After identifying jump threads, call
31921 simplify_cond_using_ranges_2.
31923 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31926 * ipa-inline.h (inline_summary): Add ctor.
31927 (create_ggc): Do not use ggc_cleared_alloc.
31929 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31930 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31932 * gcc.c (handle_braces): Support escaping in switch matching
31934 * doc/invoke.texi (Spec Files): Document it.
31935 Remove superfluous @code markup in items.
31937 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31939 * diagnostic-show-locus.c (struct column_range): New struct.
31940 (get_affected_columns): New function.
31941 (get_printed_columns): New function.
31942 (struct correction): New struct.
31943 (correction::ensure_capacity): New function.
31944 (correction::ensure_terminated): New function.
31945 (struct line_corrections): New struct.
31946 (line_corrections::~line_corrections): New dtor.
31947 (line_corrections::add_hint): New function.
31948 (layout::print_trailing_fixits): Reimplement in terms of the new
31950 (selftest::test_overlapped_fixit_printing): New function.
31951 (selftest::diagnostic_show_locus_c_tests): Call it.
31953 2017-05-03 Nathan Sidwell <nathan@acm.org>
31955 Canonicalize canonical type hashing
31956 * tree.h (type_hash_canon_hash): Declare.
31957 * tree.c (type_hash_list, attribute_hash_list): Move into
31958 type_hash_canon_hash.
31959 (build_type_attribute_qual_variant): Break out hash code calc into
31960 type_hash_canon_hash.
31961 (type_hash_canon_hash): New. Generic type hash computation.
31962 (build_range_type_1, build_array_type_1, build_function_type,
31963 build_method_type_directly, build_offset_type, build_complex_type,
31964 make_vector_type): Call it.
31966 2017-05-03 Richard Biener <rguenther@suse.de>
31968 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31969 When all DRs have unknown misaligned do not always peel
31970 when there is a store but apply the same costing model as if
31971 there were only loads.
31973 2017-05-03 Richard Biener <rguenther@suse.de>
31976 PR tree-optimization/80492
31977 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31978 compare_base_decls returning dont-know properly.
31980 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31982 * config/arm/iterators.md (CCSI): New mode iterator.
31983 (arch): New mode attribute.
31984 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31985 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31986 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31987 code iterator for success result mode.
31988 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31989 the corresponding new insn generators.
31991 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31994 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31995 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31997 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31999 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32000 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32001 (DDR_A): Wrap DDR argument in brackets.
32002 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32003 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32004 (DDR_REVERSED_P): Likewise.
32006 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32008 PR tree-optimization/79472
32009 * tree-switch-conversion.c (struct switch_conv_info): Add
32010 contiguous_range and default_case_nonstandard fields.
32011 (collect_switch_conv_info): Compute contiguous_range and
32012 default_case_nonstandard fields, don't clear final_bb if
32013 contiguous_range and only the default case doesn't have the required
32015 (check_all_empty_except_final): Set default_case_nonstandard instead
32016 of failing if contiguous_range and the default case doesn't have empty
32018 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32019 and only the default case doesn't have the required constants. Skip
32021 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32022 if default_case_nonstandard.
32023 (build_constructors): Build constant 1 just once. Assert that default
32024 values aren't inserted in between cases if contiguous_range. Skip
32026 (build_arrays): Skip virtual phis.
32027 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32028 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32029 Handle virtual phis.
32030 (gen_inbound_check): Handle default_case_nonstandard case.
32031 (process_switch): Adjust check_final_bb caller. Call
32032 gather_default_values with the first non-default case instead of
32033 default case if default_case_nonstandard.
32035 2017-05-02 Nathan Sidwell <nathan@acm.org>
32037 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32038 check. Fix formatting.
32040 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32042 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32043 errors when comparing specialized and unspecialized times.
32045 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32047 * diagnostic-show-locus.c
32048 (layout::should_print_annotation_line_p): Make private.
32049 (layout::print_annotation_line): Make private.
32050 (layout::annotation_line_showed_range_p): Make private.
32051 (layout::show_ruler): Make private.
32052 (layout::print_source_line): Make private. Pass in line and
32053 line_width, rather than calling location_get_source_line. Drop
32055 (layout::print_leading_fixits): New method.
32056 (layout::print_any_fixits): Rename to...
32057 (layout::print_trailing_fixits): ...this, and make private.
32058 Don't print newline fixits.
32059 (diagnostic_show_locus): Move logic for printing one row into...
32060 (layout::print_line): ...this new function. Move the
32061 location_get_source_line call and error-handling from
32062 print_source_line to here. Call print_leading_fixits, and rename
32063 print_any_fixits to print_trailing_fixits.
32064 (selftest::test_fixit_insert_containing_newline): Update now that
32065 newlines are partially supported.
32066 (selftest::test_fixit_insert_containing_newline_2): New test.
32067 (selftest::test_fixit_replace_containing_newline): Update comments.
32068 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32069 * edit-context.c (class added_line): New class.
32070 (class edited_line): Describe newline handling in comment.
32071 (edited_line::actually_edited_p): New method.
32072 (edited_line::print_content): Delete redundant decl.
32073 (edited_line::m_predecessors): New field.
32074 (edited_file::print_content): Call edited_line::print_content.
32075 (edited_file::print_diff): Update to support newlines.
32076 (edited_file::print_diff_hunk): Likewise.
32077 (edited_file::print_run_of_changed_lines): New function.
32078 (edited_file::print_diff_line): Convert to...
32079 (print_diff_line): ...this.
32080 (edited_file::get_effective_line_count): New function.
32081 (edited_line::edited_line): Initialize new field m_predecessors.
32082 (edited_line::~edited_line): Clean up m_predecessors.
32083 (edited_line::apply_fixit): Handle newlines.
32084 (edited_line::get_effective_line_count): New function.
32085 (edited_line::print_content): New function.
32086 (edited_line::print_diff_lines): New function.
32087 (selftest::test_applying_fixits_insert_containing_newline): New
32089 (selftest::test_applying_fixits_replace_containing_newline): New
32091 (selftest::insert_line): New function.
32092 (selftest::test_applying_fixits_multiple_lines): Add example of
32094 (selftest::edit_context_c_tests): Call the new tests.
32096 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32098 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32099 parameter cand. Update dump information.
32100 (get_computation_cost): Update uses.
32102 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32104 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32105 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32106 (get_computation_at, rewrite_use_address): Update use of
32107 get_computation_aff.
32109 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32111 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32112 (get_computation): Delete.
32113 (get_computation_cost): Implement like get_computation_cost_at.
32114 Use get_computation_at.
32115 (get_computation_cost_at): Delete.
32116 (rewrite_use_nonlinear_expr): Use get_computation_at.
32117 (rewrite_use_compare, remove_unused_ivs): Ditto.
32119 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32121 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32123 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32125 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32126 (ivopts_global_cost_for_size): Rename parameter and update uses.
32127 (iv_ca_recount_cost): Update uses.
32128 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32129 candidates seperately in n_invs and n_cands.
32130 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32132 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32134 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32135 (find_inv_vars_cb): New.
32136 (find_depends): Renamed to ...
32137 (find_inv_vars): ... this.
32138 (add_candidate_1, force_var_cost): Call find_inv_vars.
32139 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32141 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32143 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32144 inv_vars. Add inv_exprs.
32145 (struct iv_cand): Rename depends_on to inv_vars.
32146 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32147 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32148 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32149 (dump_cand): Dump inv_vars.
32150 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32151 (record_invariant, find_depends, add_candidate_1): Ditto.
32152 (set_group_iv_cost, force_var_cost): Ditto.
32153 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32154 (get_computation_cost_at, get_computation_cost): Ditto.
32155 (determine_group_iv_cost_generic): Ditto.
32156 (determine_group_iv_cost_address): Ditto.
32157 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32158 (determine_group_iv_costs): Ditto.
32159 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32160 (iv_ca_set_remove_invariants): Renamed to ...
32161 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32162 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32163 (iv_ca_set_add_invariants): Renamed to ...
32164 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32165 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32166 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32167 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32168 (create_new_ivs): Remove useless dump.
32170 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32172 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32174 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32175 (iv_ca_set_no_cp, create_new_iv): Ditto.
32177 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32179 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32181 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32183 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32184 function tree_check2.
32186 2017-05-02 Martin Liska <mliska@suse.cz>
32188 * doc/gcov.texi: Add missing preposition.
32189 * gcov.c (function_info::function_info): Properly fill up
32190 all member variables.
32192 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32194 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32196 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32198 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32200 2017-05-02 Martin Liska <mliska@suse.cz>
32203 * lto-streamer-in.c (lto_read_tree_1): Remove
32204 LTO_STREAMER_DEBUG.
32205 * lto-streamer.c (struct tree_hash_entry): Likewise.
32206 (struct tree_entry_hasher): Likewise.
32207 (tree_entry_hasher::hash): Likewise.
32208 (tree_entry_hasher::equal): Likewise.
32209 (lto_streamer_init): Likewise.
32210 (lto_orig_address_map): Likewise.
32211 (lto_orig_address_get): Likewise.
32212 (lto_orig_address_remove): Likewise.
32213 * lto-streamer.h: Likewise.
32214 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32215 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32217 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32219 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32220 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32221 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32222 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32223 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32224 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32225 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32226 (mm_maskz_sub_ss): New intrinsics.
32227 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32228 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32229 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32230 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32231 (__builtin_ia32_subss_mask_round): New builtins.
32232 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32233 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32234 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32236 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32237 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32239 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32242 2017-05-02 Martin Jambor <mjambor@suse.cz>
32244 PR tree-optimization/78687
32245 * tree-sra.c (access): New field parent.
32246 (process_subtree_disqualification): New function.
32247 (disqualify_candidate): Call it.
32248 (build_accesses_from_assign): Reset write flag if creating an
32250 (build_access_subtree): Fill in parent field and also prpagate
32251 down grp_write flag.
32252 (create_artificial_child_access): New parameter set_grp_write, set
32253 grp_write to its value.
32254 (propagate_subaccesses_across_link): Also propagate grp_write flag
32256 (propagate_all_subaccesses): Push the closest parent back to work
32257 queue if add_access_to_work_queue returned true.
32259 2017-05-02 Richard Biener <rguenther@suse.de>
32261 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32262 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32263 -fstrict-overflow documentation.
32264 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32265 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32266 flag_strict_overflow.
32267 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32268 * lto-opts.c (lto_write_options): Do not stream it.
32269 * lto-wrapper.c (merge_and_complain): Do not handle it.
32270 * opts.c (default_options_table): Do not set -fstrict-overflow.
32271 (finish_options): Likewise do not clear it when sanitizing.
32272 * simplify-rtx.c (simplify_const_relational_operation): Do not
32273 test flag_strict_overflow.
32275 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32277 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32278 using enabled attribute.
32279 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32280 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32281 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32282 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32283 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32284 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32285 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32286 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32287 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32288 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32290 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32292 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32294 2017-05-02 Richard Biener <rguenther@suse.de>
32296 PR tree-optimization/80591
32298 2017-04-10 Richard Biener <rguenther@suse.de>
32300 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32303 2017-05-02 Richard Biener <rguenther@suse.de>
32305 PR tree-optimization/80549
32306 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32307 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32308 headers if they do not have a preheader.
32310 2017-05-02 Martin Liska <mliska@suse.cz>
32313 * common.opt: Fix typo.
32314 * doc/invoke.texi: Likewise.
32316 2017-05-01 Jan Beulich <jbeulich@suse.com>
32318 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32319 swapping, add (x,x,m,x,n) alternative.
32321 2017-05-01 Nathan Sidwell <nathan@acm.org>
32323 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32324 unnecessary unadjusted_alignment check.
32326 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32329 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32331 * gimplify.c (gimplify_cilk_detach): New function.
32332 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32333 * tree-core.h: Document EXPR_CILK_SPAWN.
32334 * tree.h (EXPR_CILK_SPAWN): Define.
32336 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32338 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32339 to use new fixit_hint representation, using the "replace" logic.
32340 (get_line_span_for_fixit_hint): Likewise.
32341 (layout::print_any_fixits): Likewise.
32342 (selftest::test_one_liner_many_fixits): Rename to...
32343 (selftest::test_one_liner_many_fixits_1): ...this, and update
32344 comment and expected output to reflect that the multiple fix-it
32345 hints are now consolidated into one insertion.
32346 (selftest::test_one_liner_many_fixits_2): New test.
32347 (selftest::test_diagnostic_show_locus_one_liner): Update for
32349 (selftest::test_fixit_consolidation): Update for fix-it API
32351 * diagnostic.c (print_parseable_fixits): Likewise.
32352 * edit-context.c (edited_line::m_line_events): Convert from
32353 auto_vec <line_event *> to auto_vec <line_event>.
32354 (class line_event): Convert from abstract base class to a concrete
32355 class, taking over the role of replace_event.
32356 (class insert_event): Delete.
32357 (class replace_event): Rename to class line_event. Convert to
32359 (edit_context::add_fixits): Reimplement.
32360 (edit_context::apply_insert): Delete.
32361 (edit_context::apply_replace): Rename to...
32362 (edit_context::apply_fixit): ...this. Convert to half-open range.
32363 (edited_file::apply_insert): Delete.
32364 (edited_file::apply_replace): Rename to...
32365 (edited_file::apply_fixit): ...this.
32366 (edited_line::~edited_line): Drop deletion of events.
32367 (edited_line::apply_insert): Delete.
32368 (edited_line::apply_replace): Rename to...
32369 (edited_line::apply_fixit): ...this. Convert to half-open range.
32370 Update for change to type of m_line_events.
32371 * edit-context.h (edit_context::apply_insert): Delete.
32372 (edit_context::apply_replace): Rename to...
32373 (edit_context::apply_fixit): ...this.
32375 2017-05-01 Martin Sebor <msebor@redhat.com>
32377 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32380 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32383 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32384 __get_cpuid_max returns 0.
32385 (__get_cpuid_count): Ditto.
32387 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32389 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32390 replacement expression is another instance of one of its arguments.
32392 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32395 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32396 check for stack push/pop autoinc.
32397 * config/i386/i386.c (ix86_agi_dependent): Return false
32398 if the only reason why modified_in_p returned true is that
32399 addr is SP based and set_insn is a push or pop.
32401 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32403 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32406 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32409 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32410 (account_size_time): Use two predicates - exec_pred and
32412 (evaluate_conditions_for_known_args): Compute both clause and
32414 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32415 (inline_summary_t::duplicate): Update.
32416 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32418 (compute_inline_parameters): Likewise.
32419 (estimate_edge_size_and_time): Update caluclation of time.
32420 (estimate_node_size_and_time): Compute both time and nonspecialized
32422 (estimate_ipcp_clone_size_and_time): Update.
32423 (inline_merge_summary): Update.
32424 (do_estimate_edge_time): Update.
32425 (do_estimate_edge_size): Update.
32426 (do_estimate_edge_hints): Update.
32427 (inline_read_section, inline_write_summary): Stream both new predicates.
32428 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32430 (compute_inlined_call_time): Cleanup.
32431 (big_speedup_p): Update.
32432 (edge_badness): Update.
32433 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
32434 (size_time_entry): Replace predicate by exec_predicate and
32435 nonconst_predicate.
32436 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
32437 (estimate_edge_time): Return also nonspec_time.
32438 (reset_edge_growth_cache): Update.
32440 2017-04-29 Jakub Jelinek <jakub@redhat.com>
32442 PR rtl-optimization/80491
32443 * ifcvt.c (noce_process_if_block): When looking for x setter
32444 with missing else_bb, don't check only the insn right before
32445 cond_earliest, but look for the last insn that x is modified in
32446 within the same bb.
32448 PR rtl-optimization/80491
32449 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
32451 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
32453 PR tree-optimization/80487
32454 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
32456 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32458 PR tree-optimization/79697
32459 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
32460 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
32461 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
32463 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
32464 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
32466 2017-04-28 Martin Sebor <msebor@redhat.com>
32468 PR tree-optimization/80523
32469 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
32470 (init_target_to_host_charmap, target_to_host, target_strtol10): New
32472 (maybe_warn, format_directive, parse_directive): Use new functions.
32473 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
32475 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32477 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
32479 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
32481 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
32482 target_header_dir): Set correctly.
32483 * configure: Regenerated.
32484 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32485 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32486 instead of SYSTEM_HEADER_DIR.
32488 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
32490 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
32491 (estimate_local_effects): Likewise.
32492 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
32493 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
32494 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
32495 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
32496 do_estimate_edge_time, estimate_edge_time): Likewise.
32497 * ipa-inline-analysis.c (estimate_node_size_and_time,
32498 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
32499 (estimate_time_after_inlining): Remove.
32501 2017-04-28 Martin Liska <mliska@suse.cz>
32503 * doc/gcov.texi: Enhance documentation of gcov.
32505 2017-04-28 Martin Liska <mliska@suse.cz>
32507 * doc/gcov.texi: Sort options in alphabetic order.
32508 * doc/gcov-dump.texi: Likewise.
32509 * doc/gcov-tool.texi: Likewise.
32510 * gcov.c (print_usage): Likewise.
32511 * gcov-dump.c (print_usage): Likewise.
32512 * gcov-tool.c (print_merge_usage_message): Likewise.
32513 (print_rewrite_usage_message): Likewise.
32514 (print_overlap_usage_message): Likewise.
32516 2017-04-28 Martin Liska <mliska@suse.cz>
32518 PR gcov-profile/53915
32519 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
32521 2017-04-28 Martin Liska <mliska@suse.cz>
32523 PR gcov-profile/79891
32524 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
32525 is marked by compiler as living on a line.
32526 (get_cycles_count): Remove usage of the union.
32527 (output_intermediate_file): Likewise.
32528 (find_source): Fix GNU coding style.
32529 (accumulate_line_counts): Remove old non-all block mode.
32530 (output_lines): Remove usage of the union.
32531 * profile.c (output_location): Include all BBs, even if
32532 belonging to a same line (and file) as a previous BB.
32534 2017-04-28 Martin Liska <mliska@suse.cz>
32536 * gcov.c (process_args): Handle new argument 'w'.
32537 (read_graph_file): Assign ID to BBs.
32538 (output_branch_count): Display BB # if verbose flag is set.
32539 (output_lines): Likewise for arcs.
32540 (print_usage): Add '--verbose' option help.
32541 * doc/gcov.texi: Document --verbose (-w) option.
32543 2017-04-28 Martin Liska <mliska@suse.cz>
32545 * gcov.c (struct block_location_info): New struct.
32546 (process_file): Fill up the new structure.
32547 (read_graph_file): Replace usage of encoding by the newly added
32549 (add_line_counts): Likewise.
32550 (accumulate_line_counts): Remove usage of the union.
32551 (function_info::function_info): New function.
32552 (function_info::~function_info): Likewise.
32553 (process_file): Call delete instead of release_function.
32554 (release_function): Release the function.
32555 (release_structures): Call delete instead of release_function.
32556 (solve_flow_graph): Replace usage of num_blocks.
32557 (find_exception_blocks): Likewise.
32558 (output_lines): Fix GNU coding style.
32560 2017-04-28 Martin Liska <mliska@suse.cz>
32563 * coverage.c (coverage_remove_note_file): New function.
32564 * coverage.h: Declare the function.
32565 * toplev.c (finalize): Clean if an error has been seen.
32567 2017-04-28 Martin Liska <mliska@suse.cz>
32569 PR gcov-profile/80031
32570 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
32571 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
32572 * gcov.c (read_graph_file): Read just number of blocks.
32573 * profile.c (branch_prob): Do not stream 0 flags per a basic
32576 2017-04-28 Martin Liska <mliska@suse.cz>
32578 * gcov-dump.c (tag_*): Add new argument to declarations.
32579 (dump_gcov_file): Likewise.
32580 (tag_blocks): Add and use new argument depth.
32581 (tag_arcs): Likewise.
32582 (tag_lines): Likewise.
32583 (tag_counters): Likewise.
32584 (tag_summary): Likewise.
32585 (dump_working_sets): Use depth to do a proper indentation.
32587 2017-04-28 Jakub Jelinek <jakub@redhat.com>
32590 * cgraph.h (symtab_node::debug_symtab): No longer inline.
32591 * symtab.c (symtab_node::debug_symtab): Move definition here.
32593 2017-04-28 Richard Biener <rguenther@suse.de>
32595 * lto-streamer.h (LTO_major_version): Bump to 7.
32597 2017-04-28 Richard Biener <rguenther@suse.de>
32599 * tree-vrp.c (assert_info): New struct.
32600 (add_assert_info): New helper.
32601 (register_edge_assert_for_2): Refactor to add asserts to a vector
32603 (register_edge_assert_for_1): Likewise.
32604 (register_edge_assert_for): Likewise.
32605 (finish_register_edge_assert_for): New helper actually registering
32606 asserts where live on edge.
32607 (find_conditional_asserts): Adjust.
32608 (find_switch_asserts): Likewise.
32609 (evrp_dom_walker::try_find_new_range): Generalize.
32610 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
32612 2017-04-27 Marek Polacek <polacek@redhat.com>
32615 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
32616 arg10 and arg11 to itype.
32618 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
32620 * doc/extend.texi (Object Size Checking): Improve grammar.
32622 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
32625 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
32626 that the logic for permitting reciprocal estimates matches that
32629 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32632 * tree.c (type_cache_hasher::equal): Only compare
32633 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32634 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
32635 non-aggregate element types.
32636 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
32637 about the flag on ARRAY_TYPEs in the comment, formatting fix.
32639 2017-04-27 Richard Biener <rguenther@suse.de>
32641 PR middle-end/80533
32642 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
32643 stripping ARRAY_REFs from MEM_EXPR make sure we're not
32644 keeping a reference to a trailing array.
32646 2017-04-27 Richard Biener <rguenther@suse.de>
32648 PR middle-end/80539
32649 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
32650 being in loop-closed SSA form conservatively.
32651 (chrec_fold_multiply_poly_poly): Likewise.
32653 2017-04-27 Tamar Christina <tamar.christina@arm.com>
32655 PR middle-end/79665
32656 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
32657 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
32659 2017-04-27 Jakub Jelinek <jakub@redhat.com>
32662 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
32663 (aarch64_function_arg_alignment): Return unsigned int again, but still
32664 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
32665 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32666 Don't emit -Wpsabi note.
32667 (aarch64_function_arg_boundary): Likewise.
32668 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32671 2017-04-26 Nathan Sidwell <nathan@acm.org>
32673 * tree.h (crc32_unsigned_n): Declare.
32674 (crc32_unsigned, crc32_unsigned): Make inline.
32675 * tree.c (crc32_unsigned_bits): Replace with ...
32676 (crc32_unsigned_n): ... this.
32677 (crc32_unsigned, crc32_byte): Remove.
32678 (crc32_string): Remove unnecessary braces.
32680 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32682 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
32683 * ipa-inline-analysis.c (MAX_TIME): Remove.
32684 (account_size_time): Use sreal for time.
32685 (dump_inline_summary): Update.
32686 (estimate_function_body_sizes): Update.
32687 (estimate_edge_size_and_time): Update.
32688 (estimate_calls_size_and_time): Update.
32689 (estimate_node_size_and_time): Update.
32690 (inline_merge_summary): Update.
32691 (inline_update_overall_summary): Update.
32692 (estimate_time_after_inlining): Update.
32693 (inline_read_section): Update.
32694 (inline_write_summary): Update.
32695 * ipa-inline.c (compute_uninlined_call_time): Update.
32696 (compute_inlined_call_time): Update.
32697 (recursive_inlining): Update.
32698 (inline_small_functions): Update.
32699 (dump_overall_stats): Update.
32700 * ipa-inline.h: Include sreal.h.
32701 (size_time_entry): Turn time to sreal.
32702 (inline_summary): Turn self_time nad time to sreal.
32704 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
32706 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
32708 (sreal::stream_out, sreal::stream_in): New.
32709 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
32711 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32713 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
32716 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
32719 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
32721 (dimode_scalar_chain::compute_convert_gain): Ditto.
32722 (dimode_scalar_chain::make_vector_copies): Ditto.
32723 (dimode_scalar_chain::convert_reg): Ditto.
32724 (dimode_scalar_chain::convert_insn): Ditto.
32725 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
32726 (VI248_AVX512BW_1): New mode iterator.
32727 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
32728 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
32731 2017-04-25 Martin Sebor <msebor@redhat.com>
32733 PR tree-optimization/80497
32734 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
32735 constants are representable in HOST_WIDE_INT.
32736 (parse_directive): Ditto.
32738 2017-04-25 Martin Sebor <msebor@redhat.com>
32741 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
32742 (new_zero_array): Adjust signature.
32743 (dom_info::dom_init): Used unsigned rather that size_t.
32744 (dom_info::dom_info): Same.
32746 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32747 Jakub Jelinek <jakub@redhat.com>
32750 * config/arm/arm.c: Include gimple.h.
32751 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32752 returns negative, increment ncrn only if it returned positive.
32753 (arm_needs_doubleword_align): Return int instead of bool,
32754 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
32755 members, but if there is any such non-FIELD_DECL
32756 > PARM_BOUNDARY aligned decl, return -1 instead of false.
32757 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
32758 returns negative, increment nregs only if it returned positive.
32759 (arm_setup_incoming_varargs): Likewise.
32760 (arm_function_arg_boundary): Emit -Wpsabi note if
32761 arm_needs_doubleword_align returns negative, return
32762 DOUBLEWORD_ALIGNMENT only if it returned positive.
32764 2017-04-25 Marek Polacek <polacek@redhat.com>
32767 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32768 first argument to type.
32770 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32773 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32774 type checks to test for compatibility instead of equality.
32776 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32777 Jakub Jelinek <jakub@redhat.com>
32780 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32782 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32783 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32784 the alignment computation, but return their maximum in warn_alignment.
32785 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32786 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32788 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32789 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32792 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32794 * config/arc/simdext.md (dmpyh): Fix typo.
32796 2017-04-25 Richard Biener <rguenther@suse.de>
32798 PR tree-optimization/80492
32799 * alias.c (compare_base_decls): Handle registers with asm
32800 specification conservatively.
32801 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32802 compare_base_decls returning dont-know properly.
32804 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32806 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32807 (legitimate_offset_address_p): New function.
32808 (arc_legitimate_address_p): Use above function.
32810 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32812 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32814 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32816 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32817 ACCH registers whenever they are available.
32819 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32821 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32822 double regs fix when not used.
32824 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32826 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32828 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32829 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32831 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32833 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32834 when emitting short ADD instructions.
32836 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32838 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32840 (cmpsi_cc_c_insn): Likewise.
32841 (cbranchsi4_scratch): Compute proper instruction length using
32842 compact_hreg_operand.
32843 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32845 2017-04-25 Richard Biener <rguenther@suse.de>
32847 PR middle-end/80509
32848 * passes.c (pass_manager::pass_manager): Initialize
32849 m_name_to_pass_map.
32851 2017-04-25 Richard Biener <rguenther@suse.de>
32853 PR tree-optimization/79201
32854 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32856 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32859 * config/s390/vector.md: Split MEM->GPR vector moves for
32860 non-s_operand addresses.
32862 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32865 * config/s390/predicates.md (reload_const_wide_int_operand): New
32867 * config/s390/s390.md ("movti"): Remove d/P alternative.
32868 ("movti_bigconst"): New pattern definition.
32870 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32873 * s390-protos.h (s390_expand_cs_hqi): Removed.
32874 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32875 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32876 modes as well as CCZ1mode and CCZmode.
32877 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32878 signature of s390_emit_compare_and_swap.
32879 (s390_expand_cs_hqi): Likewise, make static.
32880 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32881 compare-and-swap, in some cases.
32882 (s390_expand_cs): Wrapper function.
32883 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32885 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32886 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32887 patterns for small and large integers. Forbid symref memory operands.
32888 Move expander to s390.c. Require cc register.
32889 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32890 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32891 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32892 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32893 symref memory operands. Remove CC mode and call s390_match_ccmode
32895 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32897 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32899 * config/s390/s390.md (define_peephole2): New peephole to help
32900 combining the load-and-test pattern with volatile memory.
32902 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32904 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32905 with CCZmode for TARGET_Z196.
32907 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32909 PR rtl-optimization/80501
32910 * combine.c (make_compound_operation_int): Set subreg_code to SET
32911 even for AND with mask of the sign bit of mode.
32913 PR rtl-optimization/80500
32914 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32915 sum's initial value.
32917 2017-04-25 Julian Brown <julian@codesourcery.com>
32918 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32920 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32922 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32924 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32926 2017-04-25 Julian Brown <julian@codesourcery.com>
32927 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32929 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32930 (thunderx2t99_sha): New Reservation.
32932 2017-04-25 Julian Brown <julian@codesourcery.com>
32933 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32935 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32936 type for 1-element load.
32938 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32940 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32942 2017-04-24 Martin Jambor <mjambor@suse.cz>
32944 PR tree-optimization/80293
32945 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32946 char arrays not totally scalarizable if it is false.
32947 (analyze_all_variable_accesses): Pass correct value in the new
32948 parameter. Add a statistics counter.
32950 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32952 PR middle-end/79931
32953 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32955 2017-04-24 Richard Biener <rguenther@suse.de>
32957 PR tree-optimization/80494
32958 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32959 out for complex types.
32961 2017-04-24 Richard Biener <rguenther@suse.de>
32963 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32964 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32965 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32966 (DFS): Adjust and never fail.
32967 (sccvn_dom_walker::fail): Remove.
32968 (sccvn_dom_walker::before_dom_children): Adjust.
32969 (run_scc_vn): Likewise and never fail.
32970 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32971 (pass_fre::execute): Likewise.
32973 2017-04-24 Richard Biener <rguenther@suse.de>
32975 PR tree-optimization/79725
32976 * tree-ssa-sink.c (statement_sink_location): Return whether
32977 failure reason was zero uses. Move that check later.
32978 (sink_code_in_bb): Deal with zero uses by removing the stmt
32981 2017-04-24 Richard Biener <rguenther@suse.de>
32984 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32985 pointer-based references.
32987 2017-04-24 Richard Biener <rguenther@suse.de>
32990 * pass_manager.h (pass_manager::operator new): Remove.
32991 (pass_manager::operator delete): Likewise.
32992 * passes.c (pass_manager::operator new): Remove.
32993 (pass_manager::operator delete): Likewise.
32994 (pass_manager::pass_manager): Zero individual pass members.
32996 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32999 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33000 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33001 Check "XEXP (src, 1)" operand here.
33002 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33003 Check "XEXP (src, 1)" operand here.
33004 (dimode_scalar_chain::make_vector_copies): Detect count register
33005 of a shift instruction. Zero extend count register from QImode
33006 to DImode to satisfy vector shift pattern count operand predicate.
33007 Substitute vector shift count operand with a DImode copy.
33008 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33011 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33013 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33014 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33015 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33016 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33017 (UNSPEC_NOREX_MEM): Remove definition.
33019 2017-04-21 Richard Biener <rguenther@suse.de>
33021 PR tree-optimization/79547
33022 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33023 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33024 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33025 without any constraints.
33027 2017-04-21 Richard Biener <rguenther@suse.de>
33029 PR tree-optimization/78847
33030 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33032 2017-04-21 Richard Biener <rguenther@suse.de>
33034 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33035 (build_distinct_type_copy): Likewise.
33036 (build_variant_type_copy): Likewise.
33037 * tree.c (build_qualified_type): Pass down mem-stat info.
33038 (build_distinct_type_copy): Likewise.
33039 (build_variant_type_copy): Likewise.
33041 2017-04-21 Richard Biener <rguenther@suse.de>
33043 PR tree-optimization/80237
33044 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33046 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33047 for a simplified result.
33049 2016-04-21 Richard Biener <rguenther@suse.de>
33051 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33052 sth as strict as a simple_iv but a chrec without symbols and an
33053 operand defined in the loop we are peeling (and not some subloop).
33054 (propagate_constants_for_unrolling): Propagate all constants.
33056 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33059 * config/i386/i386.c (print_reg): Remove assert for disalowed
33060 regno values, call output_operand_lossage instead.
33062 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33065 * config/i386/constraints.md (Yc): New register constraint.
33066 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33067 Use Yc constraint for alternative 2 of operand 0. Remove
33068 preferred_for_speed attribute.
33070 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33072 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33073 lastprivate clauses in SIMT case.
33075 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33077 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33079 2017-04-20 Richard Biener <rguenther@suse.de>
33081 PR tree-optimization/57796
33082 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33083 as N scalar stores.
33084 (vect_model_load_cost): Cost gathers as N scalar loads.
33086 2017-04-20 Richard Biener <rguenther@suse.de>
33088 * ggc-page.c (ggc_allocated_p): Rename to ...
33089 (safe_lookup_page_table_entry): ... this and return the lookup
33091 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33093 2017-04-20 Richard Biener <rguenther@suse.de>
33095 PR tree-optimization/80453
33096 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33097 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33098 from the conditions.
33099 (vn_phi_eq): Pass them down.
33100 (vn_phi_lookup): Record them.
33101 (vn_phi_insert): Likewise.
33103 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33105 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33106 uninitialized variable warning to avoid buffer overrun.
33108 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33111 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33112 is suppressed for '{ 0 }' in C.
33114 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33116 * BASE-VER: Set to 8.0.0.
33118 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33120 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33121 priority .init_array and .fini_array section with SECTION_NOTYPE
33124 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33126 PR middle-end/80423
33127 * tree.h (build_array_type): Add typeless_storage default argument.
33128 * tree.c (type_cache_hasher::equal): Also compare
33129 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33130 (build_array_type): Add typeless_storage argument, set
33131 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33133 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33134 (build_array_type): Likewise. Add typeless_storage argument.
33136 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33137 Jakub Jelinek <jakub@redhat.com>
33139 PR tree-optimization/80426
33140 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33141 operation on symbolic operands, also compute the overflow for the
33142 invariant part when the operation degenerates into a negation.
33144 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33147 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33148 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33151 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33153 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33156 * config/avr/avr.c (tree.h): Include it.
33157 (cgraph.h): Include it.
33158 (avr_encode_section_info): Don't warn for uninitialized progmem
33159 variable if it's just an alias.
33161 2017-04-19 Richard Biener <rguenther@suse.de>
33164 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33165 when needed by AutoPGO.
33167 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33170 * doc/lto.texi: Remove an extra 'that'.
33172 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33174 PR rtl-optimization/80429
33175 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33176 are only used in debug insns.
33178 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33179 Vladimir Makarov <vmakarov@redhat.com>
33181 * config/sparc/predicates.md (input_operand): Add comment. Return
33182 true for any memory operand when LRA is in progress.
33183 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33185 2017-04-18 Jeff Law <law@redhat.com>
33188 * mips.md ({return,simple_return}_internal): Do not overwrite
33191 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33193 PR tree-optimization/80443
33194 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33195 instead of adding 1, subtract -1 and similarly instead of subtracting
33198 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33200 PR rtl-optimization/80357
33201 * haifa-sched.c (tmp_bitmap): New variable.
33202 (model_recompute): Handle duplicate use records.
33203 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33204 (free_global_sched_pressure_data): Free it.
33206 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33209 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33210 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33211 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33212 instead of SYSTEM_HEADER_DIR.
33214 2017-04-18 Jeff Law <law@redhat.com>
33216 PR middle-end/80422
33217 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33218 predecessors after walking up the insn chain.
33220 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33223 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33224 sizetype type into debug info.
33226 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33229 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33230 unneeded test for TARGET_UPPER_REGS_SF.
33231 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33233 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33236 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33237 instead of gsi_after_labels.
33239 2017-04-18 Jeff Law <law@redhat.com>
33241 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33245 2017-04-13 Jeff Law <law@redhat.com>
33246 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33247 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33249 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33252 * config/avr/avr.c (intl.h): Include it.
33253 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33255 2017-04-18 Martin Liska <mliska@suse.cz>
33257 PR gcov-profile/78783
33258 * gcov-tool.c (gcov_output_files): Validate that destination
33259 file is either removed by the tool or by a user.
33261 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33262 Guy Benyei <guybe@mellanox.com>
33264 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33265 block, and do not negate it, the stored id is already negative.
33267 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33269 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33271 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33274 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33275 masks of options that should be turned off if the VSX vector
33276 options are turned off.
33277 (OTHER_P8_VECTOR_MASKS): Likewise.
33278 (OTHER_VSX_VECTOR_MASKS): Likewise.
33279 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33280 rs6000_disable_incompatible_switches to validate no type switches
33282 (rs6000_incompatible_switch): New function to disallow turning on
33283 other vector options if -mno-vsx, -mno-power8-vector, or
33284 -mno-power9-vector are specified.
33286 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33288 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33290 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33292 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33293 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33294 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33295 (ARG_POINTER_CFA_OFFSET): Likewise.
33297 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33299 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33300 conditions to take advantage of various optimizations.
33302 2017-04-13 Jeff Law <law@redhat.com>
33304 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33305 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33306 (zero_extendsidi2_dext): Likewise.
33308 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33311 * fold-const.c (fold_ternary_loc): Revert
33312 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33315 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33317 PR rtl-optimization/80343
33318 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33319 new scratch pseudo.
33321 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33324 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33325 to ubsan_encode_value.
33327 2017-04-13 Jeff Law <law@redhat.com>
33329 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33330 appearing in DEBUG_INSNs.
33332 2017-04-13 Martin Liska <mliska@suse.cz>
33334 PR gcov-profile/80413
33335 * gcov-io.c (gcov_write_string): Copy to buffer just when
33336 allocated size is greater than zero.
33338 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33341 * dwarf2out.c (decls_for_scope): Ignore declarations of
33342 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33344 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33347 * ipa-visibility.c (non_local_p): Fix typos.
33348 (localize_node): When localizing symbol in same comdat group,
33349 dissolve the group only when we know external symbols are going
33351 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33353 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33355 PR tree-optimization/79390
33356 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33357 order does not result in usable sequence, retry with reversed operand
33363 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33364 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33365 op0 instead of fold_convert_loc (loc, type, arg0).
33367 2017-04-12 Jeff Law <law@redhat.com>
33369 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33370 has a delay slot in the generated code.
33372 * config/cris/cris.md (cris_preferred_reload_class): Return
33373 GENNONACR_REGS rather than GENERAL_REGS.
33375 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33378 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33379 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33380 signedness of the result type.
33382 2017-04-12 Richard Biener <rguenther@suse.de>
33383 Jeff Law <law@redhat.com>
33385 PR tree-optimization/80359
33386 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33387 trim stores to TARGET_MEM_REFs.
33389 2017-04-12 Richard Biener <rguenther@suse.de>
33391 PR tree-optimization/79390
33392 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33393 threading case even more.
33395 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33398 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33399 for quad_address_p for TImode, instead of just not indexed_address.
33401 2017-04-12 Richard Biener <rguenther@suse.de>
33402 Bernd Edlinger <bernd.edlinger@hotmail.de>
33404 PR middle-end/79671
33405 * alias.c (component_uses_parent_alias_set_from): Handle
33406 TYPE_TYPELESS_STORAGE.
33407 (get_alias_set): Likewise.
33408 * tree-core.h (tree_type_common): Add typeless_storage flag.
33409 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33410 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33411 for types containing members with TYPE_TYPELESS_STORAGE.
33412 (place_field): Likewise.
33413 (layout_type): Likewise for ARRAY_TYPE.
33414 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33415 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33416 TYPE_TYPELESS_STORAGE.
33417 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33419 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33422 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33423 first argument to type.
33425 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33429 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33430 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33431 (rs6000_expand_binop_builtin): Likewise.
33432 (rs6000_expand_ternop_builtin): Likewise; also add missing
33433 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
33434 vshasigma built-ins.
33435 * doc/extend.texi: Document that vec_xxpermdi's third argument
33436 must be a constant.
33438 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
33440 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
33441 Use shift_const cost parameter when calculating gain of STV shifts.
33443 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
33445 PR rtl-optimization/70478
33446 * lra-constraints.c (process_alt_operands): Check memory for
33447 disfavoring memory insn operand.
33449 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33451 PR middle-end/80100
33452 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
33453 left shift in unsigned HOST_WIDE_INT type.
33455 PR rtl-optimization/80385
33456 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
33457 (not (neg X)) into (plus X -1) for complex or non-integral modes.
33460 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
33461 if they have any depend clauses.
33463 2017-04-11 Martin Liska <mliska@suse.cz>
33466 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
33467 * ipa-split.c (split_function): Create a local comdat symbol
33468 if caller is in a comdat group.
33470 2017-04-11 Martin Liska <mliska@suse.cz>
33473 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
33476 2017-04-11 Martin Sebor <msebor@redhat.com>
33478 PR middle-end/80364
33479 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
33480 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
33482 (directive::set_width, directive::set_precision, format_character):
33484 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
33487 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
33490 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
33491 conflict, set target->arch_name instead of target->cpu_name.
33493 2017-04-11 Richard Biener <rguenther@suse.de>
33495 PR tree-optimization/80374
33496 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
33497 build_zero_cst, remove fold_convertible_p check again.
33499 2017-04-11 Martin Liska <mliska@suse.cz>
33502 * ubsan.c (instrument_object_size): Do not instrument register
33505 2017-04-11 Jakub Jelinek <jakub@redhat.com>
33508 * config/i386/i386-builtin-types.def
33509 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
33510 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
33511 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
33512 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
33513 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
33514 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
33515 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
33516 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
33517 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
33518 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
33519 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
33520 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
33521 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
33522 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
33523 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
33524 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
33525 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
33526 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
33527 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
33528 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
33529 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
33530 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
33531 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
33532 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
33533 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
33534 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
33535 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
33536 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
33537 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
33538 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
33539 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
33540 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
33541 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
33542 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
33543 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
33544 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
33545 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
33546 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
33547 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
33548 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
33549 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
33550 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
33551 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
33552 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
33553 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
33554 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
33556 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
33557 flag to second_arg_count, handle 4 argument function type _COUNT
33558 aliases, handle second_arg_count on second argument rather than last.
33560 2017-04-10 Jeff Law <law@redhat.com>
33562 PR tree-optimization/80374
33563 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
33564 record anything if we can not convert integer_zero_node to the
33567 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33570 * config/rs6000/rs6000.c (rs6000_option_override_internal):
33571 Enhance special handling given to the TARGET_P9_MINMAX option in
33572 relation to certain other options.
33574 2017-04-10 Bin Cheng <bin.cheng@arm.com>
33576 PR tree-optimization/80153
33577 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
33578 remove POINTER_PLUS_EXPR's base part directly, rather than through
33581 2017-04-10 Richard Biener <rguenther@suse.de>
33582 Bin Cheng <bin.cheng@arm.com>
33584 PR tree-optimization/80153
33585 * tree-affine.c (aff_combination_to_tree): Get base pointer from
33586 the first element of pointer type aff_tree. Build result expr in
33588 (add_elt_to_tree): Convert to type unconditionally. Remove other
33589 fold_convert calls.
33590 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
33591 (rewrite_use_nonlinear_expr): Check invariant using iv information.
33593 2017-04-10 Richard Biener <rguenther@suse.de>
33595 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33598 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
33600 PR rtl-optimization/70478
33601 * lra-constraints.c (curr_small_class_check): New.
33602 (update_and_check_small_class_inputs): New.
33603 (process_alt_operands): Update curr_small_class_check. Disfavor
33604 alternative insn memory operands. Check available regs for small
33607 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33610 * config/mips/mips.opt (-mvirt): Update description.
33611 * doc/invoke.texi (-mvirt): Likewise.
33613 2017-04-10 Richard Biener <rguenther@suse.de>
33615 PR middle-end/80362
33616 * fold-const.c (fold_binary_loc): Look at unstripped ops when
33617 looking for NEGATE_EXPR in -A / -B to A / B folding.
33619 2017-04-10 Martin Liska <mliska@suse.cz>
33621 PR gcov-profile/80224
33622 * gcov.c (print_usage): Fix usage string.
33623 (get_gcov_intermediate_filename): Remove.
33624 (output_gcov_file): Use both for normal and intermediate format.
33625 (generate_results): Do not initialize special file for
33626 intermediate format.
33628 2017-04-10 Richard Biener <rguenther@suse.de>
33630 PR tree-optimization/80304
33631 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
33634 2017-04-10 Nathan Sidwell <nathan@acm.org>
33637 * config/rs6000/rs6000.c (rs6000_vector_type): New.
33638 (rs6000_init_builtins): Use it.
33640 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33642 * config/arm/arm.md (<mrc>): Add mode to SET source.
33643 (<mrrc>): Likewise.
33645 2017-04-10 Richard Biener <rguenther@suse.de>
33647 PR middle-end/80344
33648 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
33650 2017-04-10 Jakub Jelinek <jakub@redhat.com>
33653 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
33654 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
33655 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
33656 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
33657 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
33658 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
33659 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
33660 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
33661 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
33662 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
33663 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
33664 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
33665 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
33666 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
33667 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
33668 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
33669 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
33670 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
33671 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
33672 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
33673 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
33674 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
33675 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
33677 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
33679 PR rtl-optimization/70478
33680 * lra-constraints.c: Reverse the last patch.
33682 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
33684 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
33685 Add comment for WCHAR_T.
33687 2017-04-08 Martin Liska <mliska@suse.cz>
33690 2017-04-07 Martin Liska <mliska@suse.cz>
33693 * ipa-split.c (split_function): Add function part to a same comdat
33696 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33699 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
33701 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
33703 * rs6000/rs6000.c (vec_load_pendulum): Rename...
33704 (vec_pairing): ...to this.
33705 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
33706 (rs6000_sched_init): Adjust for name change.
33707 (struct rs6000_sched_context): Likewise.
33708 (rs6000_init_sched_context): Likewise.
33709 (rs6000_set_sched_context): Likewise.
33711 2017-04-07 Jakub Jelinek <jakub@redhat.com>
33717 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
33719 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
33720 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
33721 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
33723 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
33725 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
33727 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33729 PR rtl-optimization/70703
33730 * ira-color.c (update_conflict_hard_regno_costs): Use
33731 int64_t instead of HOST_WIDE_INT.
33733 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
33735 PR rtl-optimization/70478
33736 * lra-constraints.c (process_alt_operands): Disfavor alternative
33737 insn memory operands.
33739 2017-04-07 Jeff Law <law@redhat.com>
33741 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
33742 CALL and NOTE_INSN_CALL_ARG_LOCATION.
33744 2017-04-07 Martin Liska <mliska@suse.cz>
33747 * config/aarch64/aarch64.c (aarch64_process_target_attr):
33748 Show error message instead of an ICE.
33750 2017-04-07 Martin Liska <mliska@suse.cz>
33753 * ipa-split.c (split_function): Add function part to a same comdat
33756 2017-04-07 Richard Biener <rguenther@suse.de>
33758 PR middle-end/80341
33759 * tree.c (get_unwidened): Also handle ! for_type case for
33761 * convert.c (do_narrow): Split out from ...
33762 (convert_to_integer_1): ... here. Do not pass final truncation
33763 type to get_unwidened for TRUNC_DIV_EXPR.
33765 2017-04-07 Richard Biener <rguenther@suse.de>
33767 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33769 (aff_combination_const): Adjust.
33770 (aff_combination_scale): Likewise.
33771 (aff_combination_add_elt): Likewise.
33772 (aff_combination_add_cst): Likewise.
33773 (aff_combination_convert): Likewise.
33774 (add_elt_to_tree): Likewise. Remove unused argument.
33775 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33777 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33779 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33781 * config/arm/arm.c (arm_default_short_enums): Use
33782 ARM_DEFAULT_SHORT_ENUMS.
33783 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33785 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33788 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33789 members with redundant out-of-class redeclaration.
33791 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33794 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33795 * config/i386/i386.md (*zero_extendsidi2):
33796 Add (?*x,*x) and (?*v,*v) alternatives.
33798 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33801 * config/i386/i386.c (ix86_expand_builtin)
33802 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33803 mode from insn data. Convert operands to insn operand mode.
33804 Copy operands that don't satisfy insn predicate to a register.
33806 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33808 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33811 2017-04-06 Richard Biener <rguenther@suse.de>
33813 PR tree-optimization/80334
33814 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33815 preserve alignment of accesses.
33817 2017-04-06 Richard Biener <rguenther@suse.de>
33819 PR tree-optimization/80262
33820 * tree-sra.c (build_ref_for_offset): Preserve address-space
33822 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33823 Drop useless address-space information on MEM_REF offsets.
33825 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33827 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33829 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33831 PR rtl-optimization/70703
33832 * ira-color.c (update_conflict_hard_regno_costs): Use
33833 HOST_WIDE_INT instead of long.
33835 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33838 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33839 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33841 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33842 for x86_64 target. Handle -m3dnowa option.
33844 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33846 PR rtl-optimization/70703
33847 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33848 (update_conflict_hard_regno_costs): Use long instead of unsigned
33849 arithmetic for cost calculation.
33851 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33852 Bernd Edlinger <bernd.edlinger@hotmail.de>
33855 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33858 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33861 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33862 ptr_mode with Pmode throughout.
33863 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33864 into probe_stack_range and use DImode.
33866 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33869 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33870 call_eh_return is true.
33872 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33874 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33875 Initialize last_match_fntype_index.
33877 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33880 * tree-nvr.c: Include internal-fn.h.
33881 (pass_return_slot::execute): Ignore internal calls without
33884 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33885 Richard Biener <rguenther@suse.de>
33888 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33889 captures used multiple times, except for the last use.
33890 * generic-match-head.c: Include gimplify.h.
33892 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33894 PR tree-optimization/79390
33895 * target.h (struct noce_if_info): Declare.
33896 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33897 * target.def (noce_conversion_profitable_p): New target hook.
33898 * ifcvt.h (struct noce_if_info): New type, moved from ...
33899 * ifcvt.c (struct noce_if_info): ... here.
33900 (noce_conversion_profitable_p): Renamed to ...
33901 (default_noce_conversion_profitable_p): ... this. No longer
33903 (noce_try_store_flag_constants, noce_try_addcc,
33904 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33905 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33906 instead of noce_conversion_profitable_p.
33907 * config/i386/i386.c: Include ifcvt.h.
33908 (ix86_option_override_internal): Don't override
33909 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33910 (ix86_noce_conversion_profitable_p): New function.
33911 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33912 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33913 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33914 * doc/tm.texi: Regenerated.
33916 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33918 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33921 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33924 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33925 instructions for small multiply cores.
33927 2017-04-04 Jeff Law <law@redhat.com>
33929 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33931 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33932 that are not set by the loop over the elements.
33934 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33937 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33938 int mode, convert_modes it to mode as unsigned, otherwise use
33939 lowpart_subreg to mode rather than SImode.
33940 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33941 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33942 Use DImode instead of SImode for the shift count operand.
33943 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33946 2017-04-04 Richard Biener <rguenther@suse.de>
33948 PR middle-end/80281
33949 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33950 arithmetic done for the negate or the plus. Simplify.
33951 (A - (-B) -> A + B): Likewise.
33952 * fold-const.c (split_tree): Make sure to not negate pointers.
33954 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33956 PR rtl-optimization/60818
33957 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33958 a compare of comparisons with the thing compared if this results
33959 in a different machine mode.
33961 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33963 * alias.c (base_alias_check): Fix typo in comment.
33964 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33965 * cgraphunit.c (symbol_table::compile): Likewise.
33966 * collect2.c (maybe_run_lto_and_relink): Likewise.
33967 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33968 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33969 * config/avr/avr.c (avr_map_op_t): Likewise.
33970 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33971 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33972 * config/epiphany/epiphany.md (movcc): Likewise.
33973 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33974 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33976 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33977 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33978 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33979 * config/sh/sh.c (sh_rtx_costs): Likewise.
33980 * fold-const.c (fold_truth_andor): Likewise.
33981 * genautomata.c (collapse_flag): Likewise.
33982 * gengtype.h (struct type::u::s): Likewise.
33983 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33984 * input.c (FORMAT_AMOUNT): Likewise.
33985 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33986 (known_aggs_to_agg_replacement_list): Likewise.
33987 * ipa-inline-analysis.c: Likewise.
33988 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33989 * ipa-polymorphic-call.c
33990 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33991 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33992 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33994 * modulo-sched.c (apply_reg_moves): Likewise.
33995 * omp-expand.c (build_omp_regions_1): Likewise.
33996 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33997 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33998 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33999 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34000 * value-prof.c: Likewise.
34001 * var-tracking.c (val_reset): Likewise.
34003 2017-04-03 Richard Biener <rguenther@suse.de>
34005 PR tree-optimization/80275
34006 * fold-const.c (split_address_to_core_and_offset): Handle
34009 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34011 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34012 descriptors is at least equal to that of functions.
34014 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34016 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34018 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34021 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34022 (mov<IMOD4:mode>): New expander.
34023 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34025 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34027 PR rtl-optimization/79405
34028 * fwprop.c (propagations_left): New variable.
34029 (forward_propagate_into): Decrement it.
34030 (fwprop_init): Initialize it.
34031 (fw_prop): If the variable has reached zero, stop propagating.
34032 (fwprop_addr): Ditto.
34034 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34037 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34038 a FUNCTION_DECL, pass it as decl instead of origin to
34041 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34043 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34046 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34049 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34050 TARGET_VSX_SMALL_INTEGER.
34052 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34054 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34055 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34057 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34059 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34060 extraction from odd-numbered MSA register.
34062 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34064 PR middle-end/80173
34065 * expmed.c (store_bit_field_1): Don't attempt to create
34066 a word subreg out of hard registers wider than word if they
34067 have HARD_REGNO_NREGS of 1 for their mode.
34069 PR middle-end/80163
34070 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34071 conversions to integer types wider than word and pointer.
34074 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34075 (rtx_equal_for_cselib_p): Pass 0 to it.
34076 * cselib.c (cselib_hasher::equal): Likewise.
34077 (rtx_equal_for_cselib_1): Add depth argument. If depth
34078 is 128, don't look up VALUE locs and punt. Increment
34079 depth in recursive calls when walking VALUE locs.
34081 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34083 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34084 (make_gcov_file_name): Use the canonical path name for generating
34086 (read_line): Fix handling of files with ascii null bytes.
34088 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34090 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34091 to initialise a vector register instead
34092 of using a const_int.
34094 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34096 PR translation/80189
34097 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34098 diagnostic messages.
34100 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34103 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34104 (dfp_diex_<mode>): Update mode of operand 1.
34105 * doc/extend.texi (dxex, dxexq): Document change to return type.
34106 (diex, diexq): Document change to argument type.
34108 2017-03-30 Martin Jambor <mjambor@suse.cz>
34111 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34112 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34113 it reflects the signature changes performed at the callee side.
34114 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34115 to cgraph_build_function_type_skip_args.
34116 (build_function_decl_skip_args): Adjust call to the above function.
34118 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34121 * config/i386/sse.md
34122 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34123 register as dest whenever it is a MEM not rtx_equal_p to the
34124 corresponding dup operand, and when forcing into reg move the
34125 reg into the memory afterwards.
34126 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34127 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34128 for the force_reg mode.
34129 (avx512vl_vextractf128<mode>): Use register as dest either
34130 always when a MEM, or when it is a MEM not rtx_equal_p to the
34131 corresponding dup operand, or even not when it is a CONST_VECTOR
34132 depending on the mode and lo vs. hi.
34133 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34135 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34136 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34137 Likewise. Require that operands[2] is even.
34138 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34139 Remove extraneous parens. Require that operands[2] is a multiple
34141 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34142 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34143 disallow memory then.
34145 2017-03-30 Richard Biener <rguenther@suse.de>
34147 PR tree-optimization/77498
34148 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34149 to non-constants over backedges.
34151 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34153 PR rtl-optimization/80233
34154 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34155 as last_combined_insn. Do not test for BARRIER_P separately.
34157 2017-03-29 Andreas Schwab <schwab@suse.de>
34160 * calls.c (prepare_call_address): Convert funexp to Pmode before
34161 copying to temp reg.
34163 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34165 PR tree-optimization/80158
34166 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34167 Handle possible future case of more than one alternate
34169 (replace_rhs_if_not_dup): Likewise.
34170 (replace_one_candidate): Likewise.
34172 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34174 PR rtl-optimization/80193
34175 * ira.c (ira): Do not check allocation for LRA.
34177 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34179 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34180 (nvptx_output_simt_exit): Declare.
34181 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34182 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34183 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34184 (nvptx_declare_function_name): ...here. Emit declaration of local
34185 memory space buffer for omp_simt_enter insn.
34186 (nvptx_output_unisimt_switch): New.
34187 (nvptx_output_softstack_switch): New.
34188 (nvptx_output_simt_enter): New.
34189 (nvptx_output_simt_exit): New.
34190 * config/nvptx/nvptx.h (struct machine_function): New fields
34191 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34192 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34193 (UNSPECV_SIMT_EXIT): Ditto.
34194 (omp_simt_enter_insn): New insn.
34195 (omp_simt_enter): New expansion.
34196 (omp_simt_exit): New insn.
34197 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34199 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34200 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34201 (expand_GOMP_SIMT_EXIT): New.
34202 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34203 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34204 (GOMP_SIMT_EXIT): Ditto.
34205 * target-insns.def (omp_simt_enter): New insn.
34206 (omp_simt_exit): Ditto.
34207 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34209 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34210 (lower_rec_input_clauses): Likewise.
34211 (lower_lastprivate_clauses): Handle SIMT privatization.
34213 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34214 (ompdevlow_adjust_simt_enter): New.
34215 (find_simtpriv_var_op): New.
34216 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34217 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34219 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34220 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34221 (copy_decl_for_dup_finish): Ditto.
34223 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34225 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34228 * config/i386/i386.c (ix86_option_override_internal): Always
34229 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34231 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34233 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34235 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34237 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34238 mark new edge's irreducible flag accordign to it.
34239 (vect_do_peeling): Check loop preheader edge's irreducible flag
34240 and pass it to function slpeel_add_loop_guard.
34242 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34244 PR tree-optimization/80218
34245 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34246 Update block frequencies and counts.
34248 2017-03-28 Richard Biener <rguenther@suse.de>
34250 PR tree-optimization/78644
34251 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34252 of a simplification result we may not use it at all.
34254 2017-03-28 Richard Biener <rguenther@suse.de>
34257 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34258 without arguments, generate default definition of a SSA name.
34260 2017-03-28 Richard Biener <rguenther@suse.de>
34262 PR middle-end/80222
34263 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34264 TYPE_REF_CAN_ALIAS_ALL references.
34265 * fold-const.c (fold_indirect_ref_1): Likewise.
34267 2017-03-28 Martin Liska <mliska@suse.cz>
34270 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34271 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34273 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34274 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34276 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34277 (EXTRA_SPECS): Define.
34278 (SUBTARGET_EXTRA_SPECS): Likewise.
34279 (SUBTARGET_CPP_SPEC): Likewise.
34280 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34281 SUBTARGET_EXTRA_SPECS.
34282 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34284 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34286 * config/arc/simdext.md (vst64_insn): Update pattern.
34287 (vld32wh_insn): Likewise.
34288 (vld32wl_insn): Likewise.
34289 (vld64_insn): Likewise.
34290 (vld32_insn): Likewise.
34292 2017-03-28 Marek Polacek <polacek@redhat.com>
34295 * fold-const.c (fold_comparison): Use protected_set_expr_location
34296 instead of SET_EXPR_LOCATION.
34298 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34300 * tree.c (add_expr): Avoid name lookup warning.
34302 2017-03-27 Jeff Law <law@redhat.com>
34304 PR tree-optimization/80216
34305 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34306 function name. Limit recursion depth.
34307 (record_temporary_equivalences): Corresponding changes.
34309 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34311 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34314 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34317 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34319 * cfgcleanup.c (reg_note_cfa_p): New array.
34320 (insns_have_identical_cfa_notes): New function.
34321 (old_insns_match_p): Don't cross-jump in between /f
34322 and non-/f instructions. If both i1 and i2 are frame related,
34323 verify all CFA notes, their order and content.
34325 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34328 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34329 HImode and SImode with zero extend to DImode to one insn.
34330 (bswap<mode>2_extenddi): Likewise.
34331 (bswapsi2_extenddi): Likewise.
34332 (bswaphi2_extendsi): Likewise.
34333 (bswaphi2): Combine bswap HImode and SImode into one insn.
34334 Separate memory insns from swapping register.
34335 (bswapsi2): Likewise.
34336 (bswap<mode>2): Likewise.
34337 (bswaphi2_internal): Delete, no longer used.
34338 (bswapsi2_internal): Likewise.
34339 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34340 store, and gpr<-gpr swap insns.
34341 (bswap<mode>2_store): Likewise.
34342 (bswaphi2_reg): Register only splitter, combine with the splitter.
34343 (bswaphi2 splitter): Likewise.
34344 (bswapsi2_reg): Likewise.
34345 (bswapsi2 splitter): Likewise.
34346 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34347 the insns into load, store, and register/register insns.
34348 (bswapdi2_ldbrx): Likewise.
34349 (bswapdi2_load): Likewise.
34350 (bswapdi2_store): Likewise.
34351 (bswapdi2_reg): Likewise.
34353 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34355 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34356 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34358 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34361 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34363 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34364 special handling for target option conflicts between dform
34365 options (-mpower9-dform, -mpower9-dform-vector,
34366 -mpower9-dform-scalar) and -mno-direct-move.
34368 2017-03-27 Richard Biener <rguenther@suse.de>
34370 PR tree-optimization/80181
34371 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34373 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34375 * config/arc/predicates.md (move_double_src_operand): Replace the
34376 call to move_double_src_operand with a call to address_operand.
34378 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34380 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34381 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34382 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34384 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34386 * config/arc/predicates.md (long_immediate_loadstore_operand):
34387 Consider scaled addresses cases.
34389 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34391 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34392 restored when in interrupt.
34393 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34394 doesn't have delay slot.
34396 2017-03-27 Richard Biener <rguenther@suse.de>
34399 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34400 inlined thunk clones.
34402 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34405 * asan.c (instrument_derefs): Copy over last operand from
34406 original COMPONENT_REF to the new COMPONENT_REF with
34407 DECL_BIT_FIELD_REPRESENTATIVE.
34408 * ubsan.c (instrument_object_size): Likewise.
34410 2017-03-27 Richard Biener <rguenther@suse.de>
34412 PR tree-optimization/80170
34413 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34414 sure DR/SCEV didnt fold in constants we do not see when looking
34415 at the reference base alignment.
34417 2017-03-27 Richard Biener <rguenther@suse.de>
34419 PR middle-end/80171
34420 * gimple-fold.c (fold_ctor_reference): Properly guard against
34421 NULL return value from canonicalize_constructor_val.
34423 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34426 * config/i386/i386.c (ix86_expand_builtin)
34427 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34428 flags reg setting and flags reg using instructions.
34429 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34430 clobbering instructions to zero extend op2.
34432 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
34434 * doc/install.texi (Configuration) <--with-aix-soname>:
34435 Update link to AIX ld.
34437 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
34439 PR rtl-optimization/80160
34440 PR rtl-optimization/80159
34441 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
34442 reg_alternate_class into account.
34444 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
34447 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
34448 to consider in curr_insn_transform.
34450 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34452 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
34453 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
34454 and emit_mode_inner.
34456 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34458 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
34459 argument to the overloaded builtin variants. Use the new flag to
34460 deprecate certain builtin variants.
34461 * config/s390/s390-builtin-types.def: Add new builtin types.
34462 * config/s390/s390-builtins.h: Support new flags field for
34463 overloaded builtins.
34464 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
34465 (s390_macro_to_expand): Enable vector float data type.
34466 (s390_cpu_cpp_builtins_internal): Indicate support of the new
34467 builtins by incrementing the __VEC__ version number.
34468 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
34470 (s390_resolve_overloaded_builtin): Emit error messages depending
34471 on the builtin flags.
34472 * config/s390/s390.c (s390_expand_builtin): Support additional
34473 flags argument. Change error message to match the messages
34474 emitted in s390-c.c.
34475 * config/s390/s390.md: New UNSPEC_* constants.
34476 (op_type): Add new instruction types.
34477 * config/s390/vecintrin.h: Add new builtins and test data class
34479 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
34480 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
34481 (VEC_INEXACT, VEC_NOINEXACT): New constants.
34482 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
34483 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
34484 ("vec_mergel<mode>"): V_HW -> VEC_HW.
34486 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
34487 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
34488 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
34489 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
34491 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
34492 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
34493 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
34494 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
34496 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
34497 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
34498 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
34499 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
34500 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
34501 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
34502 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
34504 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
34505 ("vec_scatter_element<V_HW_4:mode>_DI")
34506 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
34507 ("vec_fpint<mode>", "vflls")
34508 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
34509 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
34510 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
34511 ("*vec_cmphe<mode>_cc"): ... these.
34513 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
34514 mode constant instead of magic value.
34516 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34518 * config/s390/s390.c (s390_expand_vec_compare): Support other
34519 vector floating point modes than just V2DF.
34520 (s390_expand_vcond): Likewise.
34521 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
34522 (s390_cannot_change_mode_class): Prevent mode changes between TF
34523 and V1TF in vector registers.
34524 * config/s390/s390.md (DF, SF): New mode attributes.
34525 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
34526 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
34527 SFmode support for VRs.
34528 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
34530 (VFT, VF_HW): New mode iterators.
34531 (vw, sdx): New mode attributes.
34532 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
34533 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
34534 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
34535 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
34536 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
34537 also the new vector floating point modes. Renaming to ...
34539 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
34540 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
34541 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
34542 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
34543 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
34544 ("vec_unordered<mode>"): ... these.
34546 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
34547 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
34548 ("*vec_extendv2df"): New insn definitions.
34550 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34552 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
34553 ("mulditi3_2", "*muldi3_sign"): New patterns.
34554 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
34555 rename the pattern definition.
34557 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34559 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
34561 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
34563 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34565 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
34566 instruction if possible.
34567 * config/s390/vector.md (vec_halfnumelts): New mode
34569 ("*vec_vllezlf<mode>"): New pattern.
34571 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34573 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
34574 ("popcountv4si2", "popcountv2di2"): Rename to ...
34575 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
34576 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
34578 ("popcount<mode>2_vxe"): New pattern.
34580 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34582 * common/config/s390/s390-common.c (processor_flags_table): Add
34584 * config.gcc: Add arch12.
34585 * config/s390/driver-native.c (s390_host_detect_local_cpu):
34586 Default to arch12 for unknown CPU model numbers.
34587 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
34588 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
34589 PROCESSOR_max sanity check.
34590 * config/s390/s390-opts.h (enum processor_type): Add
34592 * config/s390/s390.c (processor_table): Add arch12.
34593 (s390_expand_builtin): Add check for B_VXE flag.
34594 (s390_issue_rate): Add PROCESSOR_ARCH12.
34595 (s390_get_sched_attrmask): Likewise.
34596 (s390_get_unit_mask): Likewise.
34597 (s390_sched_score): Enable z13 scheduling for arch12.
34598 (s390_sched_reorder): Likewise.
34599 (s390_sched_variable_issue): Likewise.
34600 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
34602 (s390_tune_attr): Use z13 scheduling also for arch12.
34603 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
34604 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
34605 (TARGET_VXE_P): New macros.
34606 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
34607 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
34608 * config/s390/s390.opt: Add arch12 as processor_type.
34610 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34612 * config/s390/s390.md
34613 ("fixuns_truncdddi2", "fixuns_trunctddi2")
34614 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
34615 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
34617 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
34618 Rename expanders to ...
34620 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
34621 ("fixuns_truncdddi2_emu"): ... these.
34623 ("fixuns_trunc<mode>si2_emu"): New expander.
34625 ("*fixuns_truncdfdi2_z13"): Rename to ...
34626 ("*fixuns_truncdfdi2_vx"): ... this.
34628 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34630 * config/s390/2964.md: Remove the single element vector compare
34631 instructions which are no longer used.
34632 * config/s390/s390.c (s390_select_ccmode): Remove handling of
34634 (s390_canonicalize_comparison): Remove handling of DFmode
34636 (s390_expand_vec_compare_scalar): Remove function.
34637 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
34638 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
34640 ("*cmp<mode>_ccs"): Add wfcdb instruction.
34642 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34644 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
34646 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
34647 will anyway by matched by mov<mode>_64dfp.
34649 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34651 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
34652 vlef/vstef. Add missing operand to vleif.
34654 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34656 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
34657 pair for all vector types with 64 bit elements.
34658 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
34659 * config/s390/vector.md (V_HW_64): ... here.
34660 (V_128_NOSINGLE): New mode iterator.
34661 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
34662 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
34663 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
34664 ("*vec_load_pairv2di"): Change to ...
34665 ("*vec_load_pair<mode>"): ... this one.
34667 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34669 * config/s390/constraints.md: Add comments.
34670 (jKK): Reject element sizes > 8 bytes.
34671 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
34673 * config/s390/s390.md: Add the s_operand checks formerly in
34674 s390_split_ok_p to various splitters where they are still
34676 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
34677 for 128 bit vectors. Plus two splitters.
34679 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34681 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
34684 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34687 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
34688 error if the boundary argument is not constant.
34690 2017-03-24 Jakub Jelinek <jakub@redhat.com>
34692 PR rtl-optimization/80112
34693 * loop-doloop.c (doloop_condition_get): Don't check condition
34694 if cmp isn't SET with IF_THEN_ELSE src.
34696 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34698 PR tree-optimization/80158
34699 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
34700 replacing a candidate statement, also replace it for the
34701 candidate's alternate interpretation.
34702 (replace_rhs_if_not_dup): Likewise.
34703 (replace_one_candidate): Likewise.
34705 2017-03-24 Richard Biener <rguenther@suse.de>
34707 PR tree-optimization/80167
34708 * graphite-isl-ast-to-gimple.c
34709 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
34711 (translate_isl_ast_to_gimple::get_rename): Likewise.
34713 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
34715 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
34716 handling of certain combinations of target options, including the
34717 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
34718 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
34720 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34723 * config/arm/arm.md (*load_multiple): Add reload_completed to
34724 matching condition.
34726 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34727 Richard Biener <rguenth@suse.de>
34729 PR tree-optimization/79908
34730 PR tree-optimization/80136
34731 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34732 been cast away, gimplify_and_add suffices.
34734 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
34736 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
34738 2017-03-23 Richard Biener <rguenther@suse.de>
34740 PR tree-optimization/80032
34741 * gimplify.c (gimple_push_cleanup): Forced unconditional
34742 cleanups still have to go to the conditional_cleanups
34745 2017-03-22 Jakub Jelinek <jakub@redhat.com>
34747 PR tree-optimization/80072
34748 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
34750 (next_operand_entry_id): Change type to unsigned int.
34751 (sort_by_operand_rank): Make sure to return the right return value
34752 even if unsigned fields are bigger than INT_MAX.
34753 (struct oecount): Change cnt and id type to unsigned int.
34754 (oecount_hasher::equal): Formatting fix.
34755 (oecount_cmp): Make sure to return the right return value
34756 even if unsigned fields are bigger than INT_MAX.
34757 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
34760 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
34761 TREE_READONLY on result if writing it more than once.
34764 * doc/invoke.texi (-fsanitize=thread): Document that with
34765 -fnon-call-exceptions atomics are not able to throw
34769 * tsan.c: Include tree-eh.h.
34770 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34771 maybe_clean_or_replace_eh_stmt where needed.
34772 (instrument_memory_accesses): Add cfg_changed argument.
34773 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34774 if it returned true.
34775 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34777 PR rtl-optimization/63191
34778 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34779 wrapper function, moved the whole old content into ...
34780 (ix86_delegitimize_address_1): ... this. New inline function.
34781 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34782 true as last argument instead of ix86_delegitimize_address.
34784 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34786 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34787 cortexa57_branch_cost.
34789 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34791 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34793 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34796 * doc/md.texi (Constraints): Document wA constraint.
34797 * config/rs6000/constraints.md (wA): New.
34798 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34799 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34800 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34801 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34803 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34806 * gimplify.c (is_oacc_declared): New function.
34807 (oacc_default_clause): Use it to set default flags for acc declared
34808 variables inside parallel regions.
34809 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34810 declared variables.
34811 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34812 declare attribute to any decl as necessary.
34814 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34817 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34818 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34819 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34820 (arm_arch_lpae): This.
34821 * config/arm/arm.c (arm_arch7ve): Rename into ...
34822 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34823 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34826 2017-03-22 Martin Liska <mliska@suse.cz>
34829 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34830 error message instead of an ICE.
34832 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34834 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34836 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34838 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34840 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34843 2017-03-21 Martin Sebor <msebor@redhat.com>
34845 * doc/extend.texi: Use "cannot" instead of "can't."
34846 * doc/hostconfig.texi: Same.
34847 * doc/install.texi: Same.
34848 * doc/invoke.texi: Same.
34849 * doc/loop.texi: Same.
34850 * doc/md.texi: Same.
34851 * doc/objc.texi: Same.
34852 * doc/rtl.texi: Same.
34853 * doc/tm.texi: Same.
34854 * doc/tm.texi.in: Same.
34855 * doc/trouble.texi: Same.
34857 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34860 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34861 (collect_checksum_attributes): Set it.
34862 (die_checksum_ordered): Use it.
34864 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34866 PR tree-optimization/79908
34867 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34868 change: For a VA_ARG whose LHS has been cast away, use
34869 force_gimple_operand to construct the side effects.
34871 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34873 PR translation/80001
34874 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34875 more amenable to translation.
34876 (oacc_loop_auto_partitions): Likewise.
34878 2017-03-21 Marek Polacek <polacek@redhat.com>
34879 Martin Sebor <msebor@redhat.com>
34881 PR tree-optimization/80109
34882 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34883 on INTEGRAL_TYPE_P.
34885 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34886 Segher Boessenkool <segher@kernel.crashing.org>
34889 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34890 check reg_used_between_p between insn and one of succ or succ2
34891 depending on if succ is artificial insn not inserted into insn
34894 2017-03-21 Martin Liska <mliska@suse.cz>
34896 PR gcov-profile/80081
34897 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34898 * doc/gcc.texi: Include gcov-dump stuff.
34899 * doc/gcov-dump.texi: New file.
34901 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34903 PR rtl-optimization/79150
34904 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34905 conditional jump, if the jump is the last insn of the loop.
34907 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34908 Richard Biener <rguenth@suse.de>
34910 PR tree-optimization/79908
34911 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34912 been cast away, use force_gimple_operand to construct the side
34915 2017-03-21 Martin Liska <mliska@suse.cz>
34917 PR libfortran/79956
34918 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34921 2017-03-21 Brad Spengler <spender@grsecurity.net>
34924 * plugin.c (htab_hash_plugin): New function.
34925 (add_new_plugin): Use it and adjust.
34926 (parse_plugin_arg_opt): Adjust.
34927 (init_one_plugin): Likewise.
34929 2017-03-21 Richard Biener <rguenther@suse.de>
34931 PR tree-optimization/80032
34932 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34933 if set force the cleanup to happen unconditionally.
34934 (gimplify_target_expr): Push inserted clobbers with force_uncond
34935 to avoid them being removed by control-dependent DCE.
34937 2017-03-21 Richard Biener <rguenther@suse.de>
34939 PR tree-optimization/80122
34940 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34941 va_arg_pack_len when the inlined call stmt requires pack
34943 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34945 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34948 * tsan.c (instrument_builtin_call): If the memory model argument
34949 is not a constant, assume it is valid.
34952 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34955 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34957 PR rtl-optimization/79910
34958 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34959 if its dest is used by an insn before I2 (other than the combined
34960 insns themselves, which are properly handled already).
34962 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34965 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34967 * combine.c (record_used_regs): New static function.
34968 (try_combine): Handle situations where there is an additional
34969 instruction between I2 and I3 which needs to have a LOG_LINK
34973 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34975 * combine.c (try_combine): Delete redundant i1 test. Call
34976 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34978 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34981 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34982 alternatives 13/14.
34984 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34986 PR tree-optimization/80054
34987 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34988 the optimization if a PHI or any of its arguments is not dominated
34989 by the candidate's basis. Use gphi* rather than gimple* as
34991 (replace_profitable_candidates): Clean up a gimple* variable that
34992 should be a gphi* variable.
34994 2017-03-20 Martin Sebor <msebor@redhat.com>
34997 * doc/extend.texi (attribute constructor): Document present limitation.
34999 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35002 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35003 __POWER9_VECTOR__ #ifdef control, change template definition to
35004 use Power9-specific built-in function.
35005 (vec_any_eq): Likewise.
35006 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35007 to control outcomes from this test.
35008 (vector_ae_<mode>p): For VEC_F modes, likewise.
35010 2017-03-20 Ian Lance Taylor <iant@google.com>
35012 * config/i386/i386.c (ix86_function_regparm): Save an extra
35013 register for -fsplit-stack with DECL_STATIC_CHAIN.
35015 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35018 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35019 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35021 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35023 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35025 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35028 2017-03-20 Marek Polacek <polacek@redhat.com>
35031 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35033 2017-03-20 Richard Biener <rguenther@suse.de>
35035 PR tree-optimization/80113
35036 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35037 allocate extra SSA name for PHI def.
35038 (add_close_phis_to_outer_loops): Likewise.
35039 (add_close_phis_to_merge_points): Likewise.
35040 (copy_loop_close_phi_args): Likewise.
35041 (copy_cond_phi_nodes): Likewise.
35043 2017-03-20 Martin Liska <mliska@suse.cz>
35045 PR middle-end/79753
35046 * tree-chkp.c (chkp_build_returned_bound): Do not build
35047 returned bounds for a LHS that's not a BOUNDED_P type.
35049 2017-03-20 Martin Liska <mliska@suse.cz>
35053 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35054 COMPLEX_CST and VECTOR_CST.
35056 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35059 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35060 target operand. A new splitter adds the clobber statement in case
35061 the target operand is dead anyway.
35063 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35065 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35066 to age-old versions of binutils and glibc.
35068 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35070 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35072 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35074 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35076 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35078 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35079 requirement for binutils 2.13.
35081 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35083 * combine.c (try_combine): Delete redundant i1 test. Call
35084 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35086 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35088 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35089 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35091 <riscv64-*-elf>: Re-arrange section
35092 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35093 <riscv32-*-linux>: Likewise.
35094 <riscv64-*-elf>: Likewise
35095 <riscv64-*-linux>: Likewise.
35097 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35100 * aarch64.opt(verbose-cost-dump): Fix typo.
35102 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35105 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35106 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35108 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35110 * reload.c (find_reloads): When reloading a nonoffsettable address,
35111 use RELOAD_OTHER for it and its address reloads.
35113 PR rtl-optimization/79910
35114 * combine.c (record_used_regs): New static function.
35115 (try_combine): Handle situations where there is an additional
35116 instruction between I2 and I3 which needs to have a LOG_LINK
35119 2017-03-17 Jeff Law <law@redhat.com>
35121 PR tree-optimization/71437
35122 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35123 conditional in the hash table first.
35124 (vrp_dom_walker::before_dom_children): Extract condition from
35125 ASSERT_EXPR. Record condition, its inverion and any implied
35126 conditions as well.
35128 2017-03-17 Marek Polacek <polacek@redhat.com>
35129 Markus Trippelsdorf <markus@trippelsdorf.de>
35131 PR tree-optimization/80079
35132 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35135 2017-03-17 Richard Biener <rguenther@suse.de>
35137 PR middle-end/80075
35138 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35139 Properly verify the LHS before the RHS possibly claims to be
35141 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35144 2017-03-17 Martin Jambor <mjambor@suse.cz>
35146 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35147 (List of -O2 options): Likewise.
35148 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35151 2017-03-17 Tom de Vries <tom@codesourcery.com>
35153 * gcov-dump.c (print_usage): Print bug_report_url.
35155 2017-03-17 Richard Biener <rguenther@suse.de>
35157 PR middle-end/80050
35158 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35159 (parser::peek): Likewise.
35161 2017-03-17 Richard Biener <rguenther@suse.de>
35163 PR tree-optimization/80048
35164 * sese.c (free_sese_info): Properly release rename_map and
35165 copied_bb_map elements.
35167 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35169 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35170 Add linked-list forward and backlinks. Insert on
35171 construction, remove on destruction.
35172 (class pass_store_merging): Add m_stores_head field.
35173 (pass_store_merging::terminate_and_process_all_chains):
35174 Iterate over m_stores_head list.
35175 (pass_store_merging::terminate_all_aliasing_chains):
35177 (pass_store_merging::execute): Check for debug stmts first.
35178 Push new chains onto the m_stores_head stack.
35180 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35183 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35184 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35185 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35187 2017-03-16 Jeff Law <law@redhat.com>
35189 PR tree-optimization/71437
35190 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35191 member function. Implementation moved into after_dom_children
35192 member function and into the threader's thread_outgoing_edges
35194 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35195 some code into new thread_outgoing_edges.
35196 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35197 definition. Simplify marker handling (do it here). Assume we always
35198 have the available expression and the const/copies tables.
35199 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35201 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35202 * tree-vrp.c (equiv_stack): No longer file scoped.
35203 (vrp_dom_walker): New class.
35204 (vrp_dom_walker::before_dom_children): New member function.
35205 (vrp_dom_walker::after_dom_children): Likewise.
35206 (identify_jump_threads): Setup domwalker. Use it rather than
35207 walking edges in a random order by hand. Simplify setup/finalization.
35208 (finalize_jump_threads): Remove.
35209 (vrp_finalize): Do not call identify_jump_threads here.
35210 (execute_vrp): Do it here instead and call thread_through_all_blocks
35213 PR tree-optimization/71437
35214 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35216 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35218 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35219 (dom_opt_dom_walker::thread_across_edge): Remove
35220 handle_dominating_asserts argument. All callers changed.
35221 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35222 changes. Remove calls to lhs_of_dominating_assert. Other
35223 uses of handle_dominating_asserts turn into unconditional code
35224 (simplify_control_stmt_condition_1): Likewise.
35225 (simplify_control_stmt_condition): Likewise.
35226 (thread_through_normal_block, thread_across_edge): Likewise.
35227 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35228 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35229 object if it is not an SSA_NAME.
35230 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35231 before calling into the VRP specific simplifiers.
35232 (identify_jump_threads): Remove handle_dominating_asserts
35235 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35238 * tree-diagnostic.c (default_tree_printer): No longer static.
35239 * tree-diagnostic.h (default_tree_printer): New prototype.
35241 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35243 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35244 Change ins into fmov.
35246 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35248 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35249 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35250 Use h_con constraint for operand 1.
35251 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35252 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35254 2017-03-15 Jeff Law <law@redhat.com>
35256 PR tree-optimization/71437
35257 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35258 (record_temporary_equivalences): Use it.
35260 PR tree-optimization/71437
35261 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35262 tree-ssa-scopedtables.
35263 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35264 (record_conditions, record_cond, vuse_eq): Likewise.
35265 (record_edge_info): Adjust to API tweak of record_conditions.
35266 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35267 (record_temporary_equivalences, optimize_stmt): Likewise.
35268 (eliminate_redundant_computations): Likewise.
35269 (record_equivalences_from_stmt): Likewise.
35270 * tree-ssa-scopedtables.c: Include options.h and params.h.
35271 (vuse_eq): New function, moved from tree-ssa-dom.c
35272 (build_and_record_new_cond): Likewise.
35273 (record_conditions): Likewise. Accept vector of conditions rather
35274 than edge_equivalence structure for first argument.
35275 for the first argument.
35276 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35277 from tree-ssa-dom.c.
35278 (avail_exprs_stack::record_cond): Likewise.
35279 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35280 from tree-ssa-dom.c.
35281 (avail_exprs_stack): Add new member functions lookup_avail_expr
35283 (record_conditions): Declare.
35285 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35288 * lra-constraints.c (process_alt_operands): Increase reject for
35289 reloading an input/output operand.
35291 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35294 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35295 insns to convert from signed/unsigned char/short to IEEE 128-bit
35297 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35299 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35302 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35303 subreg of inner mode for values already in registers.
35305 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35307 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35308 iteration reg is used after the loop.
35310 2017-03-14 Martin Sebor <msebor@redhat.com>
35312 PR tree-optimization/79800
35313 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35314 precision in negative-positive range.
35315 (format_floating): Call non-const overload with adjusted precision.
35317 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35320 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35323 2017-03-14 Martin Sebor <msebor@redhat.com>
35325 PR middle-end/80020
35326 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35327 * builtins.def (aligned_alloc): Use it.
35330 * Makefile.in (GTFILES): Add calls.c.
35331 * calls.c: Include "gt-calls.h".
35333 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35335 PR rtl-optimization/79728
35336 * regs.h (struct target_regs): New field
35337 x_contains_allocatable_regs_of_mode.
35338 (contains_allocatable_regs_of_mode): New macro.
35339 * reginfo.c (init_reg_sets_1): Initialize it, and change
35340 contains_reg_of_mode so it includes global regs as well.
35341 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35342 rather than contains_regs_of_mode.
35344 2017-03-14 Martin Liska <mliska@suse.cz>
35346 * doc/invoke.texi: Document options that can't be combined with
35347 -fcheck-pointer-bounds.
35349 2017-03-14 Martin Liska <mliska@suse.cz>
35351 PR middle-end/79831
35352 * doc/invoke.texi (-Wchkp): Document the option.
35354 2017-03-14 Martin Liska <mliska@suse.cz>
35356 * Makefile.in: Install gcov-dump.
35358 2017-03-14 Martin Liska <mliska@suse.cz>
35360 * multiple_target.c (expand_target_clones): Bail out for
35361 an invalid attribute.
35363 2017-03-14 Richard Biener <rguenther@suse.de>
35365 * alias.c (struct alias_set_entry): Pack properly.
35366 * cfgloop.h (struct loop): Likewise.
35367 * cse.c (struct set): Likewise.
35368 * ipa-utils.c (struct searchc_env): Likewise.
35369 * loop-invariant.c (struct invariant): Likewise.
35370 * lra-remat.c (struct cand): Likewise.
35371 * recog.c (struct change_t): Likewise.
35372 * rtl.h (struct address_info): Likewise.
35373 * symbol-summary.h (function_summary): Likewise.
35374 * tree-loop-distribution.c (struct partition): Likewise.
35375 * tree-object-size.c (struct object_size_info): Likewise.
35376 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35377 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35378 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35379 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35380 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35381 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35382 (struct _stmt_vec_info): Likewise.
35384 2017-03-14 Martin Liska <mliska@suse.cz>
35387 * multiple_target.c (create_dispatcher_calls): Check that
35388 a target can create a function dispatcher.
35390 2017-03-14 Martin Liska <mliska@suse.cz>
35393 * multiple_target.c (expand_target_clones): Drop local.local
35394 flag for default implementation.
35396 2017-03-14 Richard Biener <rguenther@suse.de>
35398 PR tree-optimization/80030
35399 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35401 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35403 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35404 gcc_fallthrough() instead of __attribute__((fallthrough));
35406 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35408 * doc/gcc.texi: Remove "up" link to (DIR).
35409 * doc/gccint.texi: Ditto.
35411 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35413 * doc/install.texi (Specific) <avr>: Remove reference to
35416 2017-03-13 Jeff Law <law@redhat.com>
35418 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35419 attribute rather than comments.
35421 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35422 match_scratch operand is highest.
35424 2017-03-13 Martin Liska <mliska@suse.cz>
35426 PR middle-end/78339
35427 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35428 is a CHKP clone, use original declaration.
35430 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35432 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35433 (arc_conditional_register_usage): Use a different allocation order
35434 when optimizing for size.
35435 * common/config/arc/arc-common.c (arc_option_optimization_table):
35436 Section anchors default on when optimizing for size.
35438 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35440 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
35442 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35444 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
35445 * config/arc/arc.md (cpu_facility): Add cd variant.
35446 (*movqi_insn): Add code density variant.
35447 (*movhi_insn): Likewise.
35448 (*movqi_insn): Likewise.
35449 (*addsi3_mixed): Likewise.
35450 (subsi3_insn): Likewise.
35452 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35454 * config/arc/arc.md (movsi_cond_exec): Update constraint.
35456 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35458 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
35459 expressions with MINUS and UNARY ops.
35461 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35464 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
35466 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
35467 between vec_select and vector argument.
35468 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
35469 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
35470 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
35471 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
35472 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
35473 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
35475 2017-03-13 Richard Biener <rguenther@suse.de>
35478 * params.def (vect-max-peeling-for-alignment): Fix typo.
35480 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35482 * doc/install.texi (Specific) <mips-*-*>: Remove description of
35483 issue that only occurred with binutils below 2.18.
35485 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35487 * doc/install.texi (Specific) <cris-axis-elf>: No longer
35488 refer to binutils 2.11/2.12 minimum.
35490 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
35492 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
35493 ftp.kernel.org and simplify binutils requirement.
35495 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
35497 * doc/invoke.texi (Warning Options): Fix spelling of link-time
35499 (Optimize Options): Ditto. Also remove redundancy.
35501 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35503 PR translation/79848
35504 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
35506 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
35507 to G_ to avoid double translation.
35509 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35511 PR translation/79923
35512 * auto-profile.c (get_combined_location): Convert leading
35513 character of diagnostics to lower case and remove trailing period.
35514 (read_profile): Likewise for various diagnostics.
35515 * config/arm/arm.c (arm_option_override): Remove trailing period
35516 from various diagnostics.
35517 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
35518 (msp430_expand_delay_cycles): Likewise.
35520 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35523 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
35524 full command-line argument, rather than just "str".
35525 (aarch64_validate_march): Likewise.
35526 (aarch64_validate_mtune): Likewise.
35528 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
35530 PR rtl-optimization/78911
35531 * lra-assigns.c (must_not_spill_p): New function.
35532 (spill_for): Use it.
35534 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35536 PR tree-optimization/79981
35537 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
35538 ATOMIC_COMPARE_EXCHANGE ifn result.
35539 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
35540 IFN_ATOMIC_COMPARE_EXCHANGE.
35542 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35545 * opts.c (parse_sanitizer_options): Add missing question mark to
35546 "did you mean" message.
35548 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35550 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
35552 (VMULEUH_UNS): Likewise.
35553 (VMULOUB_UNS): Likewise.
35554 (VMULOUH_UNS): Likewise.
35555 * config/rs6000/rs6000.c (builtin_function_type): Remove
35556 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
35558 2017-03-10 David Malcolm <dmalcolm@redhat.com>
35561 * read-rtl-function.c (function_reader::read_rtx_operand): Update
35562 x with result of extra_parsing_for_operand_code_0.
35563 (function_reader::extra_parsing_for_operand_code_0): Convert
35564 return type from void to rtx, returning x. When reading
35565 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
35566 larger size containing struct block_symbol.
35568 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
35570 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
35571 -mfloat128-hardware without -m64.
35573 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
35576 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
35577 entries to the case statement that marks unsigned arguments to
35578 overloaded functions.
35580 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
35582 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
35583 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
35585 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
35588 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
35589 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
35591 2017-03-10 Martin Liska <mliska@suse.cz>
35595 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
35596 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
35599 2017-03-10 Olivier Hainque <hainque@adacore.com>
35601 * tree-switch-conversion (array_value_type): Start by resetting
35602 candidate type to it's main variant.
35604 2017-03-10 Jakub Jelinek <jakub@redhat.com>
35606 PR rtl-optimization/79909
35607 * combine.c (try_combine): Use simplify_replace_rtx on individual
35608 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
35609 of the whole CALL_INSN_FUNCTION_USAGE.
35611 PR tree-optimization/79972
35612 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
35613 get_range_info on SSA_NAMEs. Formatting fixes.
35615 2017-03-10 Richard Biener <rguenther@suse.de>
35616 Jakub Jelinek <jakub@redhat.com>
35618 PR tree-optimization/77975
35619 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
35620 edge to be constant.
35621 (get_val_for): For constant x return it. Formatting fix.
35622 (loop_niter_by_eval): Avoid pointless looping if the next iteration
35623 would use the same bases as the current one.
35625 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35627 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
35628 instead of vec_select for V1TImode.
35629 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
35631 (VSX_LE_128): Add V1TI to this mode iterator.
35632 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
35633 (*vsx_le_perm_store_<mode>): Likewise.
35634 (pre-reload splitter for VSX stores): Likewise.
35635 (post-reload splitter for VSX stores): Likewise.
35636 (*vsx_xxpermdi2_le_<mode>): Likewise.
35637 (*vsx_lxvd2x2_le_<mode>): Likewise.
35638 (*vsx_stxvd2x2_le_<mode>): Likewise.
35640 2017-03-09 Michael Eager <eager@eagercon.com>
35642 Correct failures with --enable-checking=yes,rtl.
35644 * config/microblaze/microblaze.c (microblaze_expand_shift):
35645 Replace GET_CODE test with CONST_INT_P and INTVAL test with
35646 test for const0_rtx.
35647 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
35648 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
35650 2017-03-09 Richard Biener <rguenther@suse.de>
35652 PR tree-optimization/79977
35653 * graphite-scop-detection.c (scop_detection::merge_sese):
35654 Handle the case of extra exits to blocks dominating the entry.
35656 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
35658 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
35661 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
35663 PR rtl-optimization/79949
35664 * lra-constraints.c (process_alt_operands): Check memory when
35665 trying to predict a cycle. Print about the overall increase.
35667 2017-03-09 Richard Biener <rguenther@suse.de>
35669 PR middle-end/79971
35670 * gimple-expr.c (useless_type_conversion_p): Preserve
35671 TYPE_SATURATING for fixed-point types.
35673 2017-03-09 Richard Biener <rguenther@suse.de>
35676 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
35677 alignment of BLKmode params.
35679 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35682 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
35683 (VALL_NO_V2Q): Likewise.
35685 * config/aarch64/aarch64-simd.md
35686 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
35688 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
35689 VALL_NO_V2Q mode iterator.
35690 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
35692 2017-03-09 Martin Liska <mliska@suse.cz>
35694 PR tree-optimization/79631
35695 * tree-chkp-opt.c (chkp_is_constant_addr): Call
35696 tree_int_cst_sign_bit just for INTEGER constants.
35698 2017-03-09 Martin Liska <mliska@suse.cz>
35702 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
35705 2017-03-09 Marek Polacek <polacek@redhat.com>
35708 * tree.c (inchash::add_expr): Handle TREE_VEC.
35710 2017-03-09 Martin Liska <mliska@suse.cz>
35713 (chkp_narrow_size_and_offset): New function.
35714 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
35715 (void chkp_parse_bit_field_ref): New function.
35716 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
35717 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
35719 2017-03-09 Martin Liska <mliska@suse.cz>
35722 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
35723 (chkp_find_bounds_1): Remove gcc_unreachable.
35725 2017-03-09 Jakub Jelinek <jakub@redhat.com>
35728 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
35729 BUILT_IN_SYNC*, determine the access type from the size suffix and
35730 always build a MEM_REF with that type. Handle forgotten
35731 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
35734 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
35735 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
35736 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
35737 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
35738 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
35739 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
35740 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
35741 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
35742 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
35743 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
35744 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
35745 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
35746 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
35747 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
35748 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
35749 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
35750 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
35751 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
35752 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
35753 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
35754 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
35755 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
35756 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
35757 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
35758 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
35759 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
35760 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
35761 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
35762 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
35763 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
35764 definitions outside of __OPTIMIZE__ guarded section.
35767 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
35768 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35769 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35770 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35773 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35775 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35776 ("vfenez<mode>"): Add missing constraints.
35778 2017-03-08 Martin Sebor <msebor@redhat.com>
35781 * config/nds32/nds32.c (nds32_option_override):
35782 Fix misspelled diagnostic.
35784 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35787 * gimplify.c (gimplify_omp_for): Replace index var in outer
35788 taskloop statement with an artificial variable and add
35789 OMP_CLAUSE_PRIVATE clause for it.
35791 2017-03-08 Richard Biener <rguenther@suse.de>
35793 PR tree-optimization/79955
35794 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35795 for accesses that are completely outside of the variable.
35797 2017-03-08 Andrew Haley <aph@redhat.com>
35799 PR tree-optimization/79943
35800 * tree-ssa-loop-split.c (compute_new_first_bound): When
35801 calculating the new upper bound, (END-BEG) should be added, not
35804 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35806 * config/avr/avr.md (setmemhi): Make sure match_dup
35807 operand number comes before match_scratch.
35809 2017-03-08 Richard Biener <rguenther@suse.de>
35811 PR tree-optimization/79920
35812 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35813 with ncopies == 1 to ...
35814 (vect_transform_slp_perm_load): ... here. Properly compute
35815 all element loads by iterating VF times over the group. Do
35816 not handle ncopies (computed in a broken way) in
35817 vect_create_mask_and_perm.
35819 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35822 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35823 is a uniform vector, use uniform_vector_p return value instead of
35824 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35826 2017-03-07 Marek Polacek <polacek@redhat.com>
35828 PR middle-end/79809
35829 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35830 (alloca_call_type): Likewise.
35832 2017-03-07 Martin Liska <mliska@suse.cz>
35834 * gcov.c (process_args): Put comment to correct location.
35836 2017-03-07 Martin Liska <mliska@suse.cz>
35838 PR middle-end/68270
35839 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35840 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35841 (chkp_narrow_bounds_for_field): Likewise.
35842 (chkp_parse_array_and_component_ref): Pass one more argument to
35845 2017-03-07 Richard Biener <rguenther@suse.de>
35847 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35850 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35852 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35853 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35855 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35858 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35859 to end of description.
35860 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35862 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35864 PR rtl-optimization/79901
35865 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35867 (*avx512f_<code><mode>3<mask_name>): ... this.
35868 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35869 iterator instead of VI8_AVX2_AVX512BW.
35871 PR rtl-optimization/79901
35872 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35873 min/max expander, expand it using expand_vec_cond_expr.
35876 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35879 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35882 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35883 to void * for PCH reasons.
35884 * dwarf2out.c (output_loc_operands, output_die): Cast
35885 v.val_vec.array to unsigned char *.
35887 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35890 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35893 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35895 PR rtl-optimization/79571
35896 * lra-constraints.c (process_alt_operands): Calculate static
35897 reject and subtract it from overall when only addresses will be
35900 2017-03-06 Julia Koval <julia.koval@intel.com>
35903 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35904 incoming stack boundary to 128 for 64-bit targets.
35906 2017-03-06 Richard Biener <rguenther@suse.de>
35908 PR tree-optimization/79894
35909 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35910 to NULL after folding it.
35912 2017-03-06 Richard Biener <rguenther@suse.de>
35914 PR tree-optimization/79824
35915 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35916 check disabling peeling for gaps.
35918 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35920 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35921 attributes): Document gettimeofday.
35923 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35925 * config/s390/s390.c (s390_option_override_internal): Set
35926 PARAM_MIN_VECT_LOOP_BOUND
35928 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35930 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35931 * config/s390/s390.md: Likewise.
35933 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35936 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35937 (<avx2_avx512>_perm<mode>): Rename to ...
35938 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35940 (<avx512>_perm<mode>_mask): Rename to ...
35941 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35943 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35944 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35945 instead of VI8F_256_512.
35946 (avx512f_perm<mode>): New define_expand.
35947 (avx512f_perm<mode>_mask): Likewise.
35948 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35949 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35951 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35953 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35954 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35956 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35958 2017-03-06 Martin Liska <mliska@suse.cz>
35961 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35962 when having a SSA NAME w/o VAR_DECL assigned to it.
35964 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35966 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35967 msa_dpsub_<su>_d): Fix MODE for vec_select.
35969 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35971 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35973 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35975 2017-03-06 Richard Biener <rguenther@suse.de>
35977 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35978 * plugin.c (register_plugin_info): Likewise.
35979 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35981 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35983 * config/i386/sse.md (sse_storehps, sse_storelps,
35984 avx_<castmode><avxsizesuffix>_<castmode>,
35985 avx512f_<castmode><avxsizesuffix>_<castmode>,
35986 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35987 in condition that at least one operand is not a MEM.
35989 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35991 PR middle-end/79805
35992 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35993 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35995 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35996 gimple_call_nothrow_p flag based on whether original builtin can throw.
35997 If it can, emit following stmts on the fallthrough edge.
35998 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35999 don't create new bb if inserting just debug stmts on the edge, try to
36000 insert them on the fallthru bb or just reset debug stmts.
36002 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36005 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36006 restore recog_data (including the operand rtxes inside it) around
36007 the call to get_insn_template.
36009 2017-03-03 Martin Sebor <msebor@redhat.com>
36011 PR tree-optimization/79699
36012 * context.c (context::~context): Free MPFR caches to avoid
36013 a memory leak on program exit.
36015 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36017 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36018 Use wide_int::ulow () instead of .elt (0).
36020 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36022 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36023 (*pushxf): Limit oF constraint to 32bit targets and add oC
36024 constraint for 64bit targets.
36025 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36026 (*pushdf): Change rmF constraint to rmC.
36028 2017-03-03 Martin Liska <mliska@suse.cz>
36030 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36031 Remove unused variable.
36033 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36036 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36037 is a memory operand, increase num_memory.
36038 (ix86_expand_args_builtin): Likewise.
36040 2017-03-03 Jan Hubicka <jh@suse.cz>
36043 * ipa-devirt.c (maybe_record_node): Properly handle
36044 __cxa_pure_virtual visibility.
36046 2017-03-03 Martin Liska <mliska@suse.cz>
36048 PR tree-optimization/79803
36049 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36051 (pass_loop_prefetch::execute): Disabled optimization if an
36052 assumption about L1 cache size is not met.
36054 2017-03-03 Martin Liska <mliska@suse.cz>
36056 PR rtl-optimization/79574
36057 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36058 (hash_scan_set): Likewise.
36059 (dump_hash_table): Likewise.
36060 (hoist_code): Likewise.
36062 2017-03-03 Richard Biener <rguenther@suse.de>
36064 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36065 in place of uhigh ().
36066 (fixed_convert_from_real): Likewise.
36068 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36071 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36073 2017-03-03 Richard Biener <rguenther@suse.de>
36075 PR middle-end/79818
36076 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36077 TYPE_OVERFLOW_UNDEFINED check.
36079 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36081 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36083 (vector_ae_<mode>_p): Likewise.
36084 (vector_nez_<mode>_p): Likewise.
36085 (vector_ne_v2di_p): Likewise.
36086 (vector_ae_v2di_p): Likewise.
36087 (vector_ne_<mode>_p): Likewise.
36088 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36090 (vsx_tsqrt<mode>2_fe): Likewise.
36092 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36095 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36097 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36099 PR rtl-optimization/79780
36100 * cprop.c (one_cprop_pass): When second and further conditional trap
36101 in a single basic block is turned into an unconditional trap, turn it
36102 into a deleted note to avoid RTL verification failures.
36104 2017-03-02 Richard Biener <rguenther@suse.de>
36106 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36108 2017-03-02 Richard Biener <rguenther@suse.de>
36110 PR tree-optimization/79345
36112 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36113 param and abort the walk, returning -1 if it is hit.
36114 (walk_aliased_vdefs): Take a limit param and pass it on.
36115 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36116 defaulting to 0 and return a signed int.
36117 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36118 (check_defs): New helper.
36119 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36120 about uninitialized memory.
36121 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36122 bogus uninitialized warning.
36123 (fixed_convert_from_real): Likewise.
36125 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36127 PR tree-optimization/66768
36128 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36129 iv_use if base object can't be determined.
36131 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36133 PR tree-optimization/79345
36134 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36135 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36136 (get_pattern_stats): Initialize it.
36137 * genemit.c (gen_expand): Verify match_scratch numbers come after
36138 match_operand/match_dup numbers.
36139 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36140 match_scratch numbers.
36141 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36143 * config/s390/s390.md (trunctdsd2): Likewise.
36145 2017-03-02 Richard Biener <rguenther@suse.de>
36147 * wide-int.h (wide_int_storage::operator=): Implement in terms
36150 2017-03-02 Richard Biener <rguenther@suse.de>
36152 PR tree-optimization/79777
36153 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36154 the to insert expression to sth existing.
36156 2017-03-01 Martin Sebor <msebor@redhat.com>
36158 PR middle-end/79692
36159 * gimple-ssa-sprintf.c
36160 (directive::known_width_and_precision): New function.
36161 (format_integer): Use it.
36162 (get_mpfr_format_length): Consider the full range of precision
36163 when computing %g output with the # flag. Set the likely byte
36164 count to 3 rather than 1 when precision is indeterminate.
36165 (format_floating): Correct the lower bound of precision.
36167 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36169 * doc/invoke.texi: Document default code model for 64-bit Linux.
36171 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36174 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36175 udiv rather than div since input pattern is unsigned.
36177 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36179 * config/i386/i386.c (print_reg): Warn for values of
36180 unsupported size in integer register.
36182 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36185 * config/rs6000/predicates.md (current_file_function_operand): Do
36186 not allow self calls to be local if the function is replaceable.
36188 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36191 * config/rs6000/altivec.h (vec_ctz and others): Change the
36192 preprocessor macro that controls conditional compilation from
36193 _ARCH_PWR9 to __POWER9_VECTOR__.
36194 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36195 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36196 control (instead of _ARCH_PWR9 control) so that template
36197 definition uses power9-specific function.
36198 (vec_any_eq): Likewise.
36199 (vec_all_ne): Change macro definition to use a power9-specific
36200 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36201 _ARCH_PWR9 control).
36202 (vec_any_eq) Likewise.
36203 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36204 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36205 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36206 support for xvcmpnedp instruction.
36207 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36208 macro expansion so that Power9 implementation of vec_all_ne does
36209 not use the AltiVec predicate framework.
36210 (VCMPNEH_P): Likewise.
36211 (VCMPNEW_P): Likewise.
36212 (VCMPNED_P): Likewise.
36213 (VCMPNEFP_P): Likewise.
36214 (VCMPNEDP_P): Likewise.
36215 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36216 implementation of vec_any_eq to not use AltiVec predicate
36218 (VCMPAEH_P): Likewise.
36219 (VCMPAEW_P): Likewise.
36220 (VCMPAED_P): Likewise.
36221 (VCMPAEFP_P): Likewise.
36222 (VCMPAEDP_P): Likewise.
36223 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36224 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36225 not use the AltiVec predicate framework.
36226 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36227 of vec_any_eq to not use AltiVec predicate framework.
36228 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36229 support for predefined __POWER9_VECTOR__ macro to indicate that
36230 Power9 instruction selection is enabled.
36231 (altivec_overloaded_builtins): Remove extraneous
36232 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36233 function argument types RS6000_BTI_bool_V16QI and
36234 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36235 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36236 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36237 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36238 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36239 Power9 for implementations of vec_cmpne. Change the signature for
36240 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36241 (representing vec_all_ne) to remove the previously described first
36242 argument of type RS6000_BTI_INTSI, as this was an artifact of
36243 reliance on the AltiVec predicate framework, which is no longer
36244 used in the implementation of these functions. Add
36245 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36246 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36247 since, unlike the AltiVec predicate framework implementation, we
36248 do not share function descriptors between vec_alle and vec_anyeq.
36249 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36250 set of modes that receive special treatment even when
36251 TARGET_P9_VECTOR is true. The special treatment emits code that
36252 does not depend on Power9 instructions.
36253 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36254 define_expand to not rely on AltiVec predicate framework.
36255 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36257 (vector_ne_v2di_p): Change this define_expand to not rely on
36258 AltiVec predicate framework.
36259 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36261 (vector_ne_<mode>_p): Change this define_expand to not rely on
36262 AltiVec predicate framework.
36263 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36265 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36266 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36267 define_insn pattern.
36268 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36269 define_insn pattern because the xvcmpne<VSs>. instruction is not
36271 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36272 instruction is not supported.
36274 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36276 * config/nvptx/nvptx.c: Include intl.h.
36278 2017-03-01 Martin Jambor <mjambor@suse.cz>
36281 * ipa-prop.h (ipa_bits): Removed field known.
36282 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36283 to pointers. Adjusted their comments to warn about their sharing.
36284 (ipcp_transformation_summary): Change bits to a vector of pointers.
36285 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36286 (ipa_get_ipa_bits_for_value): Declare.
36287 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36288 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36289 (ipa_bits_hash_table): Likewise.
36290 (ipa_vr_ggc_hash_traits): Likewise.
36291 (ipa_vr_hash_table): Likewise.
36292 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36293 being pointers and vr_known being removed.
36294 (ipa_set_jf_unknown): Likewise.
36295 (ipa_get_ipa_bits_for_value): New function.
36296 (ipa_set_jfunc_bits): Likewise.
36297 (ipa_get_value_range): New overloaded functions.
36298 (ipa_set_jfunc_vr): Likewise.
36299 (ipa_compute_jump_functions_for_edge): Use the above functions to
36300 construct bits and vr parts of jump functions.
36301 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36302 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36304 (ipcp_grow_transformations_if_necessary): Also allocate
36305 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36307 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36308 them. Fix too long lines.
36309 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36310 vr_known being removed.
36311 (ipa_read_jump_function): Use new setter functions to construct bits
36312 and vr parts of jump functions or set them to NULL.
36313 (write_ipcp_transformation_info): Adjust for bits being pointers.
36314 (read_ipcp_transformation_info): Likewise.
36315 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36317 Include gt-ipa-prop.h.
36318 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36320 (ipcp_store_bits_results): Likewise.
36321 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36322 Do not write to existing jump functions but use a temporary instead.
36324 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36327 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36328 attempt to use its first operand as BIT_FIELD_REF base.
36330 2017-03-01 Richard Biener <rguenther@suse.de>
36332 PR middle-end/79721
36333 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36334 interpolating formula in wrapping arithmetic.
36335 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36337 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36339 PR tree-optimization/79734
36340 * tree-vect-generic.c (expand_vector_condition): Optimize
36341 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36342 Handle VEC_COND_EXPR where comparison has different inner width from
36343 type's inner width.
36345 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36347 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36348 markup, and similar issues. Remove @opindex entries for things
36349 that aren't options. Add missing -mmpy-option entries.
36351 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36353 PR tree-optimization/79737
36354 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36355 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36356 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36357 instead of byte_size. Formatting fix.
36358 (shift_bytes_in_array_right): Formatting fix.
36360 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36363 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36364 condition on optimize for the leaf function test.
36366 2017-02-28 Martin Liska <mliska@suse.cz>
36369 * read-rtl-function.c (function_reader::handle_unknown_directive):
36370 Bail out when one uses -flto.
36372 2017-02-28 Martin Liska <mliska@suse.cz>
36374 * common.opt: Replace space with tabular for options of <number>
36376 * config/i386/i386.opt: Show <number> value for
36377 -mlarge-data-threshold.
36378 * opts.c (print_filtered_help): Do not display number in hexadecimal
36381 2017-02-28 Martin Liska <mliska@suse.cz>
36383 * common.opt: Fix --help=option -Q for options which are of
36386 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36388 * config/i386/i386.c (print_reg): Error out for values
36389 of 8-bit size in invalid integer register.
36391 2017-02-28 Martin Sebor <msebor@redhat.com>
36393 PR tree-optimization/79691
36394 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36396 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36399 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36400 gcc_unreachable with output_operand_lossage.
36402 2017-02-28 Richard Biener <rguenther@suse.de>
36404 PR tree-optimization/79740
36405 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36407 (visit_nary_op): Insert the nary into the hashtable if we
36408 pattern-matched sth.
36409 * tree-ssa-pre.c (eliminate_insert): Robustify.
36411 2017-02-28 Richard Biener <rguenther@suse.de>
36413 PR middle-end/79731
36414 * fold-const.c (decode_field_reference): Reject out-of-bound
36417 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36419 * config/i386/i386.c: Include intl.h.
36420 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36421 instead of just cond ? "..." : "...".
36422 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36423 * coverage.c (read_counts_file): Likewise.
36424 * omp-offload.c: Include intl.h.
36425 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36426 of just cond ? "..." : "...".
36427 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36428 of just cond ? "..." : "...".
36430 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36433 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
36435 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
36436 'tune for' CPU name.
36437 * config/arm/arm-cpu-data.h: Regenerated.
36439 2017-02-28 Richard Biener <rguenther@suse.de>
36441 PR tree-optimization/79732
36442 * tree-inline.c (expand_call_inline): Do not shadow var.
36444 2017-02-28 Richard Biener <rguenther@suse.de>
36446 PR tree-optimization/79723
36447 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
36448 address-space properly.
36450 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
36452 * doc/optinfo.texi (Optimization groups): Fix option used for
36454 * doc/invoke.texi (-fopt-info): Document "omp".
36455 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
36456 (OPTGROUP_ALL): Add OPTGROUP_OMP.
36457 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
36458 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
36459 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
36461 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
36463 * dumpfile.c (optgroup_options): Instead of "openmp", associate
36464 OPTGROUP_OMP with "omp".
36466 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
36469 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
36470 for arithmetic shift of unsigned V2DI.
36472 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
36474 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
36475 arc/linux.h headers.
36476 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
36477 (LINK_SPEC): Likewise.
36478 (ARC_TLS_EXTRA_START_SPEC): Likewise.
36479 (EXTRA_SPECS): Likewise.
36480 (STARTFILE_SPEC): Likewise.
36481 (ENDFILE_SPEC): Likewise.
36482 (LIB_SPEC): Likewise.
36483 (TARGET_SDATA_DEFAULT): Likewise.
36484 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
36485 (MULTILIB_DEFAULTS): Likewise.
36486 (DWARF2_UNWIND_INFO): Likewise.
36487 * config/arc/big.h: New file.
36488 * config/arc/elf.h: Likewise.
36489 * config/arc/linux.h: Likewise.
36490 * config/arc/t-uClibc: Remove.
36492 2017-02-27 Bin Cheng <bin.cheng@arm.com>
36494 PR tree-optimization/77536
36495 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
36496 (tree_transform_and_unroll_loop): Use above function to compute the
36497 estimated niter of unrolled loop and use it when scaling profile.
36498 Also use count info rather than frequency if it's non-zero.
36499 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
36500 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
36501 (vect_transform_loop): Call above function.
36503 2017-02-27 Richard Biener <rguenther@suse.de>
36505 PR tree-optimization/45397
36506 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
36507 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
36508 (visit_nary_op): Add pattern matching for CSEing sign-changed
36509 or truncated operations with wider ones.
36511 2017-02-27 Richard Biener <rguenther@suse.de>
36513 PR tree-optimization/79690
36514 * tree-vect-stmts.c (vectorizable_store): Use vector type
36515 built from the DR with address-space.
36517 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
36519 * doc/invoke.texi (Optimize Options): Refine the description
36520 of asan-use-after-return.
36522 2017-02-25 Alan Modra <amodra@gmail.com>
36524 PR rtl-optimization/79584
36525 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
36526 base, not ad->base_term, the reg within base. Remove assertion
36527 that ad->base == ad->base_term. Replace gen_int_mode using
36528 bogus mode with const0_rtx.
36530 2017-02-25 Jakub Jelinek <jakub@redhat.com>
36532 PR middle-end/79396
36533 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
36534 FMA_EXPR like tcc_binary or tcc_unary.
36536 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
36539 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
36541 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
36542 (output_loc_operands): Handle DW_OP_call_ref and
36543 DW_OP_GNU_variable_value.
36544 (struct variable_value_struct): New type.
36545 (struct variable_value_hasher): Likewise.
36546 (variable_value_hash): New variable.
36547 (string_types): Remove.
36548 (copy_loc_descr): New function.
36549 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
36550 (prepend_loc_descr_to_each): New function.
36551 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
36552 instead of add_loc_descr_to_each if the first argument is single
36553 location list and the second has multiple.
36554 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
36555 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
36556 when looking for variable value which doesn't have other location info.
36557 (loc_list_from_tree): Formatting fix.
36558 (gen_array_type_die): Simplify DW_AT_string_length handling.
36559 (adjust_string_types): Remove.
36560 (gen_subprogram_die): Don't call adjust_string_types nor test/set
36561 string_types. Call resolve_variable_values.
36562 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
36563 (resolve_addr_in_expr): Likewise. Add A argument.
36564 (copy_deref_exprloc): Remove deref argument. Adjust for the
36565 original expression being DW_OP_GNU_variable_value with optionally
36566 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
36567 optionally after it.
36568 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
36569 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
36570 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
36571 (variable_value_hasher::hash, variable_value_hasher::equal): New
36573 (resolve_variable_value_in_expr, resolve_variable_value,
36574 resolve_variable_values, note_variable_value_in_expr,
36575 note_variable_value): New functions.
36576 (dwarf2out_early_finish): Call note_variable_value on all toplevel
36579 2017-02-24 Jakub Jelinek <jakub@redhat.com>
36582 * opts.h (handle_generated_option): Add GENERATED_P argument.
36583 * opts-common.c (handle_option): Adjust function comment.
36584 (handle_generated_option): Add GENERATED_P argument, pass it to
36586 (control_warning_option): Pass false to handle_generated_option
36588 * opts.c (maybe_default_option): Pass true to handle_generated_option
36590 * optc-gen.awk: Likewise.
36592 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36594 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
36595 a REG, look at the REG it is a SUBREG of.
36596 (splitter for cmpeqsi_t): Ditto.
36598 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36600 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
36601 the special USEs with the pattern of the insn, not the insn itself.
36603 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
36606 * doc/invoke.texi: Document -mload-store-pairs.
36608 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36609 Sandra Loosemore <sandra@codesourcery.com>
36611 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
36612 argument isn't a CONST_INT.
36613 (nios2_alternate_compare_const): Assert op is a CONST_INT.
36614 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
36615 (nios2_validate_compare): Bypass alternate compare logic if *op2
36616 is not a CONST_INT.
36617 (ldstwm_operation_p): Return false if first_base is not a REG or
36618 if first_offset is not a CONST_INT.
36620 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36622 * config/cris/cris.md: Use correct operand in a define_peephole2.
36624 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36626 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
36628 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
36630 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
36631 this_insn if it is an INSN or JUMP_INSN.
36632 (force_offsettable): Look at base, not at addr.
36633 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
36634 on things that aren't necessarily CONST_INTs.
36636 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
36638 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
36639 -mfpmath=sse is the default also for x86-32 targets with SSE2
36640 instruction set when @option{-ffast-math} is enabled
36642 2017-02-24 Jeff Law <law@redhat.com>
36644 PR rtl-optimizatoin/79286
36645 * ira.c (update_equiv_regs): Drop may_trap_p exception to
36648 2017-02-24 Richard Biener <rguenther@suse.de>
36650 PR tree-optimization/79389
36651 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
36654 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
36656 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
36657 function comment to reflect reality.
36658 (loop_exits_before_overflow): Fix typo in function description.
36660 2017-02-24 Richard Biener <rguenther@suse.de>
36662 PR tree-optimization/79389
36663 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
36664 properly that a threading opportunity exists. Detect conditional
36665 copy/constant propagation opportunities.
36667 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
36669 * config/visium/visium.md (type): Add trap.
36670 (b): New mode attribute.
36671 (*btst): Rename into...
36672 (*btst<mode>): ...this and adjust.
36673 (*cbranchsi4_btst_insn): Rename into...
36674 (*cbranch<mode>4_btst_insn): ...this and adjust.
36675 (trap): New define_insn.
36677 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36679 PR tree-optimization/79389
36680 * ifcvt.c (struct noce_if_info): Add rev_cond field.
36681 (noce_reversed_cond_code): New function.
36682 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
36683 reversed_comparison_code. Formatting fix.
36684 (noce_try_store_flag): Test rev_cond != NULL in addition to
36685 reversed_comparison_code.
36686 (noce_try_store_flag_constants): Likewise.
36687 (noce_try_store_flag_mask): Likewise.
36688 (noce_try_addcc): Use rev_cond if non-NULL instead of
36689 reversed_comparison_code.
36690 (noce_try_cmove_arith): Likewise. Formatting fixes.
36691 (noce_try_minmax, noce_try_abs): Clear rev_cond.
36692 (noce_find_if_block): Initialize rev_cond.
36693 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
36694 instead of false as last argument never attempt to reverse it
36697 2017-02-23 Bin Cheng <bin.cheng@arm.com>
36699 PR tree-optimization/79663
36700 * tree-predcom.c (combine_chains): Process refs in reverse order
36701 only for ZERO length chains, and add explaining comment.
36703 2017-02-23 Jeff Law <law@redhat.com>
36705 PR tree-optimization/79578
36706 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
36707 in call to operand_equal_p.
36709 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
36712 * config/i386/cpuid.h: Fix another undefined behavior.
36714 2017-02-23 Richard Biener <rguenther@suse.de>
36716 PR tree-optimization/79683
36717 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
36718 vector types for data-refs.
36720 2017-02-23 Martin Liska <mliska@suse.cz>
36722 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
36724 2017-02-23 Jakub Jelinek <jakub@redhat.com>
36726 PR middle-end/79665
36727 * internal-fn.c (get_range_pos_neg): Moved to ...
36728 * tree.c (get_range_pos_neg): ... here. No longer static.
36729 * tree.h (get_range_pos_neg): New prototype.
36730 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
36731 are known to be in between 0 and signed maximum inclusive, try to
36732 expand both unsigned and signed divmod and use the cheaper one from
36735 2017-02-22 Jeff Law <law@redhat.com>
36737 PR tree-optimization/79578
36738 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
36739 to compare base operands.
36741 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
36744 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
36745 gpc_reg_operand instead of fpr_reg_operand.
36747 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
36749 * config/mips/mips.c (mips_return_in_memory): Force FP
36750 vector types to be returned in memory for o32 ABI.
36752 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36754 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
36755 instead of DW_TAG_member for static data member declarations and don't
36756 set no_linkage_name for static inline data members.
36757 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
36760 2017-02-22 Martin Liska <mliska@suse.cz>
36762 * doc/invoke.texi: Replace inequality signs with square brackets
36765 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
36768 * lra-constraints.c (simplify_operand_subreg): Handle
36769 WORD_REGISTER_OPERATIONS targets.
36771 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36774 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36775 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36776 elimination by swapping fld*.
36778 2017-02-22 Richard Biener <rguenther@suse.de>
36780 PR tree-optimization/79673
36781 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36782 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36783 irrelevant address-space qualifiers and avoiding a
36784 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36786 2017-02-22 Richard Biener <rguenther@suse.de>
36788 PR tree-optimization/79666
36789 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36790 to not symbolically negate if that may introduce undefined
36793 2017-02-22 Martin Liska <mliska@suse.cz>
36796 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36797 * data-streamer-out.c (streamer_write_gcov_count_stream):
36799 * value-prof.c (stream_out_histogram_value): Make assert more
36800 precise based on type of counter.
36802 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36805 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36806 Use nonimmediate_operand instead of memory_operand for operand 1.
36807 (float-extend standard_x87sse_constant_load splitter): Ditto.
36809 2017-02-21 Jeff Law <law@redhat.com>
36811 PR tree-optimization/79621
36812 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36813 blocks with edges to themselves.
36815 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36818 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36819 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36820 Use gimple_call_builtin_p.
36823 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36824 on temporarily removed DEBUG_INSNs.
36826 PR tree-optimization/79649
36827 * tree-loop-distribution.c (classify_partition): Give up on
36828 non-generic address space loads/stores.
36830 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36832 * doc/loop.texi (Loop manipulation): Remove nonexistent
36833 tree_ssa_loop_version from the documentation.
36834 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36836 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36839 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36840 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36841 * config/rs6000/rs6000.c: Include except.h.
36842 (rs6000_expand_split_stack_prologue): Call
36843 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36845 2017-02-21 Martin Jambor <mjambor@suse.cz>
36848 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36849 have been analyzed.
36851 2017-02-21 Martin Jambor <mjambor@suse.cz>
36853 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36854 for backward compatibility only.
36855 * doc/invoke.texi (Option Summary): Remove all references to
36856 -fipa-cp-alignment.
36858 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36862 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36864 * lra-constraints.c (curr_insn_transform): Handle
36865 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36867 2017-02-21 Martin Liska <mliska@suse.cz>
36869 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36871 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36873 PR translation/79638
36874 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36876 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36879 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36880 (arm_function_ok_for_sibcall): Return false for an indirect call by
36881 descriptor if all the argument registers are used.
36882 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36883 alignment of the function.
36885 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36887 PR tree-optimization/61441
36888 * simplify-rtx.c (simplify_const_unary_operation): For
36889 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36890 the sNaN unmodified.
36892 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36894 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36895 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36896 instead of SYSTEM_HEADER_DIR.
36898 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36899 Martin Liška <mliska@suse.cz>
36901 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36902 Fix typos and grammar, use active voice, and clarify.
36904 2017-02-20 Marek Polacek <polacek@redhat.com>
36906 PR middle-end/79537
36907 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36910 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36912 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36915 * config/i386/i386.c (ix86_expand_builtin): Handle
36916 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36917 ix86_builtins_isa[fcode].isa as a requirement of those
36918 flags and any other flag in the bitmask.
36919 (ix86_init_mmx_sse_builtins): Use 0 instead of
36920 ~OPTION_MASK_ISA_64BIT as mask.
36921 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36922 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36923 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36924 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36926 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36929 * lra-constraints.c (split_reg): Check requested split mode
36930 is supported by the register.
36932 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36934 * lra-constraints.c (simplify_operand_subreg): Remove early
36937 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36940 * lra-constraints.c (curr_insn_transform): Tighten condition
36941 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36943 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36946 * lra-constraints.c (curr_insn_transform): Handle
36947 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36949 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36952 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36954 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36956 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36959 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36962 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36964 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36967 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36969 * final.c (last_columnnum, override_columnnum): New variables.
36970 (final_start_function): Set last_columnnum, pass it to begin_prologue
36971 hook and pass 0 to dwarf2out_begin_prologue.
36972 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36973 to source_line debug hook.
36974 (notice_source_line): Compute last_columnnum and for debug_column_info
36975 return true on column changes.
36976 * debug.h (struct gcc_debug_hooks): Add column argument to
36977 source_line and begin_prologue hooks.
36978 (debug_nothing_int_charstar_int_bool): Remove prototype.
36979 (debug_nothing_int_int_charstar,
36980 debug_nothing_int_int_charstar_int_bool): New prototypes.
36981 (dwarf2out_begin_prologue): Add column argument.
36982 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36983 begin_prologue hooks.
36984 (debug_nothing_int_charstar_int_bool): Remove.
36985 (debug_nothing_int_int_charstar,
36986 debug_nothing_int_int_charstar_int_bool): New functions.
36987 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36988 through to dwarf2out_source_line.
36989 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36990 (dwarf2out_source_line): Add column argument, emit it if requested.
36991 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36993 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36994 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36995 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36996 through to dwarf2out_begin_prologue.
36997 (vmsdbgout_source_line): Add column argument, pass it through to
36998 dwarf2out_source_line.
36999 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37000 dbxout_source_line caller.
37001 (dbxout_source_line): Add column argument.
37003 * common.opt (gno-column-info, gcolumn-info): New options.
37004 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37005 (check_die): Also test for multiple DW_AT_decl_column attributes.
37006 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37007 DW_AT_decl_column if requested.
37008 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37010 (gen_variable_die): Likewise.
37011 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37012 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37015 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37016 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37017 (ix86_handle_option): Handle OPT_m3dnowa.
37018 * doc/invoke.texi (-m3dnowa): Document.
37019 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37020 -m3dnowa instead of -m3dnow -march=athlon.
37023 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37024 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37026 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37029 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37030 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37031 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37032 generator for vsx_xxpermdi_<mode>_be.
37033 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37034 force big-endian semantics.
37035 (vsx_xxpermdi_<mode>_be): New define_expand with same
37036 implementation as previous version of vsx_xxpermdi_<mode>.
37038 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37040 PR tree-optimization/79327
37041 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37042 variable, its initialization and use.
37044 2017-02-17 Julia Koval <julia.koval@intel.com>
37046 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37047 (OPTION_MASK_ISA_PKU_UNSET): New.
37048 (ix86_handle_option): Handle -mrdpid.
37049 * config/i386/cpuid.h (bit_RDPID): New.
37050 * config/i386/driver-i386.c (host_detect_local_cpu):
37051 Detect RDPID feature.
37052 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37053 * config/i386/i386-c.c (ix86_target_macros_internal):
37055 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37056 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37057 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37058 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37059 * config/i386/i386.md (define_insn "rdpid"): New.
37060 * config/i386/i386.opt Add -mrdpid.
37061 * config/i386/immintrin.h (_rdpid_u32): New.
37063 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37065 PR rtl-optimization/79541
37066 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37067 instead of transforming it into USE.
37069 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37071 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37072 If HONOR_SNANS (SFmode) force the input to a register.
37073 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37074 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37075 an frsp or similar insn.
37077 2017-02-17 Martin Liska <mliska@suse.cz>
37079 PR rtl-optimization/79577
37080 * params.def (selsched-max-sched-times): Increase minimum to 1.
37082 2017-02-17 Martin Liska <mliska@suse.cz>
37084 PR rtl-optimization/79574
37085 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37087 2017-02-17 Martin Liska <mliska@suse.cz>
37089 PR tree-optimization/79529
37090 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37091 ssa_defined_default_def_p to handle cases which are implicitly
37093 * tree-ssa.c (ssa_defined_default_def_p): New function.
37094 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37095 which are implicitly defined.
37096 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37098 2017-02-17 Richard Biener <rguenther@suse.de>
37100 PR middle-end/79576
37101 * params.def (max-ssa-name-query-depth): Limit to 10.
37103 2017-02-17 Richard Biener <rguenther@suse.de>
37105 PR tree-optimization/79552
37106 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37109 2017-02-17 Richard Biener <rguenther@suse.de>
37112 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37114 2017-02-17 Marek Polacek <polacek@redhat.com>
37116 PR middle-end/79536
37117 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37118 (fold_negate_expr): New wrapper.
37120 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37122 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37123 Correct terminology and de-emphasize pre-standard behavior.
37125 2017-02-16 Alan Modra <amodra@gmail.com>
37127 PR rtl-optimization/79286
37128 * ira.c (def_dominates_uses): New function.
37129 (update_equiv_regs): Don't create an equivalence for insns that
37130 may trap where the register def does not dominate the use.
37132 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37134 PR rtl-optimization/78127
37135 * lra.c (lra): Call lra_eliminate before finish the loop after
37138 2017-02-16 Richard Biener <rguenther@suse.de>
37140 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37142 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37143 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37144 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37145 (isl_val_int_from_wi): New function.
37146 (extract_affine_gmp): Rename to ...
37147 (extract_affine_wi): ... this, take a widest_int.
37148 (extract_affine_int): Just wrap extract_affine_wi.
37149 (add_param_constraints): Use isl_val_int_from_wi.
37150 (add_loop_constraints): Likewise, and extract_affine_wi.
37152 2017-02-15 Jeff Law <law@redhat.com>
37154 PR middle-end/79521
37155 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37156 ira_init_register_move_cost_if_necessary.
37158 2017-02-15 Martin Sebor <msebor@redhat.com>
37160 PR middle-end/32003
37161 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37162 removed in a prior commit.
37164 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37166 PR tree-optimization/79347
37167 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37168 counters during peeling.
37170 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37172 * Makefile.in (site.exp): Remove "set ISLVER".
37174 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37177 * real.c (real_from_integer): Call real_convert even for decimal.
37179 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37182 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37184 2017-02-14 Andrew Pinski <apinski@cavium.com>
37186 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37187 cores and change the partno/implementer to be correct.
37188 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37189 the 'B" as the implementer.
37190 * config/aarch64/aarch64-tune.md: Regenerate.
37192 2017-02-14 Carl Love <cel@us.ibm.com>
37194 * config/rs6000/rs6000.c: Add case statement entry to make the
37195 xvcvuxdsp built-in argument unsigned.
37196 * config/rs6000/vsx.md: Fix the source and return operand types so they
37197 match the instruction definitions from the ISA document. Fix typo
37198 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37201 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37204 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37205 member early_clobber_alts.
37206 * lra-lives.c (reg_early_clobber_p): New.
37207 (process_bb_lives): Use it.
37208 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37209 (debug_operand_data): Initialize early_clobber_alts.
37210 (setup_operand_alternative): Set up early_clobber_alts.
37211 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37212 alternatives to new_insn_reg.
37213 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37215 (lra_update_insn_regno_info): Pass the new arg.
37217 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37219 PR middle-end/79505
37220 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37221 (new_oacc_loop_raw): Don't clear already cleared fields.
37224 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37225 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37226 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37228 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37231 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37233 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37236 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37237 the extra instruction to the right place to store 128-bit constant
37240 2017-02-14 Martin Sebor <msebor@redhat.com>
37242 PR middle-end/79448
37243 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37244 warning for strings of unknown length.
37246 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37248 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37250 2017-02-14 Jeff Law <law@redhat.com>
37253 * ira-costs.c (scan_one_insn): Initialize register move costs
37254 for pseudos seen in USE/CLOBBER insns.
37256 PR tree-optimization/79095
37257 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37258 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37259 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37260 if the operands are known to be not equal, then the resulting range
37262 (intersect_ranges): If the new range is ~[0,0] and the old range is
37263 wide, then prefer ~[0,0].
37264 * tree-vrp.c (overflow_comparison_p_1): New function.
37265 (overflow_comparison_p): New function.
37266 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37267 if NAME is used in an overflow test.
37268 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37269 overflow check that can be expressed as an equality test, then adjust
37270 ops to be that equality test.
37272 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37274 * config/s390/s390-builtin-types.def: Remove flags argument.
37275 * config/s390/s390.c (s390_init_builtins): Likewise.
37277 2017-02-14 Martin Liska <mliska@suse.cz>
37279 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37280 vector. Fix trailing white spaces.
37282 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37284 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37287 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37289 PR rtl-optimization/68664
37290 * config/arm/arm.c (arm_sched_can_speculate_insn):
37291 New function. Declare prototype.
37292 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37294 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37296 PR rtl-optimization/68664
37297 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37299 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37301 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37303 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37304 max skip bytes for function, loop and jump.
37306 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37308 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37309 ABS_EXPR for gimple dump.
37311 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37314 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37316 PR tree-optimization/79408
37317 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37318 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37319 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37320 also if rhs1 is INTEGER_CST.
37322 2017-02-14 Richard Biener <rguenther@suse.de>
37324 PR middle-end/79432
37325 * tree-into-ssa.c (insert_phi_nodes): When the function can
37326 have abnormal edges rewrite SSA names with broken use-def
37327 dominance out of SSA and register them for PHI insertion.
37329 2017-02-13 Martin Sebor <msebor@redhat.com>
37331 PR middle-end/79496
37332 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37333 clearing info.nowrite flag when snprintf size argument is a range.
37335 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37337 * cprop.c (cprop_jump): Add missing space in string literal.
37338 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37339 (get_constraint_for_component_ref): Likewise.
37340 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37341 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37342 * lra-constraints.c (process_alt_operands): Likewise.
37343 * ipa-inline.c (inline_small_functions): Likewise.
37344 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37345 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37346 * trans-mem.c (diagnose_tm_1_op): Likewise.
37347 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37348 (grid_parallel_clauses_gridifiable): Likewise.
37350 * config/nvptx/mkoffload.c (process): Add space in between
37353 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37354 "MOD4_SSE_REGS" and "ALL_REGS".
37356 * spellcheck.c (test_data): Add , in between "foo" and "food".
37358 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37361 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37362 boundary crossing check and subsequent code generation agree.
37364 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37366 * config/aarch64/aarch64.c (has_memory_op): Delete.
37367 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37370 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37372 PR rtl-optimization/79388
37373 PR rtl-optimization/79450
37374 * combine.c (distribute_notes): When removing TEM_INSN for which
37375 corresponding dest has last value recorded, invalidate that last
37378 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37380 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37381 of explicit '@'. Add missing assembly comment marker on branch costs
37384 2017-02-13 Nathan Sidwell <nathan@acm.org>
37386 * gengtype-lex.l (<in_struct>): Add '/'.
37388 2017-02-13 Martin Liska <mliska@suse.cz>
37391 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37393 2017-02-13 Richard Biener <rguenther@suse.de>
37395 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37397 * configure: Re-generate.
37398 * config.in: Likewise.
37399 * graphite-dependences.c: Simplify as if
37400 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37401 * graphite-isl-ast-to-gimple.c: Likewise.
37402 * graphite-optimize-isl.c: Likewise.
37403 * graphite-poly.c: Likewise.
37404 * graphite-sese-to-poly.c: Likewise.
37405 * graphite.h: Likewise.
37406 * toplev.c: Include isl/version.h and use isl_version () for
37407 printing the ISL version.
37408 * doc/install.texi: Update ISL requirement.
37410 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37412 * doc/standards.texi (Standards): Update reference to
37415 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37417 * doc/extend.texi (Named Address Spaces): sourceware.org now
37419 * doc/install.texi (Binaries): Ditto.
37422 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37424 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37425 standard terminology "stringize"/"stringizing" throughout.
37426 * doc/cppinternals.texi: Likewise.
37428 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37430 * doc/extend.texi: Fix some spelling mistakes and typos.
37431 * doc/invoke.texi: Likewise.
37433 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37436 * params.def (inline-min-speedup) Change from 10 to 8.
37438 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37440 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
37443 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
37446 * ipa-inline-analysis.c (get_minimal_bb): New function.
37447 (record_modified): Use it.
37448 (remap_edge_change_prob): Handle also ancestor functions.
37450 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
37452 * doc/contrib.texi (Contributors): Remove broken link into
37453 the Mauve CVS repository.
37455 2017-02-11 Jakub Jelinek <jakub@redhat.com>
37457 PR middle-end/79454
37458 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
37459 result computation whenever lhs doesn't have vector mode, not
37460 just when it has BLKmode.
37462 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37464 * doc/makefile.texi (profiledbootstrap): Refer to the
37465 installation instructions only in textual form.
37467 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37470 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
37472 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
37474 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
37475 (Specific): Update mingw-w64 reference.
37477 (Specific): Remove broken link to Renesas RX processor.
37479 2017-02-10 Richard Biener <rguenther@suse.de>
37481 * toplev.c (process_options): Do not mention obsolete graphite
37482 options when printing sorry message about missing graphite support.
37483 Mention -floop-nest-optimize.
37485 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
37487 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
37488 (vtst_p16): Likewise.
37489 (vtstq_p8): Likewise.
37490 (vtstq_p16): Likewise.
37492 (vtstq_p64): Likewise.
37493 * config/arm/arm_neon.h (vgetq_lane_p64): New.
37494 (vset_lane_p64): New.
37495 (vsetq_lane_p64): New.
37497 2017-02-10 Jakub Jelinek <jakub@redhat.com>
37499 PR tree-optimization/79411
37500 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
37501 stmt operands are SSA_NAMEs used in abnormal phis.
37502 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
37505 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37508 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
37511 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
37513 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
37515 2017-02-09 Jakub Jelinek <jakub@redhat.com>
37517 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
37521 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
37522 not arbitrary TREE_CONSTANT.
37525 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
37526 "omp declare target link" attribute unless is_global_var.
37527 * omp-offload.c (find_link_var_op): Likewise.
37529 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
37530 Chung-Lin Tang <cltang@codesourcery.com>
37532 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
37534 (gimplify_adjust_omp_clauses): Don't delete TILE.
37535 (gimplify_omp_for): Deal with TILE.
37536 * internal-fn.c (expand_GOACC_TILE): New function.
37537 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
37539 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
37540 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
37542 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
37543 avoid DIV for outermost collapse var.
37544 (expand_oacc_for): Insert tile element loop as needed. Adjust.
37545 Remove out of date comments, fix whitespace.
37546 * omp-general.c (omp_extract_for_data): Deal with tiling.
37547 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
37548 adjust OLF_DIM_BASE value.
37549 (struct omp_for_data): Add tiling field.
37550 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
37551 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
37552 for auto loops. Remove default auto determining, moved to
37553 oacc_loop_fixed_partitions.
37554 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
37555 stmts, add e_mask field.
37556 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
37557 (oacc_thread_numbers): Use oacc_dim_call.
37558 (oacc_xform_tile): New.
37559 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
37560 (finish_oacc_loop): Adjust for ifns vector.
37561 (oacc_loop_discover_walk): Append loop abstraction sites to list,
37562 add case for GOACC_TILE fns.
37563 (oacc_loop_xform_loop): Delete.
37564 (oacc_loop_process): Iterate over call list directly, and add
37565 handling for GOACC_TILE fns.
37566 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
37568 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
37569 vector partitioning to outer loops. Assign 2 partitions to loops
37570 when available. Add TILE handling.
37571 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
37572 (execite_oacc_device_lower): Process GOACC_TILE fns,
37573 ignore unknown specs.
37574 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
37575 * tree.c (omp_clause_num_ops): Adjust TILE ops.
37576 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
37578 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
37580 * configure.ac (ACX_BUGURL): Update.
37581 * configure: Regenerate.
37583 2017-02-09 Richard Biener <rguenther@suse.de>
37585 PR tree-optimization/69823
37586 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
37587 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
37589 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37591 * config/arc/arc-c.def: Add __NPS400__ definition.
37592 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
37593 (TARGET_NPS400): Define.
37595 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
37597 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
37599 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
37600 pointer, arch_info.
37601 (arc_cpu_types): Fill the arch_info field with a pointer into the
37602 arc_arch_types table.
37603 (arc_selected_cpu): Declare.
37604 * config/arc/arc.c (arc_selected_cpu): Make global.
37605 (arc_selected_arch): Delete.
37606 (arc_base_cpu): Delete.
37607 (arc_override_options): Remove references to deleted variables,
37608 update access to arch information.
37609 (ARC_OPT): Update access to arch information.
37610 (ARC_OPTX): Likewise.
37611 * config/arc/arc.h (arc_base_cpu): Remove declaration.
37612 (TARGET_ARC600): Update access to arch information.
37613 (TARGET_ARC601): Likewise.
37614 (TARGET_ARC700): Likewise.
37615 (TARGET_EM): Likewise.
37616 (TARGET_HS): Likewise.
37617 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
37620 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
37623 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
37624 condition/operands for integer GE/LE/GEU/LEU operations.
37626 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
37628 PR translation/79397
37629 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
37632 2017-02-08 Martin Jambor <mjambor@suse.cz>
37635 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
37636 whether allocation happened.
37637 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
37638 nothing was allocated.
37640 2017-02-08 Jakub Jelinek <jakub@redhat.com>
37642 PR tree-optimization/79408
37643 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
37644 constant, but SSA_NAME with a known integer range, use the minimum
37645 of that range instead of op1 to determine if modulo can be replaced
37646 with its first operand.
37648 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37650 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
37652 2017-02-08 Richard Biener <rguenther@suse.de>
37654 PR tree-optimization/71824
37655 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37656 Check all loops contained in the merged region.
37658 2017-02-07 Andrew Pinski <apinski@cavium.com>
37660 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
37662 2017-02-07 Andrew Pinski <apinski@cavium.com>
37664 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
37665 (thunderxt88): Likewise.
37666 (thunderxt81): Disable LSE and change v8.1 to v8.
37667 (thunderxt83): Likewise.
37669 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37670 Richard Biener <rguenther@suse.de>
37672 PR middle-end/79399
37673 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
37674 type from int to size_t.
37675 * ira-costs.c (struct_costs_size): Change type from int to size_t.
37677 2017-02-07 Jakub Jelinek <jakub@redhat.com>
37679 PR rtl-optimization/79386
37680 * cprop.c (bypass_conditional_jumps): Initialize
37681 bypass_last_basic_block already before splitting bbs after
37682 unconditional traps...
37683 (bypass_conditional_jumps): ... rather than here.
37686 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
37687 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
37688 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
37689 fix -masm=intel patterns.
37691 2017-02-07 Richard Biener <rguenther@suse.de>
37693 PR tree-optimization/79256
37694 PR middle-end/79278
37695 * builtins.c (get_object_alignment_2): Use min_align_of_type
37696 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
37697 and ADJUST_FIELD_ALIGN.
37699 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
37701 * doc/tm.texi: Regenerate.
37702 * stor-layout.c (layout_decl): Adjust.
37703 (update_alignment_for_field): Likewise.
37704 (place_field): Likewise.
37705 (min_align_of_type): Likewise.
37706 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
37707 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
37708 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
37709 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
37710 * config/frv/frv.c (frv_adjust_field_align): Likewise.
37711 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
37712 * config/i386/i386.c (x86_field_alignment): Likewise.
37713 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
37714 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
37715 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
37716 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
37717 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
37718 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
37722 2017-01-30 Richard Biener <rguenther@suse.de>
37724 PR tree-optimization/79256
37725 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37726 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37729 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
37731 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
37732 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
37733 builtins to SImode and emit a zero-extend, if necessary.
37735 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37737 * docs/invoke.texi (RISC-V Options): Alphabetize.
37739 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37741 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
37744 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
37746 * config/riscv/riscv.c: New file.
37747 * common/config/riscv/riscv-common.c: Likewise.
37748 * config.gcc: Likewise.
37749 * config/riscv/constraints.md: Likewise.
37750 * config/riscv/elf.h: Likewise.
37751 * config/riscv/generic.md: Likewise.
37752 * config/riscv/linux.h: Likewise.
37753 * config/riscv/multilib-generator: Likewise.
37754 * config/riscv/peephole.md: Likewise.
37755 * config/riscv/pic.md: Likewise.
37756 * config/riscv/predicates.md: Likewise.
37757 * config/riscv/riscv-builtins.c: Likewise.
37758 * config/riscv/riscv-c.c: Likewise.
37759 * config/riscv/riscv-ftypes.def: Likewise.
37760 * config/riscv/riscv-modes.def: Likewise.
37761 * config/riscv/riscv-opts.h: Likewise.
37762 * config/riscv/riscv-protos.h: Likewise.
37763 * config/riscv/riscv.h: Likewise.
37764 * config/riscv/riscv.md: Likewise.
37765 * config/riscv/riscv.opt: Likewise.
37766 * config/riscv/sync.md: Likewise.
37767 * config/riscv/t-elf-multilib: Likewise.
37768 * config/riscv/t-linux: Likewise.
37769 * config/riscv/t-linux-multilib: Likewise.
37770 * config/riscv/t-riscv: Likewise.
37771 * configure.ac: Likewise.
37772 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37773 Waterman as RISC-V maintainers.
37774 * doc/install.texi: Add RISC-V entries.
37775 * doc/invoke.texi: Add RISC-V options section.
37776 * doc/md.texi: Add RISC-V constraints section.
37777 * configure: Regenerated.
37779 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37782 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37783 false values to be constant vectors with all 0 or all 1 bits set.
37784 (vcondu<mode><mode>): Likewise.
37785 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37787 (fpmask_comparison_operator): Update comment.
37788 (vecint_comparison_operator): New predicate.
37789 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37790 vector conditionals when the true and false values are constant
37791 vectors with all 0 bits or all 1 bits set.
37793 2017-02-06 Martin Sebor <msebor@redhat.com>
37795 PR tree-optimization/79376
37796 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37798 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37800 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37801 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37802 to simplify split condition.
37804 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37806 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37807 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37810 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37812 PR rtl-optimization/68664
37813 * target.def (can_speculate_insn): New hook.
37814 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37815 * doc/tm.texi: Regenerate.
37816 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37817 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37818 (rs6000_sched_can_speculate_insn): New function.
37820 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37822 PR tree-optimization/79284
37823 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37824 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37825 vectorizable_mask_load_store, vectorizable_operation,
37826 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37827 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37828 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37829 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37830 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37831 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37832 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37833 is_gimple_assign (stmt). Replace another such test with
37834 is_gimple_assign (stmt).
37836 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37839 * config/avr/avr.c (rtl-iter.h): Include it.
37840 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37841 (avr_legitimate_combined_insn): ...and implementation.
37843 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37845 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37846 * config/s390/s390.c (s390_const_operand_ok)
37847 (s390_canonicalize_comparison, s390_extract_part)
37848 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37849 (s390_contiguous_bitmask_p, s390_rtx_costs)
37850 (legitimize_pic_address): Likewise.
37851 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37852 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37853 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37854 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37855 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37857 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37859 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37860 REGNO($0) == REGNO($1).
37862 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37864 * config/s390/linux.h(SIZE_TYPE): Add comment.
37866 2017-02-06 Julian Brown <julian@codesourcery.com>
37867 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37868 Virendra Pathak <virendra.pathak@broadcom.com>
37870 * config/aarch64/aarch64-cores.def: Change the scheduler
37872 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37873 * config/aarch64/thunderx2t99.md: New file.
37875 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37877 * doc/standards.texi (Go Language): Update link to language
37880 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37882 * tree-eh.c (lower_resx): Sanitize profile.
37883 (cleanup_empty_eh_move_lp): Likewise.
37885 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37888 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37890 * cfgloopmanip.h (loop_version): Update prototype.
37891 * modulo-sched.c (sms_schedule): Update call of loop_version.
37892 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37893 * tree-parloops.c (gen_parallel_loop): Likewise.
37894 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37895 * tree-ssa-loop-split.c (split_loop): Likewise.
37896 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37897 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37899 2017-02-05 Martin Liska <mliska@suse.cz>
37902 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37904 (print_operand_address): Initialize a struct to zero.
37906 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37908 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37909 garbage collector only in textual form.
37911 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37913 * doc/extend.texi (x86 specific memory model extensions for
37914 transactional memory): Simplify a phrase.
37916 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37919 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37920 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37921 (atomic_storedi_1): Likewise.
37923 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37925 PR tree-optimization/79338
37926 * tree-parloops.c (gather_scalar_reductions): Don't call
37927 vect_analyze_loop_form for loop->inner before destroying loop's
37930 2017-02-03 Martin Sebor <msebor@redhat.com>
37932 PR tree-optimization/79327
37933 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37934 when precision has resulted in leading zeros.
37935 (format_integer): Adjust the likely counter to assume an unknown
37936 argument that may be zero is non-zero.
37938 2017-02-03 Jason Merrill <jason@redhat.com>
37941 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37942 avoid copying non-taken branch.
37944 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37946 PR tree-optimization/79340
37947 * tree-vect-loop.c (vectorizable_reduction): Release
37948 vec_defs elements after safe_splicing them into other vectors.
37951 PR tree-optimization/79327
37952 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37953 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37955 (format_integer): Use wide_int_to_tree instead of build_int_cst
37956 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37957 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37958 of shortest and longest sequence.
37960 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37962 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37963 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37965 2017-02-03 Walter Lee <walt@tilera.com>
37968 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37969 after initial stackframe link reg save.
37970 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37972 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37975 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37976 wu for stxssp alternative.
37978 2017-02-03 Martin Sebor <msebor@redhat.com>
37980 PR tree-optimization/79352
37981 * gimple-fold.c (get_range_strlen): Add argument.
37982 (get_range_strlen): Change return type to bool.
37983 (get_maxval_strlen): Pass in a dummy argument.
37984 * gimple-fold.h (get_range_strlen): Change return type to bool.
37985 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37986 * tree.h (array_at_struct_end_p): Add argument.
37987 * tree.c (array_at_struct_end_p): Handle it.
37989 2017-02-03 Martin Liska <mliska@suse.cz>
37992 * multiple_target.c (create_dispatcher_calls): Redirect edge
37993 from a caller of a dispatcher.
37994 (expand_target_clones): Make the clones local.
37995 (ipa_target_clone): Do both target clones and resolvers.
37996 (ipa_dispatcher_calls): Remove the pass.
37997 (pass_dispatcher_calls::gate): Likewise.
37998 (make_pass_dispatcher_calls): Likewise.
37999 * passes.def (pass_target_clone): Put as very first IPA early
38002 2017-02-03 Martin Liska <mliska@suse.cz>
38004 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38005 in case of a function with ifunc attribute.
38007 2017-02-03 Martin Liska <mliska@suse.cz>
38009 * cgraph.c (cgraph_node::dump): Dump function version info.
38010 * symtab.c (symtab_node::dump_base): Add missing new line.
38012 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38014 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38015 (ifcombine_ifandif): Use it.
38017 2017-02-03 Martin Liska <mliska@suse.cz>
38019 * doc/invoke.texi: Document default value for
38020 use-after-scope-direct-emission-threshold.
38022 2017-02-03 Martin Liska <mliska@suse.cz>
38024 PR tree-optimization/79339
38025 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38026 (format_floating): Likewise.
38028 2017-02-03 Martin Liska <mliska@suse.cz>
38031 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38033 (ipa_node_params_t::remove): Likewise.
38034 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38035 initialization from removed ipa_node_params_t::insert.
38036 (ipa_node_params::~ipa_node_params): Move from removed
38037 ipa_node_params_t::release.
38038 * symbol-summary.h (symbol_summary::m_released): New member.
38039 Do not release a summary twice. Do not allow to call finalizer
38040 for types of a summary that live in GGC memory.
38042 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38044 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38047 2017-02-02 Martin Sebor <msebor@redhat.com>
38049 PR middle-end/79275
38050 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38051 (format_string): Tighten up the range of output for non-constant
38052 strings and correct the expected range for wide non-constant strings.
38054 2017-02-02 Martin Sebor <msebor@redhat.com>
38056 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38058 PR middle-end/32003
38059 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38061 (-fdump-tree-@var): Add to index and document how to come up
38062 with pass-specific option and dump file names.
38063 (-fdump-passes): Clarify where to look for output.
38065 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38067 PR middle-end/77445
38068 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38069 statistics of the analyzed path; allow threading for speed when
38070 any of BBs along the path are optimized for speed.
38072 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38074 PR middle-end/78468
38075 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38076 settings of the virtual registers.
38079 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38081 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38082 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38085 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38087 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38088 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38090 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38092 * config/s390/s390.md: Add missing comments with the expanded
38094 * config/s390/vector.md: Likewise.
38095 * config/s390/vx-builtins.md: Likewise.
38097 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38100 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38101 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38102 conditions on a single line.
38104 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38106 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38107 __S390_VX__ to __VX__.
38109 2017-02-01 Andrew Pinski <apinski@cavium.com>
38111 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38112 stmt_info to record_stmt_cost.
38113 (vect_get_known_peeling_cost): Pass stmt_info if known to
38115 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38116 cpu_vector_cost field into
38117 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38118 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38119 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38120 splitting of scalar_stmt_cost and vec_stmt_cost.
38121 (thunderx_vector_cost): Likewise.
38122 (cortexa57_vector_cost): LIkewise.
38123 (exynosm1_vector_cost): Likewise.
38124 (xgene1_vector_cost): Likewise.
38125 (thunderx2t99_vector_cost): Improve after the splitting of the two
38127 (aarch64_builtin_vectorization_cost): Update for the splitting of
38128 scalar_stmt_cost and vec_stmt_cost.
38130 2017-02-01 Torvald Riegel <triegel@redhat.com>
38131 Richard Henderson <rth@redhat.com>
38133 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38134 conditional on existance of a fast atomic load.
38135 * optabs-query.c (can_atomic_load_p): New function.
38136 * optabs-query.h (can_atomic_load_p): Declare it.
38137 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38138 no fast atomic load is available for the particular size of access.
38139 (expand_atomic_compare_and_swap): Likewise.
38140 (expand_atomic_load): Likewise.
38141 (expand_atomic_store): Likewise.
38142 (expand_atomic_fetch_op): Likewise.
38143 * testsuite/lib/target-supports.exp
38144 (check_effective_target_sync_int_128): Remove x86 because it provides
38145 no fast atomic load.
38146 (check_effective_target_sync_int_128_runtime): Likewise.
38148 2017-02-01 Richard Biener <rguenther@suse.de>
38150 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38151 (graphite_transform_loops): Provide opt-info for optimized nests.
38152 * tree-parloop.c (parallelize_loops): Provide opt-info for
38153 parallelized loops.
38155 2017-02-01 Richard Biener <rguenther@suse.de>
38157 PR middle-end/79315
38158 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38159 was not set before.
38161 2017-02-01 Richard Biener <rguenther@suse.de>
38163 PR tree-optimization/71824
38164 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38165 Verify the loops are valid in the merged SESE region.
38166 (scop_detection::can_represent_loop_1): Check analyzing the
38167 evolution of the number of iterations in the region succeeds.
38169 2017-01-31 Ian Lance Taylor <iant@golang.org>
38171 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38172 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38174 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38176 PR preprocessor/79210
38177 * input.c (get_substring_ranges_for_loc): Replace line_width
38178 assertion with error-handling.
38180 2017-01-31 Richard Biener <rguenther@suse.de>
38182 PR tree-optimization/77318
38183 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38184 (create_pw_aff_from_tree): Take loop parameter.
38185 (add_condition_to_pbb): Pass loop of the condition to
38186 create_pw_aff_from_tree.
38188 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38190 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38191 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38193 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38197 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38199 (convert_int_to_float128): Likewise.
38200 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38201 (convert_int_to_float128): Likewise.
38202 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38203 (UNSPEC_IEEE128_CONVERT): Likewise.
38204 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38205 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38206 Use local variables for IBM extended format.
38207 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38208 (fix_trunc<mode>si2_fprs): Likewise.
38209 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38210 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38211 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38212 to know that we can now have integers of all sizes in vector
38214 (fix<uns>_<mode>di2_hw): Likewise.
38215 (float<uns>_<mode>si2_hw): Likewise.
38216 (fix_<mode>si2_hw): Likewise.
38217 (fixuns_<mode>si2_hw): Likewise.
38218 (float<uns>_<mode>di2_hw): Likewise.
38219 (float_<mode>di2_hw): Likewise.
38220 (float_<mode>si2_hw): Likewise.
38221 (floatuns_<mode>di2_hw): Likewise.
38222 (floatuns_<mode>si2_hw): Likewise.
38223 (xscvqp<su>wz_<mode>): Delete, no longer used.
38224 (xscvqp<su>dz_<mode>): Likewise.
38225 (xscv<su>dqp_<mode>): Likewise.
38226 (ieee128_mfvsrd_64bit): Likewise.
38227 (ieee128_mfvsrd_32bit): Likewise.
38228 (ieee128_mfvsrwz): Likewise.
38229 (ieee128_mtvsrw): Likewise.
38230 (ieee128_mtvsrd_64bit): Likewise.
38231 (ieee128_mtvsrd_32bit): Likewise.
38233 2017-01-31 Martin Liska <mliska@suse.cz>
38236 * ipa-prop.c (ipa_free_all_node_params): Call release method
38237 instead of ~sumbol_summary to not to trigger double times
38240 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38242 PR tree-optimization/71691
38243 * bitmap.h (class auto_bitmap): New.
38244 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38245 is_maybe_undefined instead of ssa_undefined_value_p.
38247 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38249 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38250 __S390_ARCH_LEVEL__ to __ARCH__.
38252 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38254 PR tree-optimization/79267
38255 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38256 if should_remove_lhs_p is true.
38258 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38261 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38262 (add_alignment_attribute): New.
38263 (base_type_die): Add alignment attribute.
38264 (subrange_type_die): Likewise.
38265 (modified_type_die): Likewise.
38266 (gen_array_type_die): Likewise.
38267 (gen_descr_array_type_die: Likewise.
38268 (gen_enumeration_type_die): Likewise.
38269 (gen_subprogram_die): Likewise.
38270 (gen_variable_die): Likewise.
38271 (gen_field_die): Likewise.
38272 (gen_ptr_to_mbr_type_die): Likewise.
38273 (gen_struct_or_union_type_die): Likewise.
38274 (gen_subroutine_type_die): Likewise.
38275 (gen_typedef_die): Likewise.
38276 (base_type_cmp): Compare alignment attribute.
38278 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38281 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38282 (setb_unsigned) New pattern for setb with CCUNS.
38283 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38284 subfc./subfe sequence to avoid overflow problems. Generate a
38285 shorter sequence with cmpld/setb for power9.
38286 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38287 for generating subfc. instruction.
38288 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38289 now uses this instruction.
38291 2017-01-30 Ian Lance Taylor <iant@google.com>
38294 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38295 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38297 2017-01-30 Martin Sebor <msebor@redhat.com>
38299 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38300 Move constant to the right of a relational operator.
38301 (get_mpfr_format_length, format_character, format_string): Ditto.
38302 (should_warn_p, maybe_warn): Same.
38304 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38306 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38309 * asan.c (get_translation_unit_decl): Remove function.
38310 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38312 2017-01-30 Martin Liska <mliska@suse.cz>
38314 PR gcov-profile/79259
38315 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38316 -fprofile-generate.
38318 2017-01-30 Martin Liska <mliska@suse.cz>
38321 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38322 Initialize variables with NULL value.
38324 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38327 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38329 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38331 2017-01-30 Richard Biener <rguenther@suse.de>
38333 PR tree-optimization/79276
38334 * tree-vrp.c (process_assert_insertions): Properly adjust common
38335 when removing a duplicate.
38337 2017-01-30 Richard Biener <rguenther@suse.de>
38339 PR tree-optimization/79256
38340 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38341 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38343 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38345 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38348 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38349 ("*r<noxa>sbg_<mode>_sll_bitmask")
38350 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38351 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38352 Use contiguous_bitmask_nowrap_operand.
38354 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38357 * config/rs6000/altivec.h (vec_xl): Revise #define.
38358 (vec_xst): Likewise.
38360 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38362 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38364 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38366 PR rtl-optimization/79194
38367 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38368 traps before call to bypass_conditional_jumps.
38370 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38372 PR tree-optimization/71374
38373 * lra-constraints.c (check_conflict_input_operands): New.
38374 (match_reload): Use it.
38376 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38379 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38380 account to calculate conflict_set.
38382 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38384 PR rtl-optimization/78559
38385 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38386 other_insn in combine.
38388 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38390 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38391 uint16_type_node for BT_UINT16.
38393 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38395 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38396 "RTL Tests" to menu.
38397 (GIMPLE Tests): New node.
38398 (RTL Tests): New node.
38400 2017-01-27 Richard Biener <rguenther@suse.de>
38402 PR tree-optimization/79245
38403 * tree-loop-distribution.c (distribute_loop): Apply cost
38404 modeling also to detected patterns.
38406 2017-01-27 Richard Biener <rguenther@suse.de>
38408 PR tree-optimization/71433
38409 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38410 (compare_assert_loc): New function.
38411 (process_assert_insertions): Sort and optimize assert locations
38412 to remove duplicates and push down identical assertions on
38413 edges to their destination block.
38415 2017-01-27 Richard Biener <rguenther@suse.de>
38417 PR tree-optimization/79244
38418 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38419 out SSA names even if abnormal.
38421 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38423 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38424 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38425 instead of MPFR_RNDN.
38427 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38430 * arm.c (arm_option_override): Don't call build_target_option_node
38431 until after doing all option overrides.
38432 (arm_valid_target_attribute_tree): Likewise.
38434 2017-01-27 Martin Liska <mliska@suse.cz>
38436 * doc/invoke.texi (-fprofile-arcs): Document profiling support
38437 for {cd}tors and C++ {cd}tors.
38439 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
38441 * config/s390/s390.md ("*setmem_long_and")
38442 ("*setmem_long_and_31z"): Use zero_extend instead of and.
38444 2017-01-26 Martin Sebor <msebor@redhat.com>
38446 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
38449 2017-01-26 Martin Sebor <msebor@redhat.com>
38451 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
38452 HAVE_DFmode before using XFmode or DFmode.
38453 (parse_directive): Avoid using the z length modifier to avoid
38454 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
38456 PR middle-end/78703
38457 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
38458 to accept adjustment as an array.
38459 (get_int_range): New function.
38460 (struct directive): Make width and prec arrays.
38461 (directive::set_width, directive::set_precision): Call get_int_range.
38462 (format_integer, format_floating): Handle width and precision ranges.
38463 (format_string, parse_directive): Same.
38465 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38468 * dwarf2out.c (generate_skeleton_bottom_up): For children with
38469 comdat_type_p set, just clone them, but keep the children in the
38473 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
38474 which have direct callers with -fvar-tracking-assignments enabled
38476 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
38477 inside of type units.
38479 2017-01-26 Martin Sebor <msebor@redhat.com>
38481 PR middle-end/78703
38482 * gimple-ssa-sprintf.c (struct result_range): Add likely and
38484 (struct format_result): Replace number_chars, number_chars_min,
38485 and number_chars_max with a single member of struct result_range.
38487 (format_result::operator+=): Adjust.
38488 (struct fmtresult): Remove bounded. Handle likely and unlikely
38490 (fmtresult::adjust_for_width_or_precision): New function.
38491 (fmtresult:type_max_digits): New function.
38492 (bytes_remaining): Handle likely and unlikely counters.
38493 (min_bytes_remaining): Remove.
38494 (format_percent): Simplify.
38495 (format_integer, format_floating): Set likely and unlikely counters.
38496 (get_string_length, format_character, format_string): Same.
38497 (format_plain, should_warn_p): New function.
38498 (maybe_warn): Call should_warn_p. Update diagnostic messages
38499 and handle those for all directives, including plain strings.
38500 (format_directive): Handle likely and unlikely counters.
38501 Remove unnecessary quoting from diagnostics. Add an informational
38503 (add_bytes): Remove.
38504 (pass_sprintf_length::compute_format_length): Simplify.
38505 (try_substitute_return_value): Handle likely and unlikely counters.
38507 2017-01-26 Carl Love <cel@us.ibm.com>
38509 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
38510 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
38512 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
38515 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
38516 endianess for subregs into account.
38517 * lra-constraints.c (lra_constraints): Do risky transformations
38518 always on the first iteration.
38519 * lra-lives.c (check_pseudos_live_through_calls): Add arg
38520 last_call_used_reg_set.
38521 (process_bb_lives): Define and use last_call_used_reg_set.
38522 * lra.c (lra): Always continue after lra_constraints on the first
38525 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
38527 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
38529 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
38531 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38533 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
38534 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
38535 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
38536 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
38537 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
38538 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
38539 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
38540 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
38541 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
38543 2017-01-26 Marek Polacek <polacek@redhat.com>
38546 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
38547 for the third operand.
38549 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38551 PR middle-end/79236
38552 * omp-low.c (struct omp_context): Add simt_stmt field.
38553 (scan_omp_for): Return omp_context *.
38554 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
38555 context to the _simt_ SIMD stmt.
38556 (lower_omp_for): For combined SIMD with sibling _simt_
38557 SIMD, make sure to use the same decls in _looptemp_
38558 clauses as in the sibling.
38560 2017-01-26 David Sherwood <david.sherwood@arm.com>
38562 PR middle-end/79212
38563 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
38566 2017-01-26 Jakub Jelinek <jakub@redhat.com>
38569 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
38570 emit fld b; fld a; if possible.
38572 * brig-builtins.def: Update copyright years.
38573 * config/arm/arm_acle_builtins.def: Update copyright years.
38575 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
38578 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
38579 constraint instead of o for the stxsd instruction.
38581 2017-01-25 Carl Love <cel@us.ibm.com>
38583 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
38584 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
38586 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
38588 * doc/invoke.texi (C++ Dialect Options): Fix typo.
38590 2017-01-25 Richard Biener <rguenther@suse.de>
38592 PR tree-optimization/69264
38593 * target.def (vector_alignment_reachable): Improve documentation.
38594 * doc/tm.texi: Regenerate.
38595 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
38597 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
38598 earlier changes with respect to TYPE_USER_ALIGN.
38599 (vector_alignment_reachable_p): Likewise. Improve dumping.
38601 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38604 * config/arm/arm.md (xordi3): Force constant operand into a register
38607 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38609 * doc/invoke.texi (-fstore-merging): Correct default optimization
38610 levels at which it is enabled.
38611 (-O): Move -fstore-merging from list to...
38614 2017-01-25 Richard Biener <rguenther@suse.de>
38617 * omp-expand.c: Include debug.h.
38618 (expand_omp_taskreg): Make sure to generate early debug before
38619 outlining anything from a function.
38620 (expand_omp_target): Likewise.
38621 (grid_expand_target_grid_body): Likewise.
38623 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
38626 * asan.c (get_translation_unit_decl): New function.
38627 (asan_add_global): Extract modules file name from globals
38628 TRANSLATION_UNIT_DECL name.
38630 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
38633 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
38634 for long calls with APCS frame and VFP.
38636 2017-01-24 David Malcolm <dmalcolm@redhat.com>
38638 * cfg.c (original_copy_tables_initialized_p): New function.
38639 * cfg.h (original_copy_tables_initialized_p): New decl.
38640 * cfgrtl.c (relink_block_chain): Guard the call to
38641 free_original_copy_tables with a call to
38642 original_copy_tables_initialized_p.
38643 * cgraph.h (symtab_node::native_rtl_p): New decl.
38644 * cgraphunit.c (symtab_node::native_rtl_p): New function.
38645 (symtab_node::needed_p): Don't assert for early assembly output
38646 for __RTL functions.
38647 (cgraph_node::finalize_function): Set "force_output" for __RTL
38649 (cgraph_node::analyze): Bail out early for __RTL functions.
38650 (analyze_functions): Update assertion to support __RTL functions.
38651 (cgraph_node::expand): Bail out early for __RTL functions.
38652 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
38654 * function.h (struct function): Update comment for field
38656 * gimple-expr.c: Include "tree-pass.h".
38657 (gimple_has_body_p): Return false for __RTL functions.
38658 * Makefile.in (OBJS): Add run-rtl-passes.o.
38659 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
38661 (gcc::pass_manager::get_clean_slate): New accessor.
38662 * passes.c: Include "insn-addr.h".
38663 (should_skip_pass_p): Add logging. Update logic for running
38664 "expand" to be compatible with both __GIMPLE and __RTL. Guard
38665 property-provider override so it is only done for gimple passes.
38667 (skip_pass): New function.
38668 (execute_one_pass): Call skip_pass when skipping passes.
38669 * read-md.c (md_reader::read_char): Support filtering
38670 the input to a subset of line numbers.
38671 (md_reader::md_reader): Initialize fields
38672 m_first_line and m_last_line.
38673 (md_reader::read_file_fragment): New function.
38674 * read-md.h (md_reader::read_file_fragment): New decl.
38675 (md_reader::m_first_line): New field.
38676 (md_reader::m_last_line): New field.
38677 * read-rtl-function.c (function_reader::create_function): Only
38678 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
38679 curr_properties. Set DECL_INITIAL to a dummy block.
38680 (read_rtl_function_body_from_file_range): New function.
38681 * read-rtl-function.h (read_rtl_function_body_from_file_range):
38683 * run-rtl-passes.c: New file.
38684 * run-rtl-passes.h: New file.
38686 2017-01-24 Jeff Law <law@redhat.com>
38688 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
38691 2017-01-24 Bin Cheng <bin.cheng@arm.com>
38693 PR tree-optimization/79159
38694 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
38695 (record_nonwrapping_iv): Improve boundary using above function if no
38696 value range information.
38698 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
38699 Martin Jambor <mjambor@suse.cz>
38701 * brig-builtins.def: New file.
38702 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
38703 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
38704 (DEF_HSAIL_SAT_BUILTIN): Likewise.
38705 (DEF_HSAIL_INTR_BUILTIN): Likewise.
38706 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
38707 * builtin-types.def (BT_INT8): New.
38708 (BT_INT16): Likewise.
38709 (BT_UINT8): Likewise.
38710 (BT_UINT16): Likewise.
38711 (BT_FN_ULONG): Likewise.
38712 (BT_FN_UINT_INT): Likewise.
38713 (BT_FN_UINT_ULONG): Likewise.
38714 (BT_FN_UINT_LONG): Likewise.
38715 (BT_FN_UINT_PTR): Likewise.
38716 (BT_FN_ULONG_PTR): Likewise.
38717 (BT_FN_INT8_FLOAT): Likewise.
38718 (BT_FN_INT16_FLOAT): Likewise.
38719 (BT_FN_UINT32_FLOAT): Likewise.
38720 (BT_FN_UINT16_FLOAT): Likewise.
38721 (BT_FN_UINT8_FLOAT): Likewise.
38722 (BT_FN_UINT64_FLOAT): Likewise.
38723 (BT_FN_UINT16_UINT32): Likewise.
38724 (BT_FN_UINT32_UINT16): Likewise.
38725 (BT_FN_UINT16_UINT16_UINT16): Likewise.
38726 (BT_FN_INT_PTR_INT): Likewise.
38727 (BT_FN_UINT_PTR_UINT): Likewise.
38728 (BT_FN_LONG_PTR_LONG): Likewise.
38729 (BT_FN_ULONG_PTR_ULONG): Likewise.
38730 (BT_FN_VOID_UINT64_UINT64): Likewise.
38731 (BT_FN_UINT8_UINT8_UINT8): Likewise.
38732 (BT_FN_INT8_INT8_INT8): Likewise.
38733 (BT_FN_INT16_INT16_INT16): Likewise.
38734 (BT_FN_INT_INT_INT): Likewise.
38735 (BT_FN_UINT_FLOAT_UINT): Likewise.
38736 (BT_FN_FLOAT_UINT_UINT): Likewise.
38737 (BT_FN_ULONG_UINT_UINT): Likewise.
38738 (BT_FN_ULONG_UINT_PTR): Likewise.
38739 (BT_FN_ULONG_ULONG_ULONG): Likewise.
38740 (BT_FN_UINT_UINT_UINT): Likewise.
38741 (BT_FN_VOID_UINT_PTR): Likewise.
38742 (BT_FN_UINT_UINT_PTR: Likewise.
38743 (BT_FN_UINT32_UINT64_PTR): Likewise.
38744 (BT_FN_INT_INT_UINT_UINT): Likewise.
38745 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
38746 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
38747 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
38748 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
38749 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
38750 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
38751 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
38752 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
38753 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
38754 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
38755 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
38756 * doc/frontends.texi: List BRIG FE.
38757 * doc/install.texi (Testing): Add BRIG tesring requirements.
38758 * doc/invoke.texi (Overall Options): Mention BRIG.
38759 * doc/standards.texi (Standards): Doucment BRIG HSA version.
38761 2017-01-24 Richard Biener <rguenther@suse.de>
38763 PR translation/79208
38764 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
38766 2017-01-24 Martin Jambor <mjambor@suse.cz>
38769 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38770 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38771 and known_contexts.
38773 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38775 PR middle-end/79123
38776 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38777 casts from signed to unsigned really don't have a range.
38779 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38781 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38782 GMP_RNDx for compatiblity.
38784 2017-01-24 Martin Liska <mliska@suse.cz>
38787 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38788 that would prevent us to call alloca with -1 as argument.
38790 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38792 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38793 Avoid -Wformat-security warning.
38795 2017-01-23 Andrew Pinski <apinski@cavium.com>
38797 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38800 2017-01-23 Martin Sebor <msebor@redhat.com>
38802 PR middle-end/78703
38803 * gimple-ssa-sprintf.c (warn_level): New global.
38804 (format_integer): Use it here and throughout the rest of the file.
38805 Use the same switch to compute sign as base.
38806 (maybe_warn): New function.
38807 (format_directive): Factor out warnings into maybe_warn.
38808 Add debugging output. Use warn_level.
38809 (add_bytes): Use warn_level.
38810 (pass_sprintf_length::compute_format_length): Add debugging output.
38811 (try_substitute_return_value): Same.
38812 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38814 PR middle-end/78703
38815 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38816 (struct fmtresult, format_integer, format_floating): Adjust.
38817 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38818 (get_string_length, format_string,format_directive): Same.
38819 (pass_sprintf_length::compute_format_length): Same.
38820 (try_substitute_return_value): Simplify slightly.
38822 PR middle-end/78703
38823 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38824 (fmtresult::operator+=): Outlined.
38825 (struct fmtresult): Add ctors.
38826 (struct conversion_spec): Rename...
38827 (struct directive): ...to this. Add and remove data members.
38828 (directive::set_width, directive::set_precision): New functions.
38829 (format_percent): Use fmtresult ctor.
38830 (get_width_and_precision): Remove.
38831 (format_integer): Make naming changes. Avoid computing width and
38833 (format_floating): Same. Adjust indentation.
38834 (format_character, format_none): New functions.
38835 (format_string): Moved character handling to format_character.
38836 (format_directive): Remove arguments, change return type.
38837 (parse_directive): New function.
38838 (pass_sprintf_length::compute_format_length): Move directive
38839 parsing to parse_directive.
38841 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38843 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38844 (assign_assembler_name_if_needed): ... this.
38845 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38846 (assign_assembler_name_if_needed): ... this.
38847 (free_lang_data_in_cgraph): Adjust callers.
38848 * cgraphunit.c (cgraph_node::analyze): Likewise.
38849 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38852 2017-01-23 Richard Biener <rguenther@suse.de>
38854 PR tree-optimization/79088
38855 PR tree-optimization/79188
38856 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38857 resetting loop bounds after last path deletion. Reset loop
38858 bounds of the target loop, make code match the comments.
38859 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38860 Make sure loops need no fixups.
38862 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38864 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38865 exponent support with double type for first argument.
38866 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38867 type returned by __builtin_vec_extract_sig,
38868 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38869 functions from "vector int" to "vector unsigned int" or from
38870 "vector long long int" to "vector unsigned long long int".
38871 Changed type returned by __builtin_vec_extract_exp,
38872 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38873 functions from "vector int" to "vector unsigned int" or from
38874 "vector long long int" to "vector unsigned long long int".
38875 Changed return type of __builtin_vec_test_data_class,
38876 __builtin_vec_test_data_class_sp, and
38877 __builtin_vec_test_data_class_dp from "vector int" to
38878 "vector bool int" or from "vector long long int" to "vector bool
38879 long long int" and changed second argument type from "unsigned
38880 int" to "int". Added new overloaded function forms "vector float
38881 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38882 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38883 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38884 double, vector unsigned long long int)" and "vector double
38885 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38886 long int)". Changed return type of
38887 __builtin_scalar_test_data_class and
38888 __builtin_scalar_test_data_class_sp and
38889 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38890 int" and changed second argument from "unsigned int" to "int".
38891 Changed type returned by __builtin_scalar_test_neg,
38892 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38893 from "int" to "bool int". Added new overloaded function form
38894 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38895 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38896 exponent double-precision with floating point first argument.
38897 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38898 documentation of scalar_test_data_class, scalar_test_neg,
38899 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38900 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38901 vec_test_data_class built-in functions to reflect refinements in
38902 their type signatures.
38904 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38906 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38908 (aarch64_elf_asm_destructor): Likewise.
38910 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38912 PR rtl-optimization/78634
38913 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38914 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38915 * ifcvt.c (noce_try_cmove): Add missing cost check.
38917 PR rtl-optimization/71724
38918 * combine.c (if_then_else_cond): Look for situations where it is
38919 beneficial to undo the work of one of the recursive calls.
38921 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38923 PR tree-optimization/70754
38924 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38925 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38926 combined stmt before it if not NULL.
38927 (combine_chains): Process refs reversely and compute dominance point
38930 2017-01-23 Martin Liska <mliska@suse.cz>
38932 PR tree-optimization/79196
38933 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38934 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38936 (strlen_optimize_stmt): Call the renamed function.
38938 2017-01-23 Michael Matz <matz@suse.de>
38940 PR tree-optimization/78384
38941 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38943 2017-01-23 Richard Biener <rguenther@suse.de>
38945 PR tree-optimization/79186
38946 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38947 both incoming edges before moving an assert.
38949 2017-01-23 Martin Jambor <mjambor@suse.cz>
38951 * ipa-prop.c (load_from_param_1): Removed.
38952 (load_from_unmodified_param): Bits from load_from_param_1 put back
38954 (load_from_param): Removed.
38955 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38956 with stmt. Reverted back to use of load_from_unmodified_param.
38958 2017-01-23 Martin Jambor <mjambor@suse.cz>
38961 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38962 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38963 field a pointer to garbage collected vector, mark lattices and
38964 ipcp_orig_node with GTY((skip)).
38965 (ipa_get_param_count): Adjust to descriptors being a pointer.
38966 (ipa_get_param): Likewise.
38967 (ipa_get_type): Likewise.
38968 (ipa_get_param_move_cost): Likewise.
38969 (ipa_set_param_used): Likewise.
38970 (ipa_get_controlled_uses): Likewise.
38971 (ipa_set_controlled_uses): Likewise.
38972 (ipa_is_param_used): Likewise.
38973 (ipa_node_params_t): Move into garbage collector. New methods insert
38975 (ipa_node_params_sum): Annotate wth GTY(()).
38976 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38978 (ipa_load_from_parm_agg): Adjust declaration.
38979 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38980 * ipa-profile.c (ipa_profile): Likewise.
38981 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38982 (ipa_populate_param_decls): Make descriptors parameter garbage
38984 (ipa_dump_param): Adjust to descriptors being a pointer.
38985 (ipa_alloc_node_params): Likewise.
38986 (ipa_initialize_node_params): Likewise.
38987 (load_from_param_1): Make descriptors parameter garbage collected.
38988 (load_from_unmodified_param): Likewise.
38989 (load_from_param): Likewise.
38990 (ipa_load_from_parm_agg): Likewise.
38991 (ipa_node_params::~ipa_node_params): Removed.
38992 (ipa_free_all_node_params): Remove call to delete operator.
38993 (ipa_node_params_t::insert): New.
38994 (ipa_node_params_t::remove): Likewise.
38995 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38996 copy known_csts and known_contexts vectors.
38997 (ipa_read_node_info): Adjust to descriptors being a pointer.
38998 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39000 (ipcp_transform_function): Make descriptors variable garbage
39003 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39005 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39006 * config/i386/avx512dqintrin.h: Ditto.
39007 * config/i386/avx512fintrin.h: Ditto.
39008 * config/i386/i386.c: Handle new builtins.
39009 * config/i386/i386-builtin.def: Add new builtins.
39010 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39011 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39013 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39014 Martin Liska <mliska@suse.cz>
39016 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39017 * asan.c (asan_expand_poison_ifn): Support stores and use
39018 appropriate ASAN report function.
39019 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39020 * internal-fn.def (ASAN_POISON_USE): Declare.
39021 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39022 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39023 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39024 ASAN_POISON calls w/o LHS.
39025 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39026 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39027 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39028 * gimplify.c (asan_poison_variables): Add attribute
39029 use_after_scope_memory to variables that really needs to live
39031 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39032 having the attribute.
39034 2017-01-23 Martin Liska <mliska@suse.cz>
39036 * asan.c (create_asan_shadow_var): New function.
39037 (asan_expand_poison_ifn): Likewise.
39038 * asan.h (asan_expand_poison_ifn): New declaration.
39039 * internal-fn.c (expand_ASAN_POISON): Likewise.
39040 * internal-fn.def (ASAN_POISON): New builtin.
39041 * sanopt.c (pass_sanopt::execute): Expand
39042 asan_expand_poison_ifn.
39043 * tree-inline.c (copy_decl_for_dup_finish): Make function
39045 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39046 * tree-ssa.c (is_asan_mark_p): New function.
39047 (execute_update_addresses_taken): Rewrite local variables
39048 (identified just by use-after-scope as addressable) into SSA.
39050 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39052 * doc/install.texi (Specific): opensource.apple.com uses https
39053 now. Remove trailing slash.
39055 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39057 * README.Portability: Remove note on an Irix compatibility issue.
39059 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39061 * gcov.c (INCLUDE_ALGORITHM): Define.
39062 (INCLUDE_VECTOR): Define.
39063 No longer include <vector> and <algorithm> directly.
39065 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39067 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39069 * doc/invoke.texi (Code Gen Options): Ditto.
39071 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39074 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39076 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39078 rtl-optimization/79125
39079 * cprop.c (local_cprop_pass): Handle cases where we make an
39080 unconditional trap.
39082 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39086 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39087 read from, for big endian.
39089 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39091 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39092 register pauth builtins for LP64 only.
39094 2017-01-20 Marek Polacek <polacek@redhat.com>
39097 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39100 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39102 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39104 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39105 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39106 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39108 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39111 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39112 in big-endian mode when they are not a single duplicated value.
39114 2017-01-20 Richard Biener <rguenther@suse.de>
39116 * BASE-VER: Bump to 7.0.1.
39118 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39120 * omp-low.c (omplow_simd_context): New struct. Use it...
39121 (lower_rec_simd_input_clauses): ...here and...
39122 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39123 references to idx, lane, max_vf, is_simt.
39125 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39127 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39130 2017-01-20 Martin Jambor <mjambor@suse.cz>
39132 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39133 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39135 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39136 (GTFILES): Rename hsa.c to hsa-common.c.
39137 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39138 * hsa-dump.c: Likewise.
39139 * hsa-gen.c: Likewise.
39140 * hsa-regalloc.c: Likewise.
39141 * ipa-hsa.c: Likewise.
39142 * omp-expand.c: Likewise.
39143 * omp-low.c: Likewise.
39144 * toplev.c: Likewise.
39146 2017-01-20 Marek Polacek <polacek@redhat.com>
39149 * doc/invoke.texi: Document -Wduplicated-branches.
39150 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39151 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39152 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39153 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39154 return 0 only when not OEP_LEXICOGRAPHIC.
39155 (fold_build_cleanup_point_expr): Use the expression
39156 location when building CLEANUP_POINT_EXPR.
39157 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39158 * tree.c (add_expr): Handle error_mark_node.
39160 2017-01-20 Martin Liska <mliska@suse.cz>
39163 * tree-profile.c (init_ic_make_global_vars): Do not call
39165 (gimple_init_gcov_profiler): Likewise.
39167 2017-01-20 Martin Liska <mliska@suse.cz>
39170 * cgraph.h (maybe_create_reference): Remove argument and
39172 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39174 * ipa-cp.c (create_specialized_node): Likewise.
39175 * symtab.c (symtab_node::maybe_create_reference): Handle
39176 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39178 2017-01-20 Martin Liska <mliska@suse.cz>
39180 * read-rtl-function.c (function_reader::create_function): Use
39181 build_decl instread of build_decl_stat.
39183 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39185 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39186 * config/i386/avx512dqintrin.h: Ditto.
39187 * config/i386/avx512fintrin.h: Ditto.
39188 * config/i386/i386-builtin-types.def: Add new types.
39189 * config/i386/i386.c: Handle new types.
39190 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39191 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39192 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39193 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39194 (__builtin_ia32_kshiftridi): New.
39195 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39197 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39201 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39202 define to rs6000_init_stack_protect_guard.
39203 (rs6000_init_stack_protect_guard): New function.
39205 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39206 Yunqiang Su <yunqiang.su@imgtec.com>
39208 * config.gcc (supported_defaults): Add madd4.
39209 (with_madd4): Add validation.
39210 (all_defaults): Add madd4.
39211 * config/mips/mips.opt (mmadd4): New option.
39212 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39214 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39216 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39217 (ISA_HAS_FUSED_MADD4): Likewise.
39218 * doc/invoke.texi (-mmadd4): Document the new option.
39219 * doc/install.texi (--with-madd4): Document the new option.
39221 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39223 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39224 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39225 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39226 (aarch64_init_pauth_hint_builtins): New.
39227 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39228 (aarch64_expand_builtin): Expand new builtins.
39230 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39232 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39233 * combine-stack-adj.c (no_unhandled_cfa): Handle
39234 REG_CFA_TOGGLE_RA_MANGLE.
39235 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39236 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39237 info for return address signing.
39238 (aarch64_expand_epilogue): Likewise.
39240 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39242 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39243 * config/aarch64/aarch64-protos.h
39244 (aarch64_return_address_signing_enabled): New declaration.
39245 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39247 (aarch64_expand_prologue): Sign return address before it's pushed onto
39249 (aarch64_expand_epilogue): Authenticate return address fetched from
39251 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39252 (aarch64_attributes): New function attributes for "sign-return-address".
39253 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39254 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39255 ("*do_return"): Generate combined instructions according to key index.
39256 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39257 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39259 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39260 * config/aarch64/aarch64.opt (msign-return-address=): New.
39261 * doc/extend.texi (AArch64 Function Attributes): Documents
39262 "sign-return-address=".
39263 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39265 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39267 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39268 overall option summary.
39270 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39272 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39273 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39274 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39275 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39277 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39279 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39280 -mpower9-minmax by default for -mcpu=power9.
39281 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39282 128-bit floating point.
39284 2017-01-20 Alan Modra <amodra@gmail.com>
39286 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39287 optimizing for size.
39289 2017-01-20 Alan Modra <amodra@gmail.com>
39292 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39293 for strcmp and strncmp from corresponding builtin decl.
39295 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39297 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39298 instead of i386/rtems-64.h.
39299 * config/i386/rtems-64.h: Remove.
39301 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39305 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39307 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39309 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39311 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39312 Change int to HOST_WIDE_INT.
39313 * config/aarch64/aarch64-protos.h
39314 (aarch64_simd_gen_const_vector_dup): Likewise.
39315 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39317 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39319 * langhooks-def.h (lhd_type_for_size): New decl.
39320 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39321 * langhooks.c (lhd_type_for_size): New function, taken from
39324 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39326 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39327 define_bypass for CR latency.
39328 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39329 (power9-alu2): Add define_bypass for CR latency.
39331 (power9-mul): Update insn latency.
39332 (power9-mul-compare): Update insn latency, bypass latency and remove
39335 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39337 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39339 * config/aarch64/aarch64.md
39340 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39341 aarch64_nopcrelative_literal_loads.
39342 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39344 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39346 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39347 TARGET_LOONGSON_3A.
39348 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39350 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39353 * config.gcc (supported_defaults): Add lxc1-sxc1.
39354 (with_lxc1_sxc1): Add validation.
39355 (all_defaults): Add lxc1-sxc1.
39356 * config/mips/mips.opt (mlxc1-sxc1): New option.
39357 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39359 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39360 __mips_no_lxc1_sxc1.
39361 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39362 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39363 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39365 2017-01-19 Richard Biener <rguenther@suse.de>
39367 PR tree-optimization/72488
39368 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39369 sure to restore SSA info.
39370 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39372 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39374 PR rtl-optimization/79121
39375 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39376 of the inner type when shifting an extended value.
39378 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39381 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39382 interposable aliases.
39384 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39387 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39388 Use the evmergelohi instruction.
39389 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39390 (mov_sitf_e500_subreg8_2_be): Likewise.
39391 (mov_sitf_e500_subreg12_2_le): Likewise.
39392 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39393 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39394 (mov_sitf_e500_subreg8_2_le): Likewise.
39395 (mov_sitf_e500_subreg12_2_be): Likewise.
39397 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39399 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39400 attribute from vecsimple to vecperm.
39401 (altivec_vbpermq2): Likewise.
39403 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39406 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39408 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39409 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39410 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39411 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39412 case where N arg is SIZE_MAX.
39413 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39414 (cmpstrsi): Add pattern.
39416 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39418 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39419 __builtin_vec_revb builtins.
39420 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39421 built-in functions to support generation of the ISA 3.0 XXBR<x>
39422 vector byte reverse instructions.
39423 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39424 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39425 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39426 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39427 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39428 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39429 (P9V_BUILTIN_VEC_REVB): Likewise.
39430 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39431 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39432 (p9_xxbrq_v16qi): Likewise.
39433 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
39434 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
39435 (p9_xxbrh_v8hi): Likewise.
39436 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
39437 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
39438 vec_revb built-in functions.
39440 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
39442 PR rtl-optimization/78952
39443 * config/i386/i386.md (any_extract): New code iterator.
39444 (*insvqi_2): Use any_extract for source operand.
39445 (*insvqi_3): Use any_shiftrt for source operand.
39447 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
39449 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
39451 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
39453 2017-01-18 Matthias Klose <doko@ubuntu.com>
39455 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
39457 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39459 * config/rs6000/altivec.h (vec_bperm): Change #define.
39460 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
39461 (altivec_vbpermq2): New define_insn.
39462 (altivec_vbpermd): Likewise.
39463 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
39464 function interface.
39465 (VBPERMD): Likewise.
39466 (VBPERM): New polymorphic function interface.
39467 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
39468 Add entries for P9V_BUILTIN_VEC_VBPERM.
39469 * doc/extend.texi: Add interfaces for vec_bperm.
39471 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39473 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
39474 first letter of error messages.
39475 (s390_resolve_overloaded_builtin): Likewise.
39476 * config/s390/s390.c (s390_expand_builtin): Likewise.
39477 (s390_invalid_arg_for_unprototyped_fn): Likewise.
39478 (s390_valid_target_attribute_inner_p): Likewise.
39479 * config/s390/s390.md ("tabort"): Likewise.
39481 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
39483 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
39484 (ISA_AVOID_DIV_HILO): New macro.
39485 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
39486 (ISA_HAS_DDIV): Likewise.
39488 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39490 * doc/invoke.texi (fabi-version): Correct number of occurrences.
39492 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39494 * doc/invoke.texi (fabi-version): Spelling fix.
39496 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39499 * doc/invoke.texi (fabi-version): Mention mangling fix for
39502 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
39505 * doc/invoke.texi (fabi-version): Document discriminator mangling.
39507 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
39510 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
39511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
39513 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
39515 (stack_protect_test): Ditto.
39516 * config/rs6000/rs6000.opt (mstack-protector-guard=,
39517 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
39519 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
39520 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
39521 -mstack-protector-guard-offset=.
39522 (RS/6000 and PowerPC Options): Ditto.
39524 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39526 * config/i386/i386.h (MASK_CLASS_P): New define.
39527 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
39528 there are no registers from different register sets also when
39529 mask registers are used. Update function comment.
39530 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
39531 to (*k/*r) and (*k/*km) alternatives.
39533 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
39535 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
39536 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
39537 (EH_RETURN_HANDLER_RTX): New define.
39538 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
39539 Force frame pointer in EH return functions.
39540 (aarch64_expand_epilogue): Add barrier for eh_return.
39541 (aarch64_final_eh_return_addr): Remove.
39542 (aarch64_eh_return_handler_rtx): New function.
39543 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
39545 (aarch64_eh_return_handler_rtx): New prototype.
39547 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39549 * config/rs6000/altivec.h (vec_rlmi): New #define.
39550 (vec_vrlnm): Likewise.
39551 (vec_rlnm): Likewise.
39552 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
39553 (UNSPEC_VRLNM): Likewise.
39554 (VIlong): New mode iterator.
39555 (altivec_vrl<VI_char>mi): New define_insn.
39556 (altivec_vrl<VI_char>nm): Likewise.
39557 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
39559 (VRLDNM): Likewise.
39560 (RLNM): New polymorphic function entry.
39561 (VRLWMI): New monomorphic function entry.
39562 (VRLDMI): Likewise.
39563 (RLMI): New polymorphic function entry.
39564 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
39565 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
39566 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
39569 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39572 * dwarf2out.c (field_byte_offset): Restore the
39573 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
39574 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
39575 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
39578 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39581 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
39583 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39586 * dwarf2out.c (add_data_member_location_attribute): For constant
39587 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
39588 instead of DW_AT_data_member_location, DW_AT_bit_offset and
39589 DW_AT_byte_size attributes.
39591 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
39593 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
39594 after forcing to constant memory when the code model is medium.
39596 2017-01-17 Julia Koval <julia.koval@intel.com>
39599 * config/i386/avx512fintrin.h
39600 (_mm512_i32gather_ps): Change __addr type to void const*.
39601 (_mm512_mask_i32gather_ps): Ditto.
39602 (_mm512_i32gather_pd): Ditto.
39603 (_mm512_mask_i32gather_pd): Ditto.
39604 (_mm512_i64gather_ps): Ditto.
39605 (_mm512_mask_i64gather_ps): Ditto.
39606 (_mm512_i64gather_pd): Ditto.
39607 (_mm512_mask_i64gather_pd): Ditto.
39608 (_mm512_i32gather_epi32): Ditto.
39609 (_mm512_mask_i32gather_epi32): Ditto.
39610 (_mm512_i32gather_epi64): Ditto.
39611 (_mm512_mask_i32gather_epi64): Ditto.
39612 (_mm512_i64gather_epi32): Ditto.
39613 (_mm512_mask_i64gather_epi32): Ditto.
39614 (_mm512_i64gather_epi64): Ditto.
39615 (_mm512_mask_i64gather_epi64): Ditto.
39616 (_mm512_i32scatter_ps): Change __addr type to void*.
39617 (_mm512_mask_i32scatter_ps): Ditto.
39618 (_mm512_i32scatter_pd): Ditto.
39619 (_mm512_mask_i32scatter_pd): Ditto.
39620 (_mm512_i64scatter_ps): Ditto.
39621 (_mm512_mask_i64scatter_ps): Ditto.
39622 (_mm512_i64scatter_pd): Ditto.
39623 (_mm512_mask_i64scatter_pd): Ditto.
39624 (_mm512_i32scatter_epi32): Ditto.
39625 (_mm512_mask_i32scatter_epi32): Ditto.
39626 (_mm512_i32scatter_epi64): Ditto.
39627 (_mm512_mask_i32scatter_epi64): Ditto.
39628 (_mm512_i64scatter_epi32): Ditto.
39629 (_mm512_mask_i64scatter_epi32): Ditto.
39630 (_mm512_i64scatter_epi64): Ditto.
39631 (_mm512_mask_i64scatter_epi64): Ditto.
39632 * config/i386/avx512pfintrin.h
39633 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
39634 (_mm512_mask_prefetch_i32gather_ps): Ditto.
39635 (_mm512_mask_prefetch_i64gather_pd): Ditto.
39636 (_mm512_mask_prefetch_i64gather_ps): Ditto.
39637 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
39638 (_mm512_prefetch_i32scatter_ps): Ditto.
39639 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
39640 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
39641 (_mm512_prefetch_i64scatter_pd): Ditto.
39642 (_mm512_prefetch_i64scatter_ps): Ditto.
39643 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
39644 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
39645 * config/i386/avx512vlintrin.h
39646 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
39647 (_mm_mmask_i32gather_ps): Ditto.
39648 (_mm256_mmask_i32gather_pd): Ditto.
39649 (_mm_mmask_i32gather_pd): Ditto.
39650 (_mm256_mmask_i64gather_ps): Ditto.
39651 (_mm_mmask_i64gather_ps): Ditto.
39652 (_mm256_mmask_i64gather_pd): Ditto.
39653 (_mm_mmask_i64gather_pd): Ditto.
39654 (_mm256_mmask_i32gather_epi32): Ditto.
39655 (_mm_mmask_i32gather_epi32): Ditto.
39656 (_mm256_mmask_i32gather_epi64): Ditto.
39657 (_mm_mmask_i32gather_epi64): Ditto.
39658 (_mm256_mmask_i64gather_epi32): Ditto.
39659 (_mm_mmask_i64gather_epi32): Ditto.
39660 (_mm256_mmask_i64gather_epi64): Ditto.
39661 (_mm_mmask_i64gather_epi64): Ditto.
39662 (_mm256_i32scatter_ps): Change __addr type to void*.
39663 (_mm256_mask_i32scatter_ps): Ditto.
39664 (_mm_i32scatter_ps): Ditto.
39665 (_mm_mask_i32scatter_ps): Ditto.
39666 (_mm256_i32scatter_pd): Ditto.
39667 (_mm256_mask_i32scatter_pd): Ditto.
39668 (_mm_i32scatter_pd): Ditto.
39669 (_mm_mask_i32scatter_pd): Ditto.
39670 (_mm256_i64scatter_ps): Ditto.
39671 (_mm256_mask_i64scatter_ps): Ditto.
39672 (_mm_i64scatter_ps): Ditto.
39673 (_mm_mask_i64scatter_ps): Ditto.
39674 (_mm256_i64scatter_pd): Ditto.
39675 (_mm256_mask_i64scatter_pd): Ditto.
39676 (_mm_i64scatter_pd): Ditto.
39677 (_mm_mask_i64scatter_pd): Ditto.
39678 (_mm256_i32scatter_epi32): Ditto.
39679 (_mm256_mask_i32scatter_epi32): Ditto.
39680 (_mm_i32scatter_epi32): Ditto.
39681 (_mm_mask_i32scatter_epi32): Ditto.
39682 (_mm256_i32scatter_epi64): Ditto.
39683 (_mm256_mask_i32scatter_epi64): Ditto.
39684 (_mm_i32scatter_epi64): Ditto.
39685 (_mm_mask_i32scatter_epi64): Ditto.
39686 (_mm256_i64scatter_epi32): Ditto.
39687 (_mm256_mask_i64scatter_epi32): Ditto.
39688 (_mm_i64scatter_epi32): Ditto.
39689 (_mm_mask_i64scatter_epi32): Ditto.
39690 (_mm256_i64scatter_epi64): Ditto.
39691 (_mm256_mask_i64scatter_epi64): Ditto.
39692 (_mm_i64scatter_epi64): Ditto.
39693 (_mm_mask_i64scatter_epi64): Ditto.
39694 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
39695 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
39696 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
39697 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
39698 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
39699 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
39700 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
39701 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
39702 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
39703 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
39704 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
39705 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
39706 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
39707 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
39708 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
39709 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
39710 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
39711 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
39712 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
39713 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
39714 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
39715 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
39716 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
39717 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
39718 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
39719 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
39720 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
39721 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
39722 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
39723 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
39724 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
39725 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
39726 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
39727 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
39728 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
39729 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
39730 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
39731 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
39732 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
39733 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
39734 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
39735 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
39736 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
39737 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
39738 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
39739 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
39740 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
39741 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
39742 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
39743 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
39744 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
39745 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
39746 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
39747 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
39748 definitions accordingly.
39750 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
39751 Kuan-Lin Chen <kuanlinchentw@gmail.com>
39754 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
39757 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
39760 * ira-conflicts.c (ira_build_conflicts): Update total conflict
39761 hard regs for inner regno.
39763 2017-01-17 Martin Liska <mliska@suse.cz>
39766 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
39767 assumption and add comment.
39769 2017-01-17 Nathan Sidwell <nathan@acm.org>
39771 * ipa-visibility.c (localize_node): New function, broken out of ...
39772 (function_and_variable_visibility): ... here. Call it.
39774 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39776 PR middle-end/77445
39777 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39778 correctly set frequency of oudgoing edge.
39779 (duplicate_thread_path): Fix profile updating.
39781 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39784 * configure.ac: Add GCC_BASE_VER.
39785 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39786 version from BASE-VER file.
39787 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39788 (gcc.o): Depend on $(BASEVER).
39789 * common.opt (dumpfullversion): New option.
39790 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39791 * doc/invoke.texi: Document -dumpfullversion.
39792 * doc/install.texi: Document --with-gcc-major-version-only.
39793 * configure: Regenerated.
39795 2017-01-17 Richard Biener <rguenther@suse.de>
39797 PR tree-optimization/71433
39798 * tree-vrp.c (register_new_assert_for): Merge same asserts
39799 on all incoming edges.
39800 (process_assert_insertions_for): Handle insertions at the
39803 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39805 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39806 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39808 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39811 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39814 2017-01-17 Alan Modra <amodra@gmail.com>
39817 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39818 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39819 symbolic stack limit when pic.
39821 2017-01-16 Martin Sebor <msebor@redhat.com>
39823 PR tree-optimization/78608
39824 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39826 2017-01-16 Jeff Law <law@redhat.com>
39829 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39830 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39831 for several include directories that may be relative to sysroot.
39832 * config/i386/x-mingw32 (gplus_includedir): Define.
39833 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39834 (native_system_includedir): Likewise.
39835 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39836 override if TARGET_SYSTEM_ROOT is defined.
39837 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39839 PR tree-optimization/79090
39840 PR tree-optimization/33562
39841 PR tree-optimization/61912
39842 PR tree-optimization/77485
39843 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39844 and computed trims into the dump file.
39846 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39848 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39850 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39853 * gimplify.c (gimplify_init_constructor): If want_value and
39854 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39858 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39859 sequence. Formatting fixes.
39860 (doloop_optimize): Formatting fixes.
39863 * gcc.c (debug_level_greater_than_spec_func): New function.
39864 (static_spec_functions): Add debug-level-gt spec function.
39865 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39867 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39868 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39869 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39870 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39871 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39872 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39874 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39876 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39877 QImode fixups to general and mask registers only.
39879 2017-01-16 Carl Love <cel@us.ibm.com>
39881 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39882 for built-in functions
39883 vector signed char vec_nabs (vector signed char)
39884 vector signed short vec_nabs (vector signed short)
39885 vector signed int vec_nabs (vector signed int)
39886 vector signed long long vec_nabs (vector signed long long)
39887 vector float vec_nabs (vector float)
39888 vector double vec_nabs (vector double)
39889 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39891 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39892 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39893 * doc/extend.texi: Update the documentation file for the new built-in
39896 2017-01-16 Martin Sebor <msebor@redhat.com>
39898 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39901 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39903 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39904 UNSPEC_VSX__XXSPLTD to require special splat handling.
39906 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39909 * system.h: Poison strndup.
39911 2017-01-16 Alan Modra <amodra@gmail.com>
39914 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39917 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39919 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39921 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39923 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39924 call recog here. Assert that INSN_CODE (insn) is non-negative.
39926 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39929 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39931 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39932 in the currently scheduled RTL fragment.
39934 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39936 PR rtl-optimization/78751
39937 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39940 2017-01-14 Jeff Law <law@redhat.com>
39942 PR tree-optimization/79090
39943 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39944 variable length stores.
39945 (compute_trims): Delete dead assignment to *trim_tail.
39946 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39949 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39951 PR rtl-optimization/78626
39952 PR rtl-optimization/78727
39953 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39954 of a block, and split such blocks after everything else is finished.
39956 2017-01-14 Alan Modra <amodra@gmail.com>
39959 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39960 target legitimate_combined_insn.
39961 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39962 (rs6000_legitimate_combined_insn): New function.
39963 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39965 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39966 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39967 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39969 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39971 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39973 2017-01-13 Jeff Law <law@redhat.com>
39975 PR tree-optimization/33562
39976 PR tree-optimization/61912
39977 PR tree-optimization/77485
39978 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39980 (delete_dead_assignment): Likewise.
39981 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39982 statement to delete_dead_call and delete_dead_assignment.
39984 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39987 * substring-locations.c (format_warning_va): Strengthen case 1 so
39988 that both endpoints of the substring must be within the format
39989 range for just the substring to be printed.
39991 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39993 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39994 * config/i386/i386.c (ix86_target_string): Add missing options
39995 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39996 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39997 flags_other and ix86_target_other to flags2_other. Display unknown
39999 (ix86_valid_target_attribute_inner_p): Add missing options and
40000 reorder options by implied ISAs, as in ix86_target_string.
40002 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40004 * hash-table.h (hash_table::too_empty_p): New function.
40005 (hash_table::expand): Use it.
40006 (hash_table::traverse): Likewise.
40007 (hash_table::empty_slot): Use sizeof (value_type) instead of
40008 sizeof (PTR) to convert bytes to elements. Shrink the table
40009 if the current size is excessive for the current number of
40012 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40014 * ira-costs.c (record_reg_classes): Break from the inner loop
40015 early once alt_fail is known to be true. Update outer loop
40016 handling accordingly.
40018 2017-01-13 Jeff Law <law@redhat.com>
40020 * tree-ssa-dse.c (decrement_count): New function.
40021 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40022 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40023 when we know the partially dead statement is a mem* function.
40025 PR tree-optimization/61912
40026 PR tree-optimization/77485
40027 * tree-ssa-dse.c: Include expr.h.
40028 (maybe_trim_constructor_store): New function.
40029 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40031 PR tree-optimization/33562
40032 PR tree-optimization/61912
40033 PR tree-optimization/77485
40034 * doc/invoke.texi: Document new dse-max-object-size param.
40035 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40036 * tree-ssa-dse.c: Include params.h.
40037 (dse_store_status): New enum.
40038 (initialize_ao_ref_for_dse): New, partially extracted from
40040 (valid_ao_ref_for_dse, normalize_ref): New.
40041 (setup_live_bytes_from_ref, compute_trims): Likewise.
40042 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40043 (maybe_trim_partially_dead_store): Likewise.
40044 (maybe_trim_complex_store): Likewise.
40045 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40046 Track what bytes live from the original store. Return tri-state
40047 for dead, partially dead or live.
40048 (dse_dom_walker): Add constructor, destructor and new private members.
40049 (delete_dead_call, delete_dead_assignment): New extracted from
40051 (dse_optimize_stmt): Make a member of dse_dom_walker.
40052 Use initialize_ao_ref_for_dse.
40054 PR tree-optimization/33562
40055 PR tree-optimization/61912
40056 PR tree-optimization/77485
40057 * sbitmap.h (bitmap_count_bits): Prototype.
40058 (bitmap_clear_range, bitmap_set_range): Likewise.
40059 * sbitmap.c (bitmap_clear_range): New function.
40060 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40062 2017-01-13 Martin Liska <mliska@suse.cz>
40065 * function.c (set_cfun): Add new argument force.
40066 * function.h (set_cfun): Likewise.
40067 * ipa-inline-transform.c (inline_call): Use the function when
40068 strict alising from is dropped for function we inline to.
40070 2017-01-13 Richard Biener <rguenther@suse.de>
40072 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40073 for dumping GIMPLE INTEGER_CSTs.
40075 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40077 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40078 to 201112L since C++17.
40080 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40083 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40084 if -fsanitize=kernel-address is present.
40086 2017-01-13 Richard Biener <rguenther@suse.de>
40088 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40089 as _Literal ( type ) number in case usual suffixes do not
40090 preserve all information.
40092 2017-01-13 Richard Biener <rguenther@suse.de>
40094 PR tree-optimization/77283
40095 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40096 and ssa-iterators.h.
40097 (is_feasible_trace): Implement a cost model based on joiner
40100 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40103 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40104 char or short to __float128/_Float128 directly.
40106 2017-01-12 Martin Sebor <msebor@redhat.com>
40108 to -Wformat-overflow.
40109 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40110 (min_bytes_remaining): Same.
40111 (get_string_length): Same.
40112 (format_string): Same.
40113 (format_directive): Same.
40115 (pass_sprintf_length::handle_gimple_call): Same.
40117 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40119 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40120 info.nowrite calls with no lhs that can't throw. Return bool
40121 whether gsi_remove has been called or not.
40122 (pass_sprintf_length::handle_gimple_call): Return bool whether
40123 try_substitute_return_value called gsi_remove. Formatting fix.
40124 (pass_sprintf_length::execute): Don't use gsi_remove if
40125 handle_gimple_call returned true.
40128 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40129 be removed due to side-effects, don't remove following barrier nor
40130 turn the successor edge into fallthru edge.
40132 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40135 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40136 element-reversing loads and stores as not swappable.
40138 2017-01-12 Nathan Sidwell <nathan@acm.org>
40139 Nicolai Stange <nicstange@gmail.com>
40141 * combine.c (try_combine): Don't ignore result of overlap checking
40142 loop. Combine overlap & asm check into single loop.
40144 2017-01-12 Richard Biener <rguenther@suse.de>
40146 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40147 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40149 2017-01-12 Richard Biener <rguenther@suse.de>
40151 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40152 and TS_TARGET_OPTION directly derive from TS_BASE.
40153 * tree-core.h (tree_optimization_option): Derive from tree_base.
40154 (tree_target_option): Likewise.
40156 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40158 * config/i386/i386.c (memory_address_length): Increase len
40159 only when rip_relative_addr_p returns false.
40161 2017-01-11 Julia Koval <julia.koval@intel.com>
40163 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40164 (OPTION_MASK_ISA_SGX_SET): New.
40165 (ix86_handle_option): Handle OPT_msgx.
40166 * config.gcc: Added sgxintrin.h.
40167 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40168 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40169 * config/i386/i386.c (ix86_target_string): Add -msgx.
40171 (ix86_option_override_internal): Handle new options.
40172 (ix86_valid_target_attribute_inner_p): Add sgx.
40173 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40174 * config/i386/i386.opt: Add msgx.
40175 * config/i386/sgxintrin.h: New file.
40176 * config/i386/x86intrin.h: Add sgxintrin.h.
40178 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40181 * fold-const.c (maybe_nonzero_address): Return 1 for function
40183 (tree_single_nonzero_warnv_p): Don't handle function local objects
40187 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40190 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40193 * opts.c: Include "spellcheck.h"
40194 (struct string_fragment): New struct.
40195 (struct edit_distance_traits<const string_fragment &>): New
40197 (get_closest_sanitizer_option): New function.
40198 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40200 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40202 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40204 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40205 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40206 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40207 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40208 for initial die_offset if dwarf_split_debug_info.
40209 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40210 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40212 (output_skeleton_debug_sections): Formatting fix. Use
40213 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40214 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40216 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40218 * config/arm/cortex-a53.md: Add bypasses for
40219 cortex_a53_r2f_cvt.
40220 (cortex_a53_r2f): Only use for transfers.
40221 (cortex_a53_f2r): Likewise.
40222 (cortex_a53_r2f_cvt): Add reservation for conversions.
40223 (cortex_a53_f2r_cvt): Likewise.
40225 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40227 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40228 to all inlined functions, change static to extern.
40230 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40233 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40235 (arm_assemble_integer): Likewise.
40237 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40239 * config.gcc: Use new awk script to check CPU, FPU and architecture
40240 parameters for --with-... options.
40241 * config/arm/parsecpu.awk: New file
40242 * config/arm/arm-cpus.in: New file.
40243 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40245 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40247 * config/arm/t-arm: Update dependency rules.
40248 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40249 of processing .def files.
40250 * config/arm/genopt.sh: Deleted.
40251 * config/arm/gentune.sh: Deleted.
40252 * config/arm/arm-cores.def: Deleted.
40253 * config/arm/arm-arches.def: Deleted.
40254 * config/arm/arm-fpus.def: Deleted.
40255 * config/arm/arm-tune.md: Regenerated.
40256 * config/arm/arm-tables.opt: Regenerated.
40257 * config/arm/arm-cpu.h: New generated file.
40258 * config/arm/arm-cpu-data.h: New generated file.
40259 * config/arm/arm-cpu-cdata.h: New generated file.
40261 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40264 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40266 (input_varpool_node): Unpack dynamically_initialized bit.
40268 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40270 PR rtl-optimization/79032
40271 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40272 the alignment of the adjusted memory reference against that of MODE,
40273 instead of the alignment of the original memory reference.
40275 2017-01-11 Martin Jambor <mjambor@suse.cz>
40277 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40279 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40280 decorated functions.
40282 2017-01-11 Richard Biener <rguenther@suse.de>
40284 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40285 set range/nonnull info for PHI results. Do not set it on
40286 stmts marked for removal.
40288 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40290 * expr.c (store_field): In the bitfield case, fetch the return value
40291 from the registers before applying a single big-endian adjustment.
40292 Always do a final load for a BLKmode value not larger than a word.
40294 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40297 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40298 that we correctly handle column numbers greater than
40299 LINE_MAP_MAX_COLUMN_NUMBER.
40301 2017-01-10 Martin Sebor <msebor@redhat.com>
40303 PR middle-end/78245
40304 * gimple-ssa-sprintf.c (get_destination_size): Call
40305 {init,fini}object_sizes.
40306 * tree-object-size.c (addr_object_size): Adjust.
40307 (pass_through_call): Adjust.
40308 (pass_object_sizes::execute): Adjust.
40309 * tree-object-size.h (fini_object_sizes): Declare.
40311 2017-01-10 Martin Sebor <msebor@redhat.com>
40313 PR tree-optimization/78775
40314 * builtins.c (get_size_range): Move...
40315 * calls.c: ...to here.
40316 (alloc_max_size): Accept zero argument.
40317 (operand_signed_p): Remove.
40318 (maybe_warn_alloc_args_overflow): Call get_size_range.
40319 * calls.h (get_size_range): Declare.
40321 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40323 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40324 from TI's devices.csv file as of September 2016.
40325 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40327 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40329 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40330 * doc/invoke.texi: Likewise.
40331 * doc/md.texi: Likewise.
40332 * doc/objc.texi: Likewise.
40334 2017-01-10 Joshua Conner <joshconner@google.com>
40336 * config/arm/fuchsia-elf.h: New file.
40337 * config/fuchsia.h: New file.
40338 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40339 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40341 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40343 2016-01-10 Richard Biener <rguenther@suse.de>
40345 PR tree-optimization/79034
40346 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40347 Propagate out degenerate PHIs in the joiner.
40349 2017-01-10 Martin Liska <mliska@suse.cz>
40351 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40352 (sort_congruence_classes_by_decl_uid): Likewise.
40353 (sort_congruence_class_groups_by_decl_uid): Likewise.
40354 (sem_item_optimizer::merge_classes): Sort class, groups in these
40355 classes and members in the groups by DECL_UID of declarations.
40356 This would make merge operations stable.
40358 2017-01-10 Martin Liska <mliska@suse.cz>
40360 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40361 usage of m_classes_vec.
40362 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40363 (sem_item_optimizer::get_group_by_hash): Likewise.
40364 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40365 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40366 (sem_item_optimizer::verify_classes): Likewise.
40367 (sem_item_optimizer::process_cong_reduction): Likewise.
40368 (sem_item_optimizer::dump_cong_classes): Likewise.
40369 (sem_item_optimizer::merge_classes): Likewise.
40370 * ipa-icf.h (congruence_class_hash): Rename from
40371 congruence_class_group_hash. Remove declaration of m_classes_vec.
40373 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40375 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40376 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40377 * config.gcc: Add avx512vpopcntdqintrin.h.
40378 * config/i386/avx512vpopcntdqintrin.h: New.
40379 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40380 * config/i386/i386-builtin-types.def: Add new types.
40381 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40382 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40383 __builtin_ia32_vpopcountq_v8di_mask): New.
40384 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40385 __AVX512VPOPCNTDQ__.
40386 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40387 (PTA_AVX512VPOPCNTDQ): Define.
40388 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40389 TARGET_AVX512VPOPCNTDQ_P): Define.
40390 * config/i386/i386.opt: Add mavx512vpopcntdq.
40391 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40392 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40394 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40396 PR middle-end/77484
40397 * predict.def (PRED_CALL): Set to 67.
40399 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40401 * expr.c (store_field): In the bitfield case, if the value comes from
40402 a function call and is of an aggregate type returned in registers, do
40403 not modify the field mode; extract the value in all cases if the mode
40404 is BLKmode and the size is not larger than a word.
40406 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40409 * config/i386/cpuid.h: Fix undefined behavior.
40411 2017-01-04 Jeff Law <law@redhat.com>
40413 PR tree-optimization/79007
40414 PR tree-optimization/67955
40415 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40416 conservative for pt.null when flag_non_call_exceptions is on.
40418 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40420 PR translation/79019
40421 PR translation/79020
40422 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40423 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40424 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40426 * config/avr/avr.opt (maccumulate-args): Likewise.
40427 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40428 * common.opt (freport-bug): Likewise.
40429 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40430 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40431 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40432 translatable string.
40433 * config/i386/i386.c (function_value_32): Likewise.
40434 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
40435 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
40437 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
40438 * common/config/msp430/msp430-common.c (msp430_handle_option):
40440 * symtab.c (symtab_node::verify_base): Likewise.
40441 * opts.c (set_debug_level): Likewise.
40442 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
40443 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
40444 missing whitespace to translatable strings.
40445 * config/avr/avr.md (bswapsi2): Fix typo in comment.
40446 * config/sh/superh.h: Likewise.
40447 * config/i386/xopintrin.h: Likewise.
40448 * config/i386/znver1.md: Likewise.
40449 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
40450 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
40451 * double-int.h (struct double_int): Likewise.
40452 * double-int.c (div_and_round_double): Likewise.
40453 * wide-int.cc: Likewise.
40454 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
40455 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
40456 * cfgcleanup.c (crossjumps_occured): Renamed to ...
40457 (crossjumps_occurred): ... this.
40458 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
40461 PR tree-optimization/78899
40462 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
40463 returning bool return struct loop *, NULL for failure and the new
40465 (versionable_outer_loop_p): Don't version outer loop if it has
40466 dont_vectorized bit set.
40467 (tree_if_conversion): When versioning outer loop, ensure
40468 tree_if_conversion is performed also on the inner loop of the
40469 non-vectorizable outer loop copy.
40470 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
40471 LOOP_VECTORIZED in inner loop of the scalar outer loop and
40472 prevent vectorization of it.
40473 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
40474 the outer loop vectorization of the non-scalar version is attempted
40475 before vectorization of the inner loop in scalar version. If
40476 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
40477 vectorization of its inner loop.
40478 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
40479 has 2 inner loops, rename also on edges from bb whose single pred
40480 is outer_loop->header. Fix typo in function comment.
40482 2017-01-09 Martin Sebor <msebor@redhat.com>
40485 * asan.c (asan_emit_stack_protection): Increase local buffer size
40486 to avoid snprintf truncation warning.
40488 2017-01-09 Andrew Pinski <apinski@cavium.com>
40490 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
40491 to reference thunderx2t99 for the tuning structure
40492 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
40494 (thunderx2t99_extra_costs): This.
40495 * config/aarch64/aarch64-tune.md: Regenerate.
40496 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
40497 (vulcan_addrcost_table): This.
40498 (vulcan_regmove_cost): Rename to ...
40499 (thunderx2t99_regmove_cost): This.
40500 (vulcan_vector_cost): Rename to ...
40501 (thunderx2t99_vector_cost): this.
40502 (vulcan_branch_cost): Rename to ...
40503 (thunderx2t99_branch_cost): This.
40504 (vulcan_tunings): Rename to ...
40505 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
40506 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
40508 2017-01-09 Martin Jambor <mjambor@suse.cz>
40512 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
40513 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
40514 (propagate_vr_accross_jump_function): Use the above function for all
40515 value range computations for pass-through jump functions and type
40516 converasion from explicit value range values.
40517 (ipcp_propagate_stage): Do not attempt to deduce types of formal
40518 parameters from TYPE_ARG_TYPES.
40519 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
40520 (ipa_write_node_info): Stream type of the actual argument.
40521 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
40523 2017-01-09 Martin Liska <mliska@suse.cz>
40526 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
40527 (lookup_compiler): Do not show error message with have_E.
40529 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40531 PR tree-optimization/78938
40532 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
40533 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
40534 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
40535 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
40538 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40540 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
40543 2017-01-09 Richard Biener <rguenther@suse.de>
40545 PR tree-optimization/78997
40546 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
40547 name condition properly.
40549 2017-01-09 Richard Biener <rguenther@suse.de>
40552 * dwarf2out.c (is_cxx): New overload with context.
40553 (is_naming_typedef_decl): Use it.
40555 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40557 * invoke.texi (Option Summary): Correct spacing in option lists
40558 and add line breaks to fix over-long lines.
40560 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40562 PR middle-end/17660
40564 * extend.texi (Common Variable Attributes): Add xref to GCC
40565 Internals manual to explain mode attribute keywords.
40567 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
40570 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
40571 and Preprocessor Options.
40572 (Options for Linking): Document -pthread here....
40573 (RS/6000 and PowerPC Options): ...not here.
40574 (Solaris 2 Options): ...or here.
40575 * doc/cppopts.texi: Document -pthread.
40577 2017-01-08 Martin Sebor <msebor@redhat.com>
40579 PR middle-end/77708
40580 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
40581 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
40582 New member functions.
40583 (format_directive): Used them.
40585 (pass_sprintf_length::handle_gimple_call): Same.
40586 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
40587 to avoid truncation for any argument.
40588 (extract_affine_mul): Same.
40589 * tree.c (get_file_function_name): Same.
40591 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40593 PR middle-end/77484
40594 * predict.def (PRED_INDIR_CALL): Set to 86.
40596 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40598 PR preprocessor/54124
40599 * doc/cppopts.texi: Reformat -d subtable to list the full name
40600 of the options. Add cross-reference to the docs for the general
40601 compiler -d options.
40602 * doc/invoke.texi (Developer Options): Add cross-reference to the
40603 preprocessor-specific -d option documentation.
40605 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40607 PR preprocessor/13498
40608 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
40609 redudant material, and reflect new command-line options.
40610 (System Headers): Likewise.
40612 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
40614 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
40615 -isystem, and -idirafter. Copy-edit.
40616 * doc/cppopts.texi: Copy-edit. Remove contradiction about
40617 default for -ftrack-macro-expansion. Delete obsolete and
40618 badly-formatted implementation details about -fdebug-cpp output.
40619 * doc/cppwarnopts.texi: Copy-edit.
40621 2017-01-07 David Malcolm <dmalcolm@redhat.com>
40624 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40625 that the transition from a max line width >= 1<<10 to narrower
40626 lines works correctly.
40628 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
40630 * doc/options.texi (PerFunction): New.
40631 * opt-functions.awk (switch_flags): Map both Optimization and
40632 PerFunction to CL_OPTIMIZATION.
40633 * opth-gen.awk: Test for PerFunction flag along with
40635 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
40636 it only when the latter is present. Skip those that don't in
40637 the hash function generator.
40638 * common.opt (fvar-tracking): Mark as PerFunction instead of
40640 (fvar-tracking-assignments): Likewise.
40641 (fvar-tracking-assignments-toggle): Likewise.
40642 (fvar-tracking-uninit): Likewise.
40644 2017-01-07 Jakub Jelinek <jakub@redhat.com>
40646 PR translation/79018
40647 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
40650 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
40653 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
40654 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
40655 TARGET_LONG_JUMP_TABLE_OFFSETS.
40656 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
40657 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
40658 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
40659 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
40660 * config/m68k/m68k.md (tablejump expander): Likewise.
40661 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
40662 TARGET_LONG_JUMP_TABLE_OFFSETS.
40663 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
40664 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
40666 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
40667 David Holsgrove <david.holsgrove@xilinx.com>
40669 * common/config/microblaze/microblaze-common.c
40670 (TARGET_EXCEPT_UNWIND_INFO): Remove.
40671 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
40673 * config/microblaze/microblaze.c (microblaze_must_save_register)
40674 (microblaze_expand_epilogue, microblaze_return_addr): Handle
40676 (microblaze_eh_return): New function.
40677 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
40678 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
40679 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
40680 * config/microblaze/microblaze.md (eh_return): New pattern.
40682 2017-01-06 Jakub Jelinek <jakub@redhat.com>
40684 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
40685 GCC_DIAGNOSTIC_STRINGIFY): Define.
40687 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
40689 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40691 * config/arm/arm.md (<mcrr>): New.
40693 * config/arm/arm.c (arm_arch5te): New.
40694 (arm_option_override): Set arm_arch5te.
40695 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
40697 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
40698 (arm_mcrr_qualifiers): ... this. New.
40699 (MRRC_QUALIFIERS): Define to...
40700 (arm_mrrc_qualifiers): ... this. New.
40701 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
40703 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
40704 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
40705 (MRRCI, mrrc, MRRC): New.
40706 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
40707 VUNSPEC_MRRC2): New.
40709 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40711 * config/arm/arm.md (<mcr>): New.
40713 * config/arm/arm.c (arm_coproc_builtin_available): Add
40714 support for mcr, mrc, mcr2 and mrc2.
40715 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
40716 (arm_mcr_qualifiers): ... this. New.
40717 (MRC_QUALIFIERS): Define to ...
40718 (arm_mrc_qualifiers): ... this. New.
40719 (MCR_QUALIFIERS): Define to ...
40720 (arm_mcr_qualifiers): ... this. New.
40721 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
40723 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
40724 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
40725 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
40726 VUNSPEC_MRC2): New.
40728 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40730 * config/arm/arm.md (*ldc): New.
40734 * config/arm/arm.c (arm_coproc_builtin_available): Add
40735 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
40736 (arm_coproc_ldc_stc_legitimate_address): New.
40737 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
40738 'qualifier_const_pointer'.
40739 (LDC_QUALIFIERS): Define to...
40740 (arm_ldc_qualifiers): ... this. New.
40741 (STC_QUALIFIERS): Define to...
40742 (arm_stc_qualifiers): ... this. New.
40743 * config/arm/arm-protos.h
40744 (arm_coproc_ldc_stc_legitimate_address): New.
40745 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
40746 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
40747 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
40748 stc2, stcl, stc2l): New.
40749 * config/arm/constraints.md (Uz): New.
40750 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
40751 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
40752 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
40753 VUNSPEC_STC2L): New.
40755 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40757 * config/arm/arm.md (<cdp>): New.
40758 * config/arm/arm.c (neon_const_bounds): Rename this ...
40759 (arm_const_bounds): ... this.
40760 (arm_coproc_builtin_available): New.
40761 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
40762 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
40763 (CDP_QUALIFIERS): Define to...
40764 (arm_cdp_qualifiers): ... this. New.
40766 (arm_expand_builtin_args): Add case for 6 arguments.
40767 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
40768 (arm_const_bounds): ... this.
40769 (arm_coproc_builtin_available): New.
40770 * config/arm/arm_acle.h (__arm_cdp): New.
40772 * config/arm/arm_acle_builtins.def (cdp): New.
40774 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40775 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40776 'arm_const_bounds'.
40777 * config/arm/types.md (coproc): New.
40778 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40779 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40780 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40781 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40783 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40785 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40786 (UBINOP_QUALIFIERS): New.
40788 (acle_builtin_data): New. Change comment.
40789 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40790 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40791 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40792 arm_acle_builtins.def.
40793 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40794 (arm_init_acle_builtins): New.
40795 (CRC32_BUILTIN): Remove.
40796 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40797 crc32cb, crc32ch and crc32cw.
40798 (arm_init_crc32_builtins): Remove.
40799 (arm_init_builtins): Use arm_init_acle_builtins rather
40800 than arm_init_crc32_builtins.
40801 (arm_expand_acle_builtin): New.
40802 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40803 * config/arm/arm_acle_builtins.def: New.
40805 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40807 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40808 (arm_builtin_datum): ... this.
40809 (arm_init_neon_builtin): Rename to ...
40810 (arm_init_builtin): ... this. Add a new parameters PREFIX
40811 and USE_SIG_IN_NAME.
40812 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40813 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40814 'arm_builtin_datum'.
40815 (arm_init_vfp_builtins): Likewise.
40816 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40817 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40818 (arm_expand_neon_args): Rename to ...
40819 (arm_expand_builtin_args): ... this. Rename builtin_arg
40820 enum values and differentiate between ARG_BUILTIN_MEMORY
40821 and ARG_BUILTIN_NEON_MEMORY.
40822 (arm_expand_neon_builtin_1): Rename to ...
40823 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40824 values, arm_expand_builtin_args and add bool parameter NEON.
40825 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40826 (arm_expand_vfp_builtin): Likewise.
40827 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40829 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40831 PR middle-end/77484
40832 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40833 * predict.c (tree_estimate_probability_bb): Reverse direction of
40834 polymorphic call predictor.
40836 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40838 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40839 (determine_pass_name_match): ...this new function and...
40840 (should_skip_pass_p): ...this new function.
40842 2017-01-06 Nathan Sidwell <nathan@acm.org>
40844 * ipa-visibility.c (function_and_variable_visibility): Reformat
40845 comments and long lines. Remove extrneous if.
40846 * symtab.c (symtab_node::make_decl_local): Fix code format.
40847 (symtab_node::set_section_for_node): Fix comment typo.
40849 2017-01-06 Martin Liska <mliska@suse.cz>
40852 * lra-constraints.c: Rename invariant to lra_invariant.
40853 * predict.c (set_even_probabilities): Initialize e to NULL.
40855 2017-01-05 Martin Sebor <msebor@redhat.com>
40857 PR tree-optimization/78910
40858 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40859 (format_integer): Correct off-by-one error in the handling
40860 of precision with negative numbers in signed conversions..
40862 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40864 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40866 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40868 PR tree-optimization/71016
40869 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40870 factor_out_conditional_conversion. Formatting fix.
40871 (factor_out_conditional_conversion): Add cond_stmt argument.
40872 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40873 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40876 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40878 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40879 read-rtl-function.o, and selftest-rtl.o.
40880 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40881 (selftest::aarch64_test_loading_full_dump): New function.
40882 (selftest::aarch64_run_selftests): New function.
40883 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40884 selftest::aarch64_run_selftests.
40885 * config/i386/i386.c
40886 (selftest::ix86_test_loading_dump_fragment_1): New function.
40887 (selftest::ix86_test_loading_call_insn): New function.
40888 (selftest::ix86_test_loading_full_dump): New function.
40889 (selftest::ix86_test_loading_unspec): New function.
40890 (selftest::ix86_run_selftests): Call the new functions.
40891 * emit-rtl.c (maybe_set_max_label_num): New function.
40892 * emit-rtl.h (maybe_set_max_label_num): New decl.
40893 * function.c (instantiate_decls): Guard call to
40894 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40895 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40897 * gensupport.c (gen_reader::gen_reader): Pass "false"
40898 for new "compact" param of rtx_reader.
40899 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40900 rather than an empty string for NULL strings.
40901 * read-md.c: Potentially include config.h rather than bconfig.h.
40902 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40903 (have_error): New global, copied from errors.c.
40904 (md_reader::read_name): Rename to...
40905 (md_reader::read_name_1): ...this, adding "out_loc" param,
40906 and converting "missing name or number" to returning false, rather
40908 (md_reader::read_name): Reimplement in terms of read_name_1.
40909 (md_reader::read_name_or_nil): New function.
40910 (md_reader::read_string): Handle "(nil)" by returning NULL.
40911 (md_reader::md_reader): Add new param "compact".
40912 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40913 (md_reader::read_file): New method.
40914 * read-md.h (md_reader::md_reader): Add new param "compact".
40915 (md_reader::read_file): New method.
40916 (md_reader::is_compact): New accessor.
40917 (md_reader::read_name): Convert return type from void to file_location.
40918 (md_reader::read_name_or_nil): New decl.
40919 (md_reader::read_name_1): New decl.
40920 (md_reader::m_compact): New field.
40921 (noop_reader::noop_reader): Pass "false" for new "compact" param
40923 (rtx_reader::rtx_reader): Add new "compact" param.
40924 (rtx_reader::read_rtx_operand): Make virtual and convert return
40925 type from void to rtx.
40926 (rtx_reader::read_until): New decl.
40927 (rtx_reader::handle_any_trailing_information): New virtual function.
40928 (rtx_reader::postprocess): New virtual function.
40929 (rtx_reader::finalize_string): New virtual function.
40930 (rtx_reader::m_in_call_function_usage): New field.
40931 (rtx_reader::m_reuse_rtx_by_id): New field.
40932 * read-rtl-function.c: New file.
40933 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40934 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40935 (selftest::verify_three_block_rtl_cfg): New decl.
40936 * read-rtl-function.h: New file.
40937 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40938 For host, include function.h, memmodel.h, and emit-rtl.h.
40939 (one_time_initialization): New function.
40940 (struct compact_insn_name): New struct.
40941 (compact_insn_names): New array.
40942 (find_code): Handle insn codes in compact dumps.
40943 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40944 (bind_subst_iter_and_attr): Likewise.
40945 (add_condition_to_string): Likewise.
40946 (add_condition_to_rtx): Likewise.
40947 (apply_attribute_uses): Likewise.
40948 (add_current_iterators): Likewise.
40949 (apply_iterators): Likewise.
40950 (initialize_iterators): Guard usage of apply_subst_iterator with
40951 #ifdef GENERATOR_FILE.
40952 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40953 (md_reader::read_mapping): Likewise.
40954 (add_define_attr_for_define_subst): Likewise.
40955 (add_define_subst_attr): Likewise.
40956 (read_subst_mapping): Likewise.
40957 (check_code_iterator): Likewise.
40958 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40960 (one_time_initialization): New function.
40961 (rtx_reader::read_until): New method.
40962 (read_flags): New function.
40963 (parse_reg_note_name): New function.
40964 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40965 Handle reuse_rtx ids.
40966 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40967 Add parsing support for RTL dumps, mirroring the special-cases in
40968 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40969 values, and calling handle_any_trailing_information.
40970 (rtx_reader::read_rtx_operand): Convert return type from void
40971 to rtx, returning return_rtx. Handle case 'e'. Call
40972 finalize_string on XSTR and XTMPL fields.
40973 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40974 "(nil)" values were omitted. Call the postprocess vfunc on the
40976 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40977 class ctor. Initialize m_in_call_function_usage. Call
40978 one_time_initialization.
40979 * rtl-tests.c (selftest::test_uncond_jump): Call
40980 set_new_first_and_last_insn.
40981 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40982 * selftest-rtl.c: New file.
40983 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40984 (selftest::get_insn_by_uid): New decl.
40985 * selftest-run-tests.c (selftest::run_tests): Call
40986 read_rtl_function_c_tests.
40987 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40988 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40991 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40993 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40994 operands in a special way. Assert that pos+len <= mode precision.
40996 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40998 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40999 3 argument Alias with unlimited for the negative form.
41000 (fno-vect-cost-model): Removed.
41002 2017-01-05 Martin Liska <mliska@suse.cz>
41004 * hsa-gen.c (gen_hsa_divmod): New function.
41005 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41007 2017-01-05 Martin Liska <mliska@suse.cz>
41010 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41013 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41015 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41016 small constant length operands.
41018 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41020 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41021 between loop iterations.
41023 2017-01-05 Martin Liska <mliska@suse.cz>
41026 * gimplify.c (gimplify_decl_expr): Compare to
41027 asan_poisoned_variables instread of checking flags.
41028 (gimplify_target_expr): Likewise.
41029 (gimplify_expr): Likewise.
41030 (gimplify_function_tree): Conditionally initialize
41031 asan_poisoned_variables.
41033 2017-01-04 Jeff Law <law@redhat.com>
41035 PR tree-optimizatin/78812
41036 * rtl.h (contains_mem_rtx_p): Prototype.
41037 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41038 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41039 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41040 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41041 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41043 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41045 * input.c (assert_char_at_range): Default-initialize actual_range.
41047 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41049 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41050 to match the caller.
41052 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41054 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41055 insns after final jump in test to emit dummy move.
41057 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41059 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41060 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41062 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41064 * multiple_target.c (create_dispatcher_calls): Init e_next.
41065 * tree-ssa-loop-split.c (split_loop): Init border.
41066 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41069 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41074 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41075 (altivec_register_operand): Do not return true if the operand
41076 contains a SUBREG mixing SImode and SFmode.
41077 (vsx_register_operand): Likewise.
41078 (vsx_reg_sfsubreg_ok): New predicate.
41079 (vfloat_operand): Do not return true if the operand contains a
41080 SUBREG mixing SImode and SFmode.
41081 (vint_operand): Likewise.
41082 (vlogical_operand): Likewise.
41083 (gpc_reg_operand): Likewise.
41084 (int_reg_operand): Likewise.
41085 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41086 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41087 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41089 (rs6000_emit_move_si_sf_subreg): New helper function.
41090 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41091 fixup SUBREGs involving SImode and SFmode.
41092 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41093 numbers for the new peephole2 optimization.
41094 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41095 the GLIBC math library that do AND/IOR/XOR operations on single
41096 precision floating point.
41097 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41098 target macros to say whether we need to avoid SUBREGs mixing
41100 (TARGET_ALLOW_SF_SUBREG): Likewise.
41101 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41102 (UNSPEC_SI_FROM_SF): Likewise.
41103 (iorxor): Change spacing.
41104 (and_ior_xor): New iterator for AND, IOR, and XOR.
41105 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41106 (movdi_from_sf_zero_ext): Likewise.
41107 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41108 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41109 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41110 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41111 (fms<mode>4): Likewise.
41112 (fnma<mode>4): Likewise.
41113 (fnms<mode>4): Likewise.
41114 (nfma<mode>4): Likewise.
41115 (nfms<mode>4): Likewise.
41117 2017-01-04 Marek Polacek <polacek@redhat.com>
41120 * doc/invoke.texi: Document -Wpointer-compare.
41122 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41124 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41127 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41128 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41131 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41134 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41135 documentation of the powerpc_popcntb_ok attribute.
41136 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41137 code to issue warning messages if a requested CPU configuration is
41138 not supported by the binary (assembler and loader) toolchain.
41139 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41140 made to define a built-in function that has been disabled.
41141 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41142 made to define a built-in function that has been disabled.
41143 (altivec_init_builtins): Add comment explaining why definition
41144 of the DST built-in functions is not preceded by an assertion
41145 check. Add assertions to prevent ICE if attempts are made to
41146 define an altivec predicate or an abs* built-in function that has
41148 (htm_init_builtins): Add comment explaining why definition of the
41149 htm built-in functions is not preceded by an assertion check.
41151 2017-01-04 Jeff Law <law@redhat.com>
41153 PR tree-optimizatin/67955
41154 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41155 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41156 the points-to solution does not include pt_null. Use DECL_PT_UID
41159 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41161 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41162 Use gen_int_mode instead of gen_lopwart for const_int operands.
41164 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41166 PR tree-optimization/71563
41167 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41168 out of range value - has low bits known to be zero.
41170 2017-01-04 Alan Modra <amodra@gmail.com>
41172 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41173 * configure: Regenerate.
41174 * config.in: Regenerate.
41176 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41179 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41180 a substring of the message, but strcmp with the whole message. Ifdef
41181 ENABLE_NLS, translate the message first using dgettext.
41183 2017-01-03 Jeff Law <law@redhat.com>
41185 PR tree-optimizatin/78856
41186 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41187 (mark_threaded_blocks): Remove code to truncate thread paths that
41188 cross multiple loop headers. Instead invalidate the cached loop
41189 iteration information and handle case of a thread path walking
41190 into an irreducible region.
41192 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41195 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41196 assertions. Add support for doing the signbit if the IEEE 128-bit
41197 floating point value is in a GPR.
41198 * config/rs6000/rs6000.md (Fsignbit): Delete.
41199 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41200 Update the length attribute if the value is in a GPR.
41201 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41202 the sign or zero extension instruction, since the value is always 0/1.
41203 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41206 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41207 extracting SImode to a GPR register so that we can generate a
41208 store, limit the vector to be in a traditional Altivec register
41209 for the vextuwrx instruction.
41211 2017-01-03 Ian Lance Taylor <iant@google.com>
41213 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41215 2017-01-03 Martin Sebor <msebor@redhat.com>
41217 PR tree-optimization/78696
41218 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41219 precision. Use MPFR for %f for greater fidelity. Correct handling
41221 (pass_sprintf_length::compute_format_length): Set width and precision
41222 specified by asrerisk to void_node for vararg functions.
41223 (try_substitute_return_value): Adjust dump output.
41225 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41227 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41229 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41231 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41232 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41233 -mlra/-mno-lra was passed to the compiler.
41235 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41237 PR rtl-optimization/65618
41238 * emit-rtl.c (try_split): Move initialization of "before" and
41239 "after" to just before the call to emit_insn_after_setloc.
41241 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41243 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41245 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41247 * dwarf2out.c (gen_enumeration_type_die): When
41248 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41250 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41252 PR tree-optimization/78965
41253 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41254 Change first argument from const call_info & to call_info &. For %n
41255 set info.nowrite to false.
41257 PR middle-end/78901
41258 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41259 possibly throwing calls.
41261 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41262 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41263 and fns handling, rather than in a separate case SSA_NAME.
41265 2017-01-02 Jeff Law <law@redhat.com>
41267 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41268 fixes for first_period and second_period variables.
41270 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41273 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41274 (*insvqi_1): New insn pattern.
41275 (*insvqi_1_mem_rex64): Ditto.
41276 (*insvqi_2): Ditto.
41277 (*insvqi_3): Rename from *insvqi.
41279 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41281 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41283 * doc/cfg.texi (Edges): Remove reference to Java.
41284 (Maintaining the CFG): Ditto.
41286 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41288 PR middle-end/77674
41289 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41290 transparent aliases.
41292 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41294 PR middle-end/77484
41295 * predict.def (PRED_CALL): Update hitrate.
41296 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41297 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41298 into direct/indirect/polymorphic variants.
41300 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41302 Update copyright years.
41304 * gcc.c (process_command): Update copyright notice dates.
41305 * gcov-dump.c (print_version): Ditto.
41306 * gcov.c (print_version): Ditto.
41307 * gcov-tool.c (print_version): Ditto.
41308 * gengtype.c (create_file): Ditto.
41309 * doc/cpp.texi: Bump @copying's copyright year.
41310 * doc/cppinternals.texi: Ditto.
41311 * doc/gcc.texi: Ditto.
41312 * doc/gccint.texi: Ditto.
41313 * doc/gcov.texi: Ditto.
41314 * doc/install.texi: Ditto.
41315 * doc/invoke.texi: Ditto.
41317 Copyright (C) 2017 Free Software Foundation, Inc.
41319 Copying and distribution of this file, with or without modification,
41320 are permitted in any medium without royalty provided the copyright
41321 notice and this notice are preserved.